Engineering is the Best

Toys for Kids: A Needle and Cap Compass

2010-11-07T14:49:00.000-05:00

At some point in my life, I became a big fan of beer. I don't mean the beer that comes in three-dimensional, easily stacked structures, or the beer that lets you know when it's cold (that deserves a post to itself, although not about the technology behind the indicator, but about the psychology of the people who are actually attracted to it) but the thousands of small breweries around the country, and out of the country, that make and bottle stuff in small quantities. In the end, on the kitchen counter there is invariably a small pile of bottle caps getting ready to do absolutely nothing. Now there's no shortage of creative uses for these caps, but the way I see it, there's no nobler than to play a role in teaching a child something about our physical world.

The things you need for this little experiment are:

sewing needle (a pin is okay too, but not one with a plastic head, it's oddly balanced)
a cold one
bowl or dish
water
refrigerator magnet (or any magnet really, including an electromagnet)

From the above discussion it looks might hypocritical to have a Heineken in the photo, you'll just have to forgive me and not tell anyone. The first thing you want to do is to use the needle to demonstrate how delicate the vitreous body of the human eyeball is. Kidding. Stroke the needle, in one direction only, across one of the magnetic poles. The way to see how the magnet is oriented is simple: just get the magnet near the needle, and see to which side the point of the needle is most attracted. There will be two such points on diametrically opposing sides of the magnet.

It's important not to draw the needle in both directions; pick one direction, and lift the needle, reposition, and repeat. Keep in mind that there are instructions on the internet that say to rub the needle with silk-this is dead wrong. After you've done 40-50 strokes of the needle, take a break and have the cold one. But, when you open it, be very careful to remove the cap by prying up evenly all around the rim-this will leave a cap that's nicely symmetric, and looking like a mini tart tin.

Now, take the needle, and lay it in the middle of the upturned top, and carefully float this in the bowl of water. Within a minute or so, the magnetized needle will align itself with the earth's magnetic field (remember to move the refrigerator magnet away from the bowl!).

Timing Loops in MATLAB

2010-10-12T11:31:00.003-04:00

This probably wouldn't work as the symbol of a church, more than anything because it would be a pain in the ass to hew from a block of wood.

More than Windows vs. Linux, Macs vs. PCs, or Stanford vs. Cal, engineers who simulate with Matlab versus those who use a real programming language (C,C++) have arguments over their respective merits with religious zeal. I try to straddle both camps; unfortunately most of the time I feel like a jack of all trades, master of none. For some of the systems I try to simulate, namely PLLs and ΔΣ modulators, the competing timescales of the slow-er baseband dynamics versus the high speed oscillator/modulator requires very optimized code to get any reasonable speed. Now there are lots of tricks to play to accelerate things, but the point is if you need raw speed Matlab is not the way to go. On the other hand, if you aren't sure how to build your system up and you just want to play around a bit, Matlab has a much more friendly environment to construct incrementally rather than planning everything first. Programming with Matlab is sort of like driving an automatic car; the car takes care of the shifting for you, and all you have to do is press "go" or "stop" with your foot. With Matlab you can define and re-define variables all you want and call any number of the million inbuilt scripts, but you give up a bit of performance and control over the implementation of functions and allocation of memory.

Anyway, if you ever get stuck running a dumb for loop in Matlab, and want to know if you have enough time to get a beer before it finishes, you can try this code. It uses tic to estimate time to completion by measuring how long it took to do a certain number of iterations, and then divides that by how many more iterations there are left to go.

if (mod(ITER_VAR,NUM_ITER) == 0)
    clc;
    tmp = toc;
    tic;
    tmp = tmp*(TOTAL_ITER-ITER_VAR)/NUM_ITER;
    h_rem = floor(tmp/3600);
    m_rem = floor( (tmp - 3600*h_rem)/60);
    s_rem = round(tmp - 3600*h_rem - 60*m_rem);
    h_rem = num2str(h_rem);
    if (m_rem < 10)
        m_rem = strcat('0',num2str(m_rem));
    else
        m_rem = num2str(m_rem);
    end
    if (s_rem < 10)
        s_rem = strcat('0',num2str(s_rem));
    else
        s_rem = num2str(s_rem);
    end
    Time_Remaining = strcat(h_rem,':',m_rem,':',s_rem)

end

The looping variable is ITER_VAR. The number NUM_ITER should be about the number of iterations your loop does per second, so that the countdown refreshes once per second. If you make it update faster, it will appear more real-time, but it adds execution overhead. Adjust to taste. the number TOTAL_ITER is the final value of ITER_VAR, and is used to estimate how much time is remaining in the simulation.

VGA Video from a Sega Dreamcast

2010-09-30T14:13:00.000-04:00

I have a bipolar relationship with video games. Growing up, I was not allowed to touch them (only watch my friend play Zelda on the original NES, with its shiny gold cartridge), and once I got to college and had some money saved up I didn't go buy the PS2, which was new at the time, but the old classics like the Neo Geo. In much the same way that old cars are much more appealing from a tinkerer's perspective, I could actually understand how a Nintendo, Genesis, and Playstation worked and that itself made the system more interesting, regardless of how much I actually cared about the games. As my knowledge of computers and architecture grew, so correspondingly did my interest in more advanced machines. So far, I'm up to about the PS2/Gamecube, which isn't to say I fully understand everything by a long shot, but at least I can wrap my head around the basic principles. I particularly appreciate Sony's machines, even if I think they're architecturally inferior, since they publish their latest silicon in respectable journals and conferences. For example, here is the PS1 CPU, and the PS2 CPU, and the PS3 CPU.

The Sega Dreamcast was a half generation behind the PS2, and in my mind is a marvelous piece of engineering. Designed in the later 1990's, it was exceptionally powerful and forward-thinking, especially considering that it was developed before industry standard consumer graphics platforms were mature. Although it looks unmistakeably dated, it was probably the last machine that featured graphics better than PCs. One of its forward thinking features is its ability to natively render the framebuffer at 640x480, 60Hz which is full VGA resolution, about equivalent to 480p but in 4:3 format. To get a hold of this video stream usually meant purchasing a breakout box from Sega, but ingenious hackers discovered the pinout of the A/V port, allowing direct connection of the video.

It's not entirely clear to me the purpose of the diodes on the sync lines-I can only guess that they give some ESD protection. It was discovered that the 74HCT244s, which are just octal buffers, are not strictly necessary with most displays, and you can drive the H and V sync directly through to the display with 4.7k pullup resistors.

So, the first job is to find a spot on the back of the Dreamcast enclosure to fit the mod, which consists of an HD15 female connector, two RCAs for stereo audio, and a toggle switch.

A few minutes with a drill and a dremel makes short work of the soft ABS plastic.

Although not strictly the most professional technique, I like to make simple circuits inline with heatshrink and wire. Because the Dreamcast is pretty dense internally, I soldered the necessary wires to the copper side of the PCB, and routed them through the ventilation holes in the front of the RF shield by the controller sub-module. Here are the series blocking caps, which are 220u electrolytics in parallel with 2.2u metal films, and below that are the sync resistors.

Even though it's a bit clumsy to route the wires all the way from the back to the front and back again, there's no real signal loss/matching issues since these are VGA bandwidth (and they'll most likely drive a VGA cable which is 10x longer than these wires). All together and ready to go:

And finally, from the back, all closed up.

As I mentioned, I take no credit for the technical discovery of this trick, I just was bored one weekend and searched for information, then hacked my own machine.

Toys for Kids: An Electromagnet

2010-06-21T08:03:00.000-04:00

Although I've never been told I lack any measure of self-esteem, at the same time I don't think (I hope) I boast excessively about my accomplishments, whatever they may be. There is one area, however, where I will happily say with confidence I excelled; manually (by hand) winding electromagnets, transformers, solenoids, chokes, you name it. Of course with the proliferation of the internet there's a good chance that if you googled hard enough you would find someone who's devoted their entire lives, family and children to winding coils, but if you're talking about "normal" people, and I say normal with great hesitance since we are after all discussing me, I don't think there's anyone else out there who has so perfected the art of winding copper wire.

I haven't lived at home for about 10 years. My electronics bench at home is more or less how I left it when I departed for college. In the drawer I still have spools of magnet wire, from #18 to #30.

Underneath the desk I have a 15 pound spool of #36 magnet wire, for winding induction coils, as well as some copper drain pipe for tesla coil primaries.

Just a glance to the shelf on the left reveals more coils (with a backdrop, embarassingly enough, of my Neo-Geo game collection). This is just a small collection; I have wound hundreds of coils, most I just threw out or unwound to use the wire for something else.

Although all the coils you see are linear--with exception of winding coils on a horseshoe, of all things--I've made conical, pancake, toroidal, you name it. I have special bobbins of various sizes depending on the size of the toroidal core I'm using. At this point I think my parents, when they saw me doing this, secretly wished I just liked Nintendo like everyone else.

My most impressive coil was to wind an induction coil from #35 magnet wire, by hand, on a former 8" long by 3.5" wide. Every day for 3 months I would come home from school (I was in 3rd grade or something) and wind an even layer, wrap with a sheet of wax paper, dip it in varnish, and wind another layer. I started with a 5 pound spool of wire, and was almost empty when I finished. I literally wound miles upon miles of perfect, laminar windings. And then once I finished, the input lead to the first layer broke off. I didn't even cry, I just started over.

My favorite coil of all time is this one. I wound it when I was 8, it is made from a bolt, two fiber washers, and exactly 1/4 pound of #26 wire. Even today I think the windings aren't half bad.

Anyway, so you get the picture. Faucet's leaking? Call a plumber. Need to off someone? Hire a hitman. Need a high-performance, aesthetically-pleasing coil of wire, and some asshat to wax poetic about it? I'm your man.

See the thing is, I'm not convinced that I'm the only one with this problem. I think, and this is a totally biased perspective, that lots of little kids would have fun playing with magnets, coils, batteries, just the same way I did. They just need someone to start them off. Eccomi qui! So I've decided to start a section of the blog devoted to small toys, projects and experiments with the little ones in mind. If this turns out to be a total disaster, keep in mind that I was an odd little one.

The first project I have in mind, if you haven't already gotten the sense, is to make an electromagnet. So great, go get some wire, bolt, and be done with it. Not quite so simple. See, back in the day, kits and experiments for kids like HeathKit were very close to the "real thing".

A -> ? -> ? -> ? -> B... oh hell. You want me to build what?

You got the raw parts, some excellent instructions, and were expected to have the manual dexterity, engineering insight, and experience to correctly assemble a complex system. In a sense, they expected a very high level of expertise and maturity out of the hobbyist. For the advanced experimenter, it is nice to be working on things that are challenging but with just enough guidance to enable one to build a device that would be otherwise unattainable; for the beginning hobbyist, it is simply too daunting and most likely would end in frustration and failure.

These days, it's the total opposite. Electronic kits dumb everything down to a sequence of manual operations with zero insight, learning, or even fun. In short, you might as well be pressing buttons on a video game. Take for example, "Snap Circuits":

Wait so what you're saying is, everytime I fasten my pants the wheel will turn and the light will go on?

On that board are R's, C's, L's, and even a trimcap. But all the kid is going to know is that "when I put the white block, green block and one of the yellow blocks together and connect it to the lower right portion of the board near the batteries, some hum comes out of the red box." Where's the learning there? Making a young kid strip, tin, and solder wires is a bit too much to ask, but snapping circuit elements together is banal. Somewhere between these two extremes is a happy medium. Here's my list of good and bad:

GOOD

Mechanically reinforced wires and leads.
Durable, modular connectors like banana plugs, alligator clips, etc. along with bare wire leads.
Wires! Lots of them, different colors, sizes, etc.
The possibility of irreversible damage, in controlled amounts. Too many kits make sure devices are more or less electrically bulletproof. This is what we professionals do because we know that although we take every precaution, sometimes accidents happen, and we don't want to destroy our work. But kids haven't learned good safety practices, and should therefore be exposed to the idea that sometimes mistakes can't be undone. I'm not talking about sticking metal forks into every power outlet and saying, "Don't touch, it's dangerous" rather say a lightbulb with no ballast resistor with the warning "do not apply more than 3V."

BAD

Rigid, lego-like construction; limits configuration options, gives the wrong idea (that electricity flows in fixed paths).
Too much abstraction; inner workings should be exposed as much as possible, even if it compromises durability a bit (after all, learning to be careful is a big lesson we all need to learn).
Exposing too much at once. Hours of fun can be had with a lightbulb, a battery, paperclips and some wire. Are a potentiometer and trimcap really necessary? Never introduce anything that can't be explained, at least in a basic intuitive manner.

Following these basic tenants, I proceed to make a toy electromagnet. First, you're going to need some #26 magnet wire, packing tape, a 3/8x3" bolt (threading doesn't matter) washers and nuts.

The first task is to make an electrically insulated former, on which the windings will be wound. To do this, you first need to wrap a layer or two of packing tape on the bolt, as well as put a layer on each inside surface of the washers. This can be trimmed with a razor blade.

Winding begins by affixing one end securely, and winding one laminar layer atop the previous.

When winding coils, it is important to rotate the former rather than winding by wrapping the wire around the stationary former. This is because the latter will introduce twisted tension into the wire, which will make your windings uneven. If you're a pro at this, you can wind each layer atop the previous without putting an insulating layer in between; otherwise, a single layer of tape between each layer will provide some mechanical stability (and also a safety net, in case you drop the coil only the current layer will unwind). When the topmost layer is reached, wrap the windings in several layers of clear tape.

Now you've got to mechanically reinforce the leads to the electromagnet, because the inner wire is especially vulnerable to breaking off, and it if does, all is lost. To do this, first you have to make a physically strong anchor for the leads, so that if you pull the leads from the electromagnet it will not put force on the magnet wire windings.

Next we employ a common technique called "potting" whereby the entire wiring is enclosed in a heat-shrink tube, and all the air inside is replaced with epoxy. This is done for many reasons: waterproofing, vibration damping, physical durability, to name a few. Here is the finished result:

So there it is: a fairly kid-proof toy that is still visibly electromechanical.

Repairing an Adcom GFP-565 Preamplifier

2010-02-07T18:03:00.000-05:00

My dad's stereo setup is probably what got me interested in electronics to begin with. All those exposed wires, vibrating speaker diaphragms, display panel lights, turning records, oh dear I'm going on like it's the latest SI Swimsuit issue, I've got major problems. Moving on, We recently took this wonderful preamp out of storage, and no sound came out. Naturally, I will put my degree to work in a practical manner (this is most rare) and diagnose, then repair the problem. I will attempt to demonstrate what I think is a logical, methodical debugging protocol.

The Adcom GFP-565 stereo preamplifier debuted around 1990 and was billed as the "first affordable stereophile preamp" retailing for $900 (though clearly "affordable" is a subjective term, being $1543 adjusted for inflation). A modest design, it has three line-level inputs, a phono input, an a tape dubbing loop. It also has tone controls (which can be bypassed) and no remote control. The construction, inside and out is fantastic.

On the PCB, you can see that every square inch not used for signals forms a ground plane. Some will argue this is not a good idea, clearly Adcom disagreed.

You can also see those silver and blue ribbons that look like ladders. Those are the coolest switches I have seen. What they are are switches which are electrical at one end, and mechanical at the other. A common problem with preamplifiers is that the input selector switches are very early in the signal path, right after the input jacks, which are typically on the back of the chassis. The switches are usually conveniently located on the front. The problem then is that it is bad practice to route the electrical signals to the front, especially because at that stage they are very low level. A standard practice is to have rotary switches with very long shafts, that extend from the back of the case to the front. In this preamp, the knobs at the front cause the metal ribbon to move within the blue plastic sleeve, repositioning the switch. In spite of this complicated mechanical system, the switches have a very firm, positive feel. Here is another view of the switches.

The power supply is a discrete linear circuit, with a nicely shielded transformer.

The PCB underside is unsurprising. This is a two-layer PCB.

The most important first step, and this will save hours of labor, is to try and find the circuit schematic. Fortunately, on some questionable Russian webserver (thanks Google!) I obtained, by first uploading some schematics of Carver power amps, a decent schematic of this preamp.

If the schematic is hard to read, click the image for the full resolution.

So looking at the schematic, we make these deductions.

If there is no sound from either channel, probably either the power supply is out, or the output relay circuity is blown. That is because the L/R channels don't share any ICs, they are independent paths.
If there is sound from one channel not the other, most likely the ouput transistors of one of the power op-amps has failed. These are the Adcom 1A op amps, they are mounted on a heat sink.
The only switch besides the input selector that is in the signal path is the tone control. This is a likely point of failure.

Here is my test setup. I have a scope, multimeter (Not the fanciest one, but one I got from Radio Shack recently because I didn't have one at home. Plus, I love analog multimeters.) and a signal source, aka. iPod. Since I am not doing any quantitative measurements aside from seeing if there is a signal or not, looking at an audio wave bouncing around on a scope is more than than a boing ol' sine wave.

Looking for a problem is best done with a "binary partition search" or so I think it's called. Basically, you keep dividing your circuit in two, to narrow down the problem location.

So first, I tested the supplies. Both the +/-18V and +/-15V regulated supplies were functioning correctly. So now I want to probe the midpoint of the circuit, and I choose that to be the "Bypass" output. There is still no signal. So, I need to go back one. To see if the input selectors are working, I probed R205/R206, and there was a signal. This means there is a problem within the first buffer stage, off the volume/balance control. The outputs of IC201/IC202, the volume control buffer, were also dead. That means we've isolated the problem to be in the balance or volume control pots. Looking at the output of the balance pot, there was no signal. Bingo, the problem was within the balance pot. The pot was made by Alps, a well respected maker of precision resistors. How both wipers of the pot died at the same time is beyond me, but after 20 years of faithful service I suppose it's allowed. Anyway, fixing that and the amp works good as new. I have summarized the steps of my debugging process in the marked-up schematic.

What's Coming in 2010

2010-02-07T16:41:00.000-05:00

Well hello, I'm back! It's been almost exactly a year, hasn't it. Don't worry, I haven't abandoned this blog, in fact in the coming months I'm going to go in two directions: one is going to be mostly experimental tinkering projects, and the other is going to be much more technical engineering.

While I am principally occupied with my research, I still get a lot of ideas for circuits, though they are not part of my main focus are still worth investigating in their own right. These will be no-compromise, very throrough investigations. At the same time, I realize that not everyone has the time, resources or interest to complete such involved projects, and on the other side of the spectrum from time to time I do something fun that takes an afternoon, and I will post those as well.

For a sneak preview, projects that are currently in the pipeline include a stereo integrated amplifier with remote control, bi-amplified desktop speakers, a compact high-fidelity class-D amplifier, and a frac-N PLL demonstration board.

A Table Top Tesla Coil, mk. 2

2009-02-05T15:03:00.003-05:00

Salt shaker shown for size reference.

As I've mentioned in past posts, of all my passing nerd interests none capture my heart like sparks and explosions. Which currently brings to mind the internal combustion engine, but somehow that's a bit too contained to be really exciting. About four years ago I made this mini tesla coil, and as of three years ago I'd been planning on re-designing the secondary. Of course one project spawns another, so I wound the secondary coil (a blog post in its own right, here) which has now been sitting on my shelf, collecting dust.

Anyway I recently found myself with some spare time, and with a new year's resolution to catch up on old projects, I finally finished what I started four years ago. So let's start at the beginning.

The primary coil is wound from 3/16" copper pipe. There are three reasons for this. Visually, copper pipe is a winner. Mechanically it's very malleable yet holds its shape well. Electrically, a hollow copper pipe has almost twice the surface area as solid pipe of the same diameter, and by the skin effect at high frequencies, poses a much lower resistance. Check out Litz wire, if you don't believe me. The form is 1/4" acrylic, notched and glued to an acrylic cylinder. To the left, you can see the secondary standing around.

Here you see the primary section, waiting to be re-wired. The upper right is the 4kV transformer. I have no idea what that was originally for. It's not a NST, because it has no internal current limiting, and wasn't designed for 100% duty cycle. My best guess would be an OBIT, since it provides fairly high current, and was just meant to be pulsed on. Below that is the homemade RF filter, the spark gap (minus electrodes) and capacitor bank.

The secondary ground return is shown, a small screw post. Yes, as the glue is currently drying, the thing is resting on a box of kitty litter.

The secondary and primary mounted together. I haven't put the arc shield between the primary and secondary yet, as I hadn't powered it up.

1/4-20 mounting bolt for the topload. As you saw in the first photo, I opted for a simple spherical topload rather than the toroid I was going to make. I don't have access to the Alpha CNC lathe at the moment, as soon as I do I'll turn that.

With the thing properly grounded in dry weather I'm getting 4" sparks, so by the rule of thumb that's around 120kV. I'm still having problems with strikes to the primary, so next time I either make the primary more dense, or put another shield between the two. Click for high resolution images.

I put a CD on the topload, and a ground terminal above it. Sparks are very pretty, if you ask me. If only I were a better photographer...

Installing Ubuntu Eee (Hardy Heron) on the EeePC 900

2008-07-01T05:16:00.010-04:00

DISCLAIMER: All the information in this post can be found on the internet: though Google, or by following links from the site http://www.eeeuser.com/ and http://www.ubuntu-eee.com/. I simply wrote this because when I installed Ubuntu, there was no straightforward list of things to do, and some suggestions conflicted with others. So I simply am writing down the steps I took which resulted in a nearly perfect system.

My Macbook Pro kicked the bucket a week after moving to Italy. In need of something to tide me over while I'm here (and not wanting to send my expensive laptop through the Italian post for repair) I went and purchased an EEEPC 900. There's some bait-and-switch fiasco going on with the battery, but I'm happy to report that mine came with a 5800mAh battery.

The Xandros OS was fine for a while, especially when I switched to advanced desktop mode, but I didn't like the fact that the Asus repository for packages was pretty out of date. For example, to run Firefox 3 you need gtk 2.10 or above, and the EEE comes with gtk 2.8, as many people have noted on forums.

I just want Firefox, why not just build gtk 2.10 instead of installing a whole new OS? I guess it's because I'm a lazy Linux user. I like the fact that Linux is mostly free, and infinitely configurable, but I really hate doing it. The thing I love about Ubuntu (and maybe other distros, I wouldn't know) is the automatic package manager is just like the software update on the Mac; unlike maintaining the RedHat Enterprise machine at work, which requires manual package compatibility and requirement checking, the Synaptic package manager does all that for you, and "just works".

I chose to install Ubuntu Eee, a streamlined version of Hardy Heron (8.0.4) designed specifically for the Eee. The instructions on the Ubuntu Eee Wiki are somewhat wrong, at least in my experience. First of all, I found that my 16GB Corsair Voyager wasn't recognized by the BIOS as a bootable device; instead I used a 1GB USB key. Here's what I did, on a plain desktop Ubuntu 8.0.4 system.

INSTALLATION

Download the ISO from the Ubuntu Eee site. There's also a torrent link.
Stick the USB drive in the machine (make sure there's nothing on it you want to keep) and let the machine auto-mount it. When it does, use the file browser to find the path (mine was /media/disk) then type in a terminal
```
mount | grep /media/disk
```
which should spit something out like,
```
cwu@ceeepc:~$ mount | grep /media/disk
/dev/sdd1 on /media/disk type vfat (rw,nosuid,nodev,uhelper=hal,shortname=mixed,uid=1000,utf8,umask=077,flush)
```
So now I know the drive is /dev/sdd1, although yours may be different. Now I'm going to unmount it (it's about to be erased...) and get to work.
```
cwu@ceeepc:~$ umount /media/disk
```
First we need some helper programs. Do:
```
sudo -s
apt-get install syslinux
wget http://startx.ro/sugar/isotostick.sh
chmod +x isotostick.sh
parted /dev/sdX
```
The first command will ask you for your password, this way you don't have to use sudo each time. The second command retrieves a set of tools from the net, and the third retrieves a script with which we can install the ISO image onto the USB stick. The fourth command makes the script executable, and finally we launch the partitioning program on the USB stick. Make sure to replace X with whatever drive letter you had from the previous step.

Here's where my method departs from the instructions in the Wiki. For some reason, mkfs.vfat doesn't create a partition 1 for me... I have no idea why. So in parted, you get a prompt. Here's what to do. First, take a look at the partition table. On my key, it looks like this (I'm recreating the steps with the 16GB drive):

(parted) print                                   

Disk /dev/sdd1: 16.1GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos

Number  Start  End  Size  Type  File system  Flags

(parted)

Now, create a partition:

(parted) mkpart                                   
Partition type?  primary/extended? primary        
File system type?  [ext2]? fat32                  
Start? 1                                          
End? 16.1GB                                       
(parted) print

Disk /dev/sdd1: 16.1GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos

Number  Start   End     Size    Type     File system  Flags
1      32.3kB  16.1GB  16.1GB  primary               lba

(parted)

Now we have to format this partition.

(parted) mkfs                                      
Partition number? 1                                
File system?  [ext2]? fat32
(parted) print                                     

Disk /dev/sdd1: 16.1GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos

Number  Start   End     Size    Type     File system  Flags
1      32.3kB  16.1GB  16.1GB  primary  fat32        lba

(parted)

Finally, we have to make this partition bootable.

(parted) toggle 1 boot                             
(parted) print

Disk /dev/sdd1: 16.1GB
Sector size (logical/physical): 512B/512B
Partition Table: msdos

Number  Start   End     Size    Type     File system  Flags
1      32.3kB  16.1GB  16.1GB  primary  fat32        boot, lba

(parted)

Now quit from parted, and we install the Ubuntu image onto the key with:
```
./isotostick.sh ubuntu-eee-804.iso /dev/sdX
syslinux /dev/sdX
```
(replace X with your drive letter as before) The first command takes a while, as it has to copy a 700MB ISO image onto the key. The second command has something to do with the booting from the key, I'm not totally sure.
Once this is done, you stick the key in your Eee and reboot. As the first EeePC splash screen comes up, press the ESC key a bunch of times to make the boot selector come up. Choose your USB key, and it will boot into Ubuntu. Installation is really easy, the only thing I changed was the partition management:
My EeePC has two drives, a 4GB SSD and a 16GB SSD. when asked for the partition setup, I did the following: I removed the swap, for reasons of SSD longevity (debatable, given the average life cycle of an electronic gadget these days) I mounted the 4GB drive a / with an EXT2 filesystem (no journaling, for aforementioned reasons) and the 16GB mounted to /home, also EXT2.
The installation takes about 10 minutes, and when it's done, it asks you to reboot the system with the USB key removed. Assuming you don't want to re-install, your USB key is now free.

POST-INSTALL TWEAKS

Auto-Update
Pretty much everything works out of the box. The wired network, wireless, screen resolution all seemed fine, and I let Synaptic update the system with the newest packages, which took a good hour because it had about 160MB of stuff to download (including the release version of Firefox 3!).

Camera
I noticed that the webcam still wasn't working with Skype even after I re-enabled it in the BIOS (somehow when I installed Ubuntu it turned itself off), so I found a fix for this on the net, which I'll summarize below:

sudo aptitude install subversion
svn co svn://svn.berlios.de/linux-uvc/linux-uvc/trunk linux-uvc
cd linux-uvc
sudo make
sudo make install
sudo modprobe -r uvcvideo
sudo mv /lib/modules/$(uname -r)/ubuntu/media/usbvideo/uvcvideo.ko /lib/modules/$(uname -r)/ubuntu/media/usbvideo/uvcvideo.ko.original
sudo cp uvcvideo.ko /lib/modules/$(uname -r)/ubuntu/media/usbvideo/uvcvideo.ko
sudo modprobe uvcvideo

This makes the webcam with with skype, although at around 5fps. (not a Macbook Pro)

Reducing Disk Writes, and Automounting the SD Card Slot
This requires modifying the /etc/fstab file so that it looks something like this:

# /etc/fstab: static file system information.
#
#                      
proc                /proc           proc    defaults          0       0
# /dev/sda1
UUID=718516b9-c26d-470d-b99d-231ecd0927aa  /               ext2    relatime,errors=remount-ro  0       1
# /dev/sdb1
UUID=e5be350e-f189-4503-a992-bd8c9c9217a0  /home           ext2    relatime          0       2
# /dev/sdc1           /media/cdrom0   udf,iso9660 user,noauto,exec,utf8  0       0
tmpfs      /var/log tmpfs defaults,noatime  0 0
tmpfs      /tmp  tmpfs defaults,noatime  0 0
tmpfs      /var/tmp tmpfs defaults,noatime  0 0

Specifically, comment out the line beginning with /dev/sdc1, that tries to mount SD cards as CD drives and fails, then add the three tmpfs lines below. This writes things to RAM instead of the SSD.

The Shutdown Process, Audio and Hotkeys
Next, I followed the instructions here (a slightly cleaner solution) to make the Eee shut down properly.

Then to get audio to work, add:
```
options snd-hda-intel model=auto
```
to the file /etc/modprobe.d/snd-hda-intel or create it if it doesn't exist. To make all the Hotkeys work, I created the file /etc/acpi/eeepc-hotkeys.sh containing
```
#!/bin/sh

code=$3

case $code in
# Fn+F2 -- enable/disable wifi
0000001[01])
/etc/acpi/actions/wireless-toggle.sh
;;
# Fn+F7 -- mute/unmute speakers
00000013)
acpi_fakekey 113
;;
# Fn+F8 -- decrease volume
00000014)
acpi_fakekey 114
;;
# Fn+F9 -- increase volume
00000015)
acpi_fakekey 115
;;
esac
```
then made it executable with
```
chmod a+x /etc/acpi/eeepc-hotkeys.sh
```
Then, I created the file /etc/acpi/events/hotkey containing
```
event=hotkey ATKD
action=/etc/acpi/eeepc-hotkeys.sh %e
```
and rebooted the system.
Boot Speed and Battery Life, Wifi LED
I'm not sure how much of a difference these things make, but I first I changed the writeback interval by adding the lines
```
vm.dirty_writeback_centisecs=1500
dev.wifi0.ledpin=1
dev.wifi0.softled=1
```
to the file /etc/sysctl.conf. (I slipped in the Wifi LED fix)

Then, I changed this line, appending the clocksource=hpet
```
# defoptions=quiet splash clocksource=hpet
```
to the file /boot/grub/menu.lst. You need to reload the file with
```
sudo update-grub
```
Finally, I profiled the bootup by following the instructions here. I don't know how fast it's supposed to be, but I boot up in about 50 seconds.

Vodafone HSDPA USB Modem (Huawei E272)

At my current apartment, I don't have a fixed internet connection, so I went to Vodafone and purchased a 3G USB modem. The plan's pretty good; it's pay-as-you-go, 15€ a month for 30 hours unlimited data per month, and the first month is free and I get 100 hours. The modem is a Huawei E272, the latest and greatest and cost 200€, and it seems to be unlockable from what I'm reading on the net. I considered the key-style E172, but decided against it because if I ever get a Macbook Air, I've heard the larger dongles have problems fitting in the single (sigh) USB port of the Air. Besides, the E272 has an external antenna connector.What worried me about buying the newest modem (instead of the E220, which can't go as fast) is it might not be supported in Linux yet. But, I took a chance. In Synaptic, I installed the Vodafone Mobile Connect software version 1.99.17 (search "vodafone") which also installed a few python packages. I know there's a beta 3 version 2.0 on the net, but I figured I'd go with the stable version given it claimed to have been tested with the E272.

The first time I launched the program (Applications -> Internet) I accidentally jiggled the USB connector and lost connection, and after that the program would hang on the first splash screen. I found the solution was to excecute

rm -rf ~/.vmc2

to remove any preferences files. After that, the software correctly recognized the modem as an E272, and everything works fine. I don't even need to reboot the Eee with the modem plugged in. The only problem seems to be that since I am dialing up with wvdial and bypassing the Ubuntu Network Manager, Firefox starts up in offline mode. It's not a real problem, I just change that and go.

That's pretty much it, it's a very light sturdy computer, it's not exactly as cheap as it was promised to be, but I feel a lot better carrying this around and the diminutive AC adapter than my MBP. It's running a full desktop OS (minus the bells and whistles) and does everything I need, with aplomb.

The New Ride

2008-05-08T10:24:00.002-04:00

Barring a miserable job/life that you can't bear to think about, I can't imagine any reason to go bicycling in the winter in New York. Being fortunately without either of these motivations, my bike laid in disrepair during the cold months. At the end of last season I'd discovered that somehow the spline-spindle interface of my cranks had partially sheared on the drive side, owing in all probability to my massively strong left thigh (this is only partially a joke, I imagine a lightweight component like a Dura-Ace track crank was not designed to sustain braking forces for two years) which left my bike unsafe to ride.

I intended to take this opportunity and rebuild my bike with a freewheel, since I was getting sick of the proliferation of fixies without brakes and their invariable association with hipsters (right now you can hear Darwin laughing in his grave), but the real motivation came a month ago when my friend tricked me into signing up for the Westchester Jarden Triathlon, an Olympic distance event: 1.5km swim, 40km bike, 10km run. I hate running almost as much as I hate swimming. The reasonable thing to do would be to invest in some gears for the bike. Then I figured, I'm going to shoot the moon anyway, I might as well go all out and do the bike portion with one speed.

So, here's the plan: I'm really attached to my frame, so that stays. The saddle and seatube also stay, given there's nothing wrong with them. The bottom bracket is fine, having just replaced it in the fall. The front rim, a Mavic Open Pro is in great shape, and I've got a Mavic Open SUP (an older but still good rim) in storage that I'm going to use instead of the CXP22, which I currently have. Everything else goes.

The rear wheel is an Ultegra 10 speed hub laced with 14/15 DT Competition spokes 2X to the Mavic Open SUP. On it are a bunch of spacers, and a Surly 14T cog.

The front wheel is an Ultegra hub laced to the Open Pro radially with DT Revolution 14/17 spokes. I didn't mean to buy black, but I'm in favor of supporting the local bike store and they have a spoke cutting machine, but only had these spokes in black so I paid a small premium but whatever.

My old front fork was a clunky steel thing, and the headset bearings were about to die, so I figured as long as I'm upgrading the fork I might as well put in a threadless headset. Pressing the bearing cups into the frame was a bit of a chore, but with a few clamps it worked out.

Then I discovered that the stupid steerer tube on the Nashbar carbon fork (not the lightest I know, but it was a bargain on sale) was too wide at the base to accept the crown race, so I had to spend 20 minutes sanding it down so it'd fit.

This must have made my office look pretty strange from the street.

I'm going to be training all summer in Italy, so I figured I might as well be patriotic. It could also be that I got the handlebar tape in a bargain bin for $5. The computer is a simple Cateye model, and while it doesn't have a Cadence meter, who needs one when you've only got one gear? The handlebar is a Profile Airwing, and the brake levers are Syntace space controls, which were just on sale at Colorado Cyclist.

The cranks are Ultegra doubles, with a 53T front ring. This is a pretty insane gear ratio, but again, we're going to shoot the moon here. The pedals are also Ultegra.

The brakes I stole off another bike project, I think they're house components from a Trek 1000 series bike. They get the job done.

At 17.4lbs she's no spring chicken, all the same she's pretty cute, eh? The goal: 42.9kph@90rpm.

Understanding Thiele/Small Parameters

2008-01-01T18:51:00.000-05:00

Although it may pale in comparison to the wonders of the PS3, which can push upwards of 100 million textured, occluded, bump-mapped, shaded anti-aliased polygons per second, it wasn't many decades ago that recorded sound--and the reproduction of the recording--was still pure magic. Heck even now, what's not absolutely incredible about the concept that a conical (well, usually) contraption of metal, paper, plastic and wire can vibrate with such precision that the digital audio processing pathways require 16 and 24 bits? As you might imagine, the design of loudspeakers has benefited tremendously from modern computing power, with FEM solvers and acoustic modeling packages providing super accurate design resources. However, just because we have Cadence doesn't mean we don't need to learn KVL. Furthermore, for hobbyists like me--or more precisely physics morons like me--a complete 3D simulation of a loudspeaker is not only unfeasible it kinda takes the fun out of it.

Before all this simulation business, in the 60's and 70's, a standardized set of parameters was developed which forms the bare minimum of information needed to design a loudspeaker with the driver in question. Think of these parameters as the SPICE "Level-0" model equivalent, with μ_n, C_ox, V_t and the like. What the T/S parameters do for drivers is the same as what the hybrid-π or lumped element models do for circuits; they abstract a very complex mechanism in a way that still has some physical meaning, so that we may design our speakers in an intuitive, methodical manner.

The complexity of a loudspeaker driver is obvious; the magnetic field formed by the magnet and pole pieces is nonlinear, the diaphragm has multiple resonance modes, the surround has nonlinear compliance and damping, the forward and backward radiation impedances are unequal, and the dust cap and cone have a nonuniform radiation pattern, to name a few. The thing is, awesome speakers were designed before we had to think about all these issues, so I maintain it's not absolutely necessary to deal with all the second order effects, and we can do quite well with just T/S parameters. Therefore we ought to draw as much insight as possible from these parameters, and understand what they mean.

I should mention that in the process of organizing my notes to make a coherent blog post, I did some Wikipedia searching to standardize my notation, at which point I discovered that as with most things I do, someone else has already done it better (and written about it). So if you want to see what I would have liked to be able to say I'd written, check here.

Alright so let's start at the beginning. I'm going to look at this from a circuit point of view, since I know most of nothing about physics. We're going to formulate a lumped element model of the driver; that means that things which are supposed to be solid stay solid (woofer cones) and things which are supposed to bend do so linearly (driver suspensions). We'll start with the parameters we know, namely those given by the manufacturer or acquired from direct measurement, then determine the composite parameters and what they mean.

Mechanical Parameters

S_d (m²): The cone surface area.
X_max (m): The maximum linear excursion of the woofer, which although goes by different definitions, is generally the maximum excursion in which the voice coil stays in the linear portion of the magnetic field.
M_md (kg): This is the effective cone mass, which is comprised of the actual cone mass plus the mass of the volume of air surrounding the cone, which is related to S_d.
C_ms (m/N): This is the compliance of the driver suspension system. Compliance is the inverse of stiffness, known best as the "spring constant k, such that f = -kx" in that problem on that exam I failed freshman year.
R_ms (Ns/m): This is the mechanical resistance of the suspension system, it's analogous to the damping factor of the "dashpot" in the standard second order resonator.

Electrical Parameters

R_evc (Ω): The parasitic resistance of the voice coil.
L_evc (H): The actual electrical inductance of the voice coil. This is difficult to measure, because any current you put through the voice coil will cause the cone to move, thereby inducing a back-EMF from the magnet structure. It's most common to measure this at a very high frequency well beyond the cone resonant frequency, but this leads to other problems as it is a very nonlinear (frequency dependent) parameter.
Z_pk (Ω): Maximum driver impedance, measured at the driver resonant frequency F_s.
B_l (N/A): This is an electromechanical parameter, really. It's the efficiency of the linear motor, the force per amp of the voice coil. It's a function of the number of turns of the voice coil, the area of the voice coil, and the flux strength in the gap.
P_e (W): This is the rated power handling of the driver.

Let's list the mechanical equations which govern the cone motion. If we define x to be the cone displacement (along the axis it is supposed to move) then for the cone mass we have:

which can also be written as

The compliance of the suspension exerts a restoring force on the cone by the function,

But if we differentiate once this becomes:

And finally the mechanical resistance of the suspension resists the cone motion with a force proportional to the cone's velocity:

At this point, we need to decide what sort of analogy we want to make to circuits. The impedance analogy says velocity is equivalent to current, and force is equivalent to voltage. The mobility analogy says that force is equivalent to current, and velocity is equivalent to voltage. The impedance analogy is useful for acoustic circuits, because you get series circuits which have physical correspondence to the real world; the mobility analogy is useful for visualizing mechanical circuits, where there is an almost visual correspondence between KCL, which says all currents into a node sum to zero, and the law of conservation of momentum, which pretty much says the same thing but with different variables.

It goes without saying that all our results should be valid regardless of the analogy we make; as a matter of fact, the resultant network under one analogy can be transformed to the network under the other analogy by taking the network conjugate. Still, for ease of visualization I will choose the mobility analogy. Under this analogy, we can think of circuit branches as linkages through which force is transferred, which isn't too far from what's actually happening:

Under this analogy, the following transformations are made:

Now let's consider how to model the linear motor which drives the cone. The way I think of it is as an ideal electromechanical transformer, with a B_l:1 turns ratio. That is, for one amp of current flowing into the dot on the primary, B_l×1 Newtons of force flow out of the dot on the secondary. Thus the complete driver model looks like this:

Now we are ready to investigate the composite T/S parameters, namely F_s, Q_ms, Q_es, Q_ts, and V_as.

F_s, Driver Resonance Frequency

The cone mass and suspension compliance form a second order resonator, and as such it has a resonant frequency. We know the resonant frequency of an LC tank, it's given by the following equation:

Q_ms, Driver Mechanical Q

Loosely speaking, the Q of a system is a measure of how lossy it is. To measure this quantity, you need to know both how much energy it can store, and how much energy is dissipated. Note that nowhere in this definition is the requirement that the system is any sort of oscillator or resonator; an RC or LC has a perfectly well defined Q. Of course, the value of Q will be frequency dependent, since L and C impedance is a frequency dependent parameter. For resonators, we have a more specific definition of Q, since resonators are designed to operate at a particular frequency, the Q is defined as the energy stored by the energy lost over one period at the resonant frequency.

For a parallel RLC, the Q works out to be defined as

where ω_s is the resonant frequency in radians^-1. Looking at the mechanical portion of our circuit (ignoring the transformer action and back-EMF for a moment), we have the following expression:

However, there are often times that R_ms is not given, as driver datasheets are notoriously non-standardized, and some manufacturers think an illegible handwritten chart is enough. There is a way to compute R_ms using what we know about resonance and transformers. At mechanical resonance, the inductor and capacitor cancel each other perfectly, so the circuit looks like this:
But using the impedance transformation properties of the transformer, we redraw the circuit as
So the value of the impedance at the resonant frequency is simply Z_pk = R_evc + 2πf_sL_evc + B_l²/R_ms. With woofers which have resonant frequencies in the 100Hz range and lower, the impedance from the inductor is negligible, so we simply have two resistance terms. It's very easy to measure Z_pk with a multimeter and frequency generator, and R_evc as well; therefore, an alternative formulation for Q_ms is as follows:

Qes, Driver Electrical Q

Just as the suspension losses dampen the cone oscillation, as the cone oscillates a back-EMF is generated in the voice coil, which is then dissipated in R_evc. Since we assume that we are driving the voice coil with a voltage source (not always valid once the crossover network is introduced) the circuit can be redrawn as follows, this time focusing only on R_evc and ignoring the effects of R_ms, which we have accounted for with Q_ms.
Again we can use the transformer to derive the equivalent resistance seen at the mechanical resonator side, resulting the the following simplified circuit:
From this, we can see immediately that

Q_ts, Driver Total Q

The formula we're given for Q_ts is
But what does this mean? We'd expect Qts to be the driver Q accounting for both mechanical and electrical damping, let's see if this is true. We substitute in expressions for Qms and Qes:

This is the identical expression as if we'd computed the Q of this circuit directly:

Which makes intuitive sense.

V_as, Air Volume Equivalent to Driver Compliance

I'm not sure what the "as" stands for, but I would guess it's "acoustic suspension". Way back in the day, speakers were designed such that the drivers were mounted in sealed enclosures large enough so that dominant restorative force on the driver cone is provided by the suspension of the driver, not the springiness of the air pressing against the cone. This is called an "infinite baffle" enclosure. Then in the 50's Henry Kloss and a few others turned it around, and built the cabinets smaller so that the air pressure on the cone dominates over the suspension force (hence the "acoustic" in "acoustic suspension) which has the main advantage that it is a much more linear spring, thus lowering distortion levels. Anyway the V_as value is the volume of the cabinet such that if it were any larger the speaker would be infinite baffle, and if it were any smaller it becomes an acoustic suspension. So let's derive the expression for V_as.

Starting with atmospheric pressure which is ρc² where ρ is air density and c is the speed of sound at normal conditions, we know that for well behaved gasses (air, at STP)

So if we move the cone by Δx, that displaces a volume of S_dΔx. This results in a pressure change inside the cabinet of

which when multiplied by the area of the driver gives the total force on the cone:

But, we want that to be equal to the suspension restorative force, which is given by

So if we equate the two, we end up with:

So how is all this useful? Well, I want to eventually write a program to do loudspeaker design, like LEAP. This is the starting point, now that we have a model of the loudspeaker driver it just takes some adjustments to add crossover networks, cabinet effects, passive radiators, etc.. That is, provided I don't have to work on my thesis.

Bibliography

Beranek, Leo L., "Acoustics", New York : McGraw-Hill 1954
Thiele, A.N., "Loudspeakers in Vented Boxes, Parts I and II," J. Audio Eng. Soc., vol. 19, pp. 382-392 (May 1971)
Thiele, A.N., "Loudspeakers in Vented Boxes, Parts I and II," J. Audio Eng. Soc., vol. 19, pp. 471-483 (June 1971)
Small, R.H., "Direct-Radiator Loudspeaker System Analysis," J. Audio Eng. Soc., vol. 20, pp. 383-395 (June 1972)
Dickason, Vance, 'The Loudspeaker Design Cookbook", Audio Amateur Press: New Hampshire 1995
Leach, W. M. Jr, "Introduction to Electroacoustics & Audio Amplifier Design", Dubuque: Kendall Publishing, 2003
Image taken from: http://www.djsociety.org/Graphic/speakerdriver.gif

The Lizzyville Coaster: Finished!

2007-12-03T02:40:00.000-05:00

Did you want to start at the beginning?

Here is the finished bicycle, the one-of-a-kind 2007 Lizzyville Coaster. A few more shots from different angles:

The life-saving Nashbar long-reach caliper. Certainly for the price they're very nice brakes, being dual-pivot and all. They also allow a large amount of clearance. Although the bike's currently got 700x23's on it (a tad narrow I know) it probably supports up to 700x35's, which would make this practically an off road bike.

Here's the rear drivetrain. You can see the pine-cone lacing, but it's really only noticeable when looked at head-on, which is a pity. I usually do a round-up summary at the end of my projects, but I think I'll skip it this time, the bike speaks for itself. Maybe I'll just do a quick one...
What went right:

The paint job looks pretty professional, only a few runs and minor mottled patches.
Finding parts on eBay is pretty easy, as long as you take your time.
Having built several bikes before helps, as you've got a stock of spare parts.
The Vuelta rims are pretty, and seem very tough.
Having a basket is awesome.

What went wrong:

No matter how pretty the paint is, it's not durable like a powdercoat. It'll be 6 months before it's destroyed.
The decals look like they were drawn on by a 3-year-old with no hands.
Using a frame that's so old you need special long-reach calipers.
Although steel's a nicer ride, I probably should have used an aluminum frame, as the rainy winters will rust the bike out quickly.
A bit heavier than I would have liked, but that's probably only in comparison to my bike.

All in all, a worthwhile addition to the team of projects here at commonmode.

This is the bike as it looked waiting to be delivered. Just chilling in sub-zero weather in Park Slope, at 11am on December 1, 2007.

And for a moment of levity, on the way back from delivering the bike, a block from my apartment on 114th street in Harlem, I came across this. I think the moral of the story is New York will break your bike in half.

I'll miss you Lizzyville.

A month or so ago, it happened that I needed a floor lamp for my room. I was coming back from lunch in midtown and I saw a huge flashy "sale" sign emblazoned on the window of Restoration Hardware. I didn't know anything about Restoration Hardware, so I walked right in expecting to find a great deal on a floor-standing torchiere. I picked one, a reasonably understated modern design with a halogen bulb and dimmer, and asked the sales clerk how much it was. Oh, she replied. You're in luck, today it's 33% off. And that would make it...? $699 for the floor model.

The reason I bothered to bring this up at all is while I was browsing the rows of illuminating devices, I noticed next to particularly expensive items were small stacks of brochures. In these brochures was practically nothing useful/pertinent to aid you in making your purchasing decision. Instead, it was filled with "glamour shots" if you will of the item in question set in various lights and poses, with various good-looking people. It seems as if when you're about to put down a month's rent for a light fixture, you're buying not only the ability to see your own two hands at bedtime but a little chunk of status as well. Gosh, I thought, this here bike is certainly swanky enough to warrant a brochure, after all it's got status and sex appeal written all over it. Well, you can see how that little idea turned out here.

The Lizzyville Coaster: Assembly

2007-12-03T02:33:00.000-05:00

Were you looking to paint a frame?

Ahh, truing wheels while simulating a high-Q fourth order switched capacitor bi-quad.

Here's where the creativity comes in: I wanted to do a unique wheel lacing pattern. I originally intended to do a snowflake pattern for the front wheel at least, which is beautiful but totally impractical and hard to keep true. Since I can't hang around to keep the wheels true (do you know of a bike shop that would true a snowflake wheel?), I opted for "two-leading, two-trailing" also known as "pine cone" lacing. It's still eye-catching, but totally practical as it's really just a re-distribution of 2-cross wheel lacing.

It's pretty when looked at head-on, but that's about it.

A direct closeup of the rear hub shows the pattern. Unfortunately it's a whole lot more pronounced on the rear hub, which is high-flange, than the front. The truth is that "3-leading, 3-trailing" is a whole lot more impressive as it groups all the spokes into only 3 groups of 12, whereas this wheel has 4 groups of 8.

Before anything, unpack the frame and make sure it wasn't destroyed by FedEx.

The actual bike assembly is straightforward. First, I greased and installed the bottom braket and fork. It's important to liberally grease the headset races and bearings. Then you take the cranks, bolt them on, put the wheels on, measure the proper chain length and then attach the pedals. Mount the saddle to the seatpost, grease it and insert it into the seat tube. Mount the brake calipers on the frame--oh, nuts.

Oh, SNAP.

If it's not clear from the photo, these calipers are too short for this frame. What I hadn't taken into account is this frame was designed for 27" wheels, which are just a tad larger than 700c's. So I needed long-reach calipers. I was a week away from a deadline, so I went into panic and ordered a pair of Nashbar long-reach calipers for $24.99, and had them over-nighted. Crisis averted.

Something's still missing...

Continuing onward, grease the stem quill, insert into steerer tube and tighten. Bolt the bullhorn bars (which are really hacksawed drop bars) to the stem. Attach brake levers--awh, nuts. The problem is that a standard handlebar diameter is around 24mm, but flanges to 26mm at the bolting interface to the stem. The brake levers bolt to the 24mm section, except that on this handlebar the 26mm section extends left and right in both directions, making the lever placement awkwardly wide. It's the day after Thanksgiving, and you desperately need to assemble a bike and you're in Manhattan. What do you do? You go to Larry's 2nd Ave. Bicycles Plus, the best bike store I've found so far in Manhattan. Not only do you get the usual selection of high end parts you'd expect to find in a boutique store, there's a huge assortment of secondhand parts. Moreover Larry speaks French and Spanish, in case you have been reading this blog and haven't had the slightest clue what's been going on. I got new drop bars (which I promptly sawed down to bullhorns) for $8. I also bought grip tape, cable ferrules and a bike basket there.

After these initial hiccups, it was smooth sailing thereon out. Mounting the basket to the fender bosses on the front fork, routing the brake cables, fixing the brake cable down with zip ties (I'm a prize idiot, we know) then wrapping the bars with grip tape, and the bike is more or less done.

Click ahead to see the finished product...

The Lizzyville Coaster: Painting a Frame

2007-12-03T02:31:00.000-05:00

Wanna know where I got all this junk?

Before even thinking of painting, I did two things to the frame. First, I removed the brake cable bosses from the top tube, reasoning that they were ugly and we didn't need a rear brake (which is patently wrong, as this is a coaster and thus has fore and aft brakes) [no one should ever say I'm smart] so I guess I'll have to zip-tie the brake cable housing to the top tube. Speaking of ugly. Second, I sawed off the rear deraileur hanger, as that's something we really won't need, and it makes the bike more symmetric.

Dremel tool use #1001: making irreversible mistakes faster than you can think them through.

So we've got a grungy frame that's 20-odd years old, and we want to re-paint it. How do we go about doing this? First, we need to clean the excess grease off, any of the variety of orange grease cutting products work fine for this purpose. Once we get it more or less clean, we start sanding. I used a combination of 150 and 220 papers. Here you can see the four different levels:
bare metal, the white primer, turquoise paint, and clear topcoat.

Yeah, I was pretty unhappy at this point with the lugged frame.

That's all well and easy for the straightaways, but the lugs and stays were a total nightmare. After about 10 hours of laborious sanding and swearing, I got the frame and fork more or less down to the original primer. I didn't want to go for bare metal all around, since I felt I might weaken the frame by taking off too much metal in some places.

An almost naked frame.

My primary concern is the long-term durability of the finish. Haha.

First I put on several coats of primer, then let it dry for a week. During the week, I went back to New York for school and stuffs. While there, I researched how to best put a durable layer of color on this frame. It seemed that the best solution is a baked-on powdercoat, which is more or less like the glaze on ceramics, and is super durable. The only drawback: very toxic, and oh, did I mention the need for a large oven? So I did the next best thing: automotive epoxy paint. They make low-temperature curing finishes, and the guys promised me it's the hardest spraypaint there is. Problem: still need a spraygun and air compressor. So, a trip to Sears covered that. The irony here is that the bike frame cost around $50, and the paint ended up costing $150, and the compressor and spraygun also cost a total of $150. At least now my mother can fill her car tires faster and with more precision than ever before thought possible.

Either I was "Ghostbusters" for Halloween, or I'm in way over my head.

Anyway, I went ahead and bucked conventional wisdom by doing the darker color (blue) first, since I wanted to do the logos on top in orange. Here's the first coat drying, so far so good.

This color is unofficially known as "masking tape blue".

I had a vague idea of how I wanted to do the logos. I was going to freehand transfer the logo from a printed image to masking tape, then use an exacto knife to cut the stencil out. Then, out of the blue, the logo changed.

Hahaha uhmm... oh eff.

Daunted but not discouraged (read: foolhardy), I went ahead and tried to do this anyway. Here's the stenciled masking tape, ready to be traced out with a knife.

I'm sure there was a better way to do this, and I don't want to know what it is.

It was by and large an unmitigated disaster. The loops of the L's, e's, and y's and z's were closed. My eyes also started to tear from the intense concentration during the cutting/tracing. Once those were cut out, I sprayed the second base layer of color.

I'd say this is more of a pumpkin orange.

I let this dry for four times the recommended hour, since I had some fine detail around the logos and I didn't want to peel off any paint when I removed the logo masks. Once I did, it came time to apply the final clearcoat, which is basically liquid epoxy to put a durable hard finish on the paint. Once I finished coating the frame, I had some extra left in the spray gun, so I decided Arches and Parmesean cats could use a "durable, weatherproof finish" and I clearcoated them. Kidding!

I'm clearly neglecting some other work doing all this.

Here's the final finished frame. Honestly, I goofed up a good deal, it's far from perfect. There are several epoxy runs because I didn't properly adjust the spraygun to account for the higher viscosity of the epoxy versus the base colors. I also make the mistake of painting the frame in too close proximity to the fork, resulting in a somewhat matte finish in some areas. due to they spray-off painting one on the other. Nevertheless, it was a good effort. Here are some detail shots.

Let's put it all together!

The Lizzyville Coaster: How to win at eBay

2007-12-03T02:30:00.000-05:00

Want to go back one?

With the surging popularity of the fixed gear cycle comes hordes of vendors looking to make a quick buck off newcomers who don't know better. Just because a fixie has fewer parts than a conventional road bike doesn't mean it will cost any less, and by tapping into the fanaticism and mythos surrounding these bikes they've managed to open a whole market for parts which perform the same function as extant products, but at absurdly inflated prices.

Take for example the front hub. Unlike the fixed rear hub which is fundamentally different than a casette hub, all modern forks accept the same 100mm front hubs. Knowing this, and assuming you're going to be building your fixie from scratch (versus a conversion, where you get the whole front wheel for free), are you going to choose a top-of-the line road hub like the Campagnolo Record front, which weighs 116 grams and retails for $94.99 at Colorado Cyclist--notorious for selling things at MSRP in this internet markdown day and age--or this Phil Wood high-flange track hub, which weighs 225 grams and retails for $138? As a matter of fact, the Ultegra 6500 front hub is currently on sale for $19.95 at Bike Nashbar and weighs 197 grams. Although it's unfair to compare full retail prices with sale prices, it illustrates a related point which is that boutique items rarely go on sale, whereas you can't Google without finding a sale on last year's road component group.

On the opposite end of the spectrum, and this is not a phenomenon unique to fixies but more generally tied to the explosion of eBay as a venue for re-birthing secondhand items, the word "vintage" is finding new use in place of the word "junk". Although the heyday of US velodrome racing was between the two world wars meaning many high-end parts aren't made anymore, unless you're interested in owning a piece of history (a legitimate endeavor in itself, but then maybe you should think twice about riding it around Manhattan) there's no real reason to pay triple the price for something that's available newer but lacks the "vintage, has been collecting dust for 50 years behind someone's lawnmower" flavor. Even worse is when they find things that were already crap when they were new 50 years ago, and try to pass them off as high-end collectors' items.

Perhaps the best advice for online shopping is caveat emptor, which is sound advice for life itself.

COLLECTING THE PARTS

The best way to buy parts for a project like this is slowly. Impulsive bidding is the best way to get a bad deal; usually the thing to do is search one item, add a dozen of them to your watch list, then let them all end. This gives you a good idea on the price spread, and from this you can self-impose spending limits for the particular part.

STRUCTURE

This is the most interesting sale, and as such should be done first. The reason it's interesting is it most likely the most expensive, and also might come with some extra parts. It's highly improbable that someone selling caliper brakes will "throw in" a set of hubs and pedals, but with a frame sale any of those are possibilities. The frame I ended up getting is a 52cm Schwinn Traveller frame, constructed in the mid-80's. I did a bit of research on these frames, and the general consensus is they're excellent durable frames, if weight isn't a big concern. As this isn't anything near a racing bike and won't have the deraileurs or cassettes to weigh it down, this frame should suit the purpose just fine. This frame came in decent shape with a fork, headset, seatpost, crappy seat (I threw it out), stem, handlebars and brake calipers. The headset is in excellent condition and I suspect it's newer than the frame itself; the bearings are round and the races are completely unmarred.

What a delicious color... not.

When I build this bike up, depending on the final geometry taking wheels into account I may have to swap stems, but for now I plan on using the stem and bars the bike came with. For saddle I just pulled one from my pile of parts at home.

DRIVETRAIN

And this right here is pretty much reason enough to go single-speed.

I'm partial to Shimano Octalink® cranks. Unlike the conventional square taper cranks, the Octalink system uses an oversized axle, and engages the two cranks with a toothed interface, and on the whole feels a whole lot stiffer. One drawback, and I've noticed this on my Dura-Ace Octalink track-specific cranks, is the crank bolts can undo themselves due to the backpedaling from braking. This leads to some play, which can quickly ruin the cranks which are made from a relatively soft aluminum alloy, as they're bitten into by the steel bottom bracket axle. Fortunately this isn't a problem with a coaster.

Anyway for the cranks I splurged a bit. I got an Ultegra FC-6500 39/52 double crankset for $28.07, which means I don't have to buy a separate chainring. Ultegra's a bit high-end, but at that price they're a steal and it's a sexy looking crank. The other advantage of going with a Shimano road crankset versus a "vintage" track-specific crankset is the Shimano BCD of 130mm ensures there will always be a availability of different sized chainrings to experiment with. The crankarm length is 172.5mm, which is longer than the traditional track length of 155-165, but I felt that since it's a coaster, having a little extra torque at your disposal is helpful. Think of it as a more user-friendly introduction to single-speeding.

To mate this crankset I got an Ultegra BB-6500, for which I paid $31.01. That's steep and arguably the worst deal I got, but I justified it by saying that the bike's not going to go without it and it's a good bottom bracket. What actually happened is I forgot to bid on an Shimano 105 BB which ended up selling for $17, and in a fit of frustration I overbid on the Ultegra BB so here we are. When the cranks arrived, I discovered they didn't come with bolts, so I headed down to a local bike shop (LBS) Missing Link, where I got a pair for $6.

The pedals and chain I had extra from my pile of parts from home, so they came free. As mentioned before, the 16 tooth freewheel also came from my pile of parts.

WHEELS

It's actually a two-pack of hula-hoops for midgets.

This is the most exciting part of the construction, as there's most room for originality. The rims I chose were Vuelta Stylus rims, which are cheap "beater" aero rims. I chose them for two reasons: at $60 a pair new, they were a reasonably good deal, and second they are red and matching. Never underestimate style as a factor in a single speed design. I originally wanted to go with a Mavic Open Pro front and MA3 rear, which are unquestionably better rims, except then I had to find 36 hole hubs, for which no good deals existed that I could find, so in the end these were more cost-effective. The other advantage is they're pretty much bomb-proof, a necessity in New York. For hubs I ventured outside of eBay and got a surplus Specialized front hub for $5.99, and a sale re-branded Dimension flip-flop rear hub for $29.99, both from Bike Nashbar.

High-flange bolt-on rear and a low-flange quick-release front.

Spokes are an interesting question. They're impossible to buy off eBay, because for a given rim and hub configuration, the spoke length needs to be specified down to the millimeter. Good online programs exist to aid in spoke length calculation, although my favorite is the Excel spreadsheet program Spocalc. You really don't want to be buying spokes used anyway. For the spokes, I found amazingly that Colorado Cyclist had the best deal, at $0.50 per 15ga straight spoke, sold in increments of one with brass nipples. I needed 267mm and 275mm, and I was just about to buy the spokes when I remembered that I actually have those lengths by DT Swiss (go to the page, it's got a cool intro movie) in my bottomless pile of parts. What's even better is instead of straight 15's, they're 14/15 butted spokes with alloy nipples. They're actually way nicer than the application calls for, but hey when they're free...

The rim tape, tires and tubes I also had in my pile of goodies, so these came free too. You might begin to imagine that I have racks upon racks of surplus parts of every possible variety (imagine if you will the first scene in The Matrix where Morpheus takes Neo down the armory in his introduction, and you get the idea) and you'd be pretty much right. I'm a pack-rat much to everyone I know's chagrin.

BRAKES

Oh, stop it.

Because the frame came with calipers, I didn't have to worry about them. It looks like these are actually calipers off a Trek 1000 series bike, which would make them pretty snazzy for a bike like this. Unfortunately (this is a recurring problem with eBay) the calipers were missing the cable adjustment ferrules, so I went to another excellent LBS Mike's Bikes and they gave me a pair for free! All I needed now were levers and cables. The cables, like spokes, are cut to measure so can't be bought used. Fortunately they're pretty cheap at $3.99 each from Nashbar. The levers are Cane Creek in-line levers designed to give drop bars some stopping flexibility when riding erect. (oh stop) They come in two variants, machined aluminum and carbon fiber. As appealing as saving 9 grams per pair with the super sexy carbon, it just didn't justify the $30 premium they called, so I got the aluminum pair for $18 on eBay.

Now that we've got everything we need, are you ready to paint a frame?

The Lizzyville Coaster: Inception

2007-12-03T02:13:00.000-05:00

If you hang out in the city you can't escape noticing these odd looking bikes, often straddled by even odder looking riders. Unlike the bike your midlife-crisis afflicted dad just bought, which is described by a plethora of buzzwords (OCLV, V-brake, Octalink, SUP, SPD, Exa-Drive, Hyperglide, SiS, ISIS, Hollowtech, every year there's a new few) the beauty of these bikes lies in their singularity of purpose. Designed for the Velodrome (a bicycle racetrack), they can only do one thing: go very fast on smooth, flat surfaces. They don't have any braking mechanism--save for jamming backward on the pedals, since a fixie by definition can't coast--because during a race where riders are speeding upwards of 40mph inches away from each other, one person stopping suddenly can end disastrously for all.

The evolution of these tuned, sleek machines took a sudden turn when someone decided to bring one out onto the streets: a decidedly more hostile environment for both the bike, which is accustomed to well-maintained surfaces, and the rider, who has to deal with cell-phone talking Starbucks-carrying drivers oblivious to anything not blinking on a Blackberry screen. Numerous pitfalls notwithstanding, there are many valid reasons why fixies are great for city riding:

Less to worry about while riding. With no gears to change, you'll never think about them! Besides, in most flat cities low gears aren't necessary.
Less likely to get stolen. This is less true now than it was 10 years ago, thanks to the boom of fixies and the ingenuity of the aftermarket part vendor devising ways to make something as simple as a chainring look precious. (The word "bling" comes to mind.) In the past though, a bike on the street with one speed was usually a piece of crap, the money hits were the carbon-titanium-aluminum kitchen sinks.
Easier maintenance. I don't care what people say, maintaining a multi-speed bike, especially one with index shifting and a compact 9 or 10 speed drivetrain is a lot of work. Something's always scraping, getting out of true, or not shifting on a dime. Worse, even if you manage to find a sweet spot, all it takes is a small bump against a parking meter or a Chihuahua getting caught in the chain to send it all to hell. Not so with a fixie! All you have to do is check your chain's in good condition, and grease the bearings once a month or so.
Good workout. Stopping the bike with your legs is hard work! I've heard this is detrimental for real riders because it teaches bad habits and messes with the circularity of your stroke, but for regular folk I see no such drawbacks.
Teaches excellent riding technique. This is what fixie riders elaborate upon ad nauseam, unfortunately they're right. If you decide to get on a bike without brakes and take it out into the street, provided you're not flattened in the first few days you begin to develop a heightened awareness for the road. You stop looking a body's length in front of your tire and start analyzing activity two, three blocks ahead. If you watched too many action movies as a kid, you also start imagining targeting reticules floating around your vision, an HUD displaying pertinent tactical information and a three-dimensional map overlay (think Google Earth).

One must also keep in mind that it's possible (recommended) to ride a fixie with a front brake and a single brake lever. This is often colloquially referred to as the "oh shit handle" or some variant thereon.

Following this brief and altogether too opinionated introduction to fixies, let's turn out attention to the particular bike we'll be assembling.

DESIGN GOALS

The rider is a 5'8" female, about 115lbs. Thanks to her litheness I can relax a bit over the abuse the bike will take, although that's only half the picture as the bike will be ridden around Brooklyn and Manhattan. As the introductory vehicle to single speed riding, I'm opting to go for a single speed freewheel, which means you can coast. There are two reasons for this. One, the initial fixed gear experience is disconcerting. It's like one of those volunteer psychology studies you did in college, where some grad student grasping at a meaningful dissertation topic puts you in a room and make you play a driving video game except the screen's reversed so left means right and vice versa: completely doable with all your concentration, except after 10 minutes you have a splitting headache. Being a gift, I'd like the initial experience to be something other than "nauseating". I dunno, something positive, like "collarbone breaking". Kidding! The other reason I don't want to make it a fixie is that sometimes you want to use the bike to get somewhere without expending a lot of energy. Call me out of shape, but when I ride my fixe I always work up a bit of a sweat braking all the time. Maybe I have poor fixie riding technique, and I should really be coasting through everything totally disregarding the rights of way of others. Whatever. The point is, in the interest of efficiency it's going to be coaster. I'll use a flip-flop hub so that when curiosity strikes there won't be any technical obstacles to going fixed.

To choose the right frame, we need to analyze the rider's body geometry. Fortunately I have a good photo for that purpose. With judicious application of Photoshop, we arrive at the following result:

Help, I'm a ghost!

As I said, this is going to be a freewheel affair. As such, it's commonplace to use a ratio slightly lower than comfortable on a fixed, since going downhill fast won't require spinning the legs. I have a spare 16 tooth freewheel, so I need to determine a chainring size to mate that. I get by just fine with a 48/16, which gives an even 3:1 ratio. The rider in question hasn't done a lot of riding recently, so to make things a bit easier in the beginning I'm going to go with a 42/16 or 39/16. Since 39 is often the second chainring on a multi-speed crankset, I'll go with 39/16, which yields a ratio of 2.43:1. To get an idea of the cruising speed, if we assume there's going to be a 700x28 tire, the approximate wheel circumference is 84 inches. If we further assume she's going at an 85 rpm cadence (she's a runner, and the optimal running/cycling cadence according to my triathlete friends is around 90) that's going to be 85 revs/min × 60 mins/hour × 2.43 rotations/rev × 84 inches/rotation ÷ 12 inches/foot ÷ 5280 feet/mile = 16.5 miles/hour. Okay that's a tad slow, but we're in New York, taking it easy never hurt anyone.

THINGS YOU'LL NEED

Let's go over the BOM (bill of materials). For the structure, we'll need:
A frame, fork, headset, stem, handlebars, seatpost, saddle.

Then the drivetrain, which is usually a whole complicated affair in a multi-speed bike, is reduced to the fundamentals in a single-speed:
Cranks, bottom bracket, chainring, chain, freewheel, pedals.

And this is going to be driving the wheels, which are comprised of:
Rims, rim tape, inner tubes, tires, spokes, hubs.

Next, we need some way to stop, so we're going to need:
Brakes, brake cables, brake levers.

Finally, the finishing touches:
Bar grips/bar tape. Basket. Decals. Lock. Lights.

Find out how to procure these items in the next installment, when we tackle the caprices of eBay!

Visual Acuity Exam

2007-10-30T23:23:00.000-04:00

I finally brought my speakers to New York! Can you spot them in the photograph?

New York Speakers Part 5: Construction

2007-07-10T09:41:00.001-04:00

Construction occurred over several visits home each lasting a day or two each, so the steps had to be partitioned and synchronized carefully. Given the abundance of time I had in a small apartment in Italy (without any power tools) I went to the pains of making cut lists in SketchUp, to minimize pilot error. Also, my dado set isn't the greatest so instead of using 3/4" MDF and rabbetting madly, I made that piece from 1/2" MDF and 1/4" MDF glued together. The rest of the 1/2" pieces have 1/4" deep rabbets cut into them.

SATELLITE CONSTRUCTION

I first did the satellites, since they're the simplest and I wanted to re-acclimate myself to the subtleties of my pathological table saw. It's all butt joints except the faceplate, which has a 1/4" rabbet all around. Both drivers and terminal cup are recessed for a clean look, which I did using a router and Jasper Jig.
Here the satellites are being clamped (poor man's clamp) and glued. While this was going on, I populated the crossover boards. As we've agreed, getting silkscreened crossover boards is a total waste so let's not beat a dead horse.
After the glue dried, I filled the seams with wood putty, and let them sit for two weeks to harden. Actually, I left and didn't come back until two weeks later (damn day job) but it certainly didn't hurt the finish. Often with wood putty people will find that if they apply paint too soon after filling then over time the putty will expand and press out the seams, leaving unsightly bulges. By re-sanding after the two week curing time, I hope I've minimized this possibility. Here're the crossovers and drivers being mounted:
I'm a big fan of PC-7 epoxy, shown above. It's got a great shelf life, easy handling (thickness of peanut butter, according to the press release), and cures to an incredible hardness. The crossovers are bolted to metal standoffs which are epoxied down to the bottom panel, and all the screw threads are loktite-ed to keep from vibrating loose. When all's said and done, here are the two satellites of my system, posed tastefully beside a bottle of sherry for size comparison.

SUBWOOFER CONSTRUCTION

As the satellites moved off the table saw to the paint booth (you'll laugh in a second when you see what exactly that entails) I began the arduous task of constructing a precision multi-chamber enclosure with my table saw, which from observation while building the satellites had gained 1/32" of skew over 24" since I last used it. After cutting all the pieces from the cut list I made in SketchUp, I had to trim a few pieces to get them all to play nicely together. I ended up deepening the dado cuts a bit reasoning that I could always reinforce the joints later with a filler of sawdust and glue. Test fittings:

The piece I'm holding is a support dividing the two chambers. It's got two 10-32 T-nuts embedded which will provide anchors for the removable cover. As shown they are being glued in place with Gorilla Glue, which for all the awesomeness that is it's space-filling expansion, requires a hefty bit of positioning as provided by the clamp and screw. The transformer and driver are simply making temporary appearances so I can test clearances. The finished raw cabinet looks like this:
You can see the silver T-Nuts in the holes. While I don't expect to be taking the cover on and off frequently, machine bolts as anchors provide strength that wood screws would lose after the first few removals. Now we move on to the painting process, and you can laugh at my interpretation of a painting booth:
A note on the finish. I had always wanted to try a micro-pebble finish, but from reading literature that would require too much fancy equipment to do right. Instead, I bought a gallon of DuraTex paint, and a few cheap textured rollers. The rollers are really coarse, and at first I didn't like the finish, but in retrospect it's durable as hell and doesn't took too bad. While this was drying, I moved on to...

PLATE AMPLIFIER CONSTRUCTION

I'll be honest, I like working with metal, but only when I have a full shop. I have a long chunk of extruded heatsink and some sheets of 6061 in the closet but I'll be damned if I'm going to cut all that with a hacksaw. My recourse is the mecca of all DIY project parts, I speak of course of eBay. I browsed until I found exactly what I was looking for: a surplus plate amp pulled from some crummy subwoofer. I paid $31 for it, including shipping.
I gotta admit that for a purported cheap plate amp, it's unique. Most low power plate amps use monolithic chip amps (guilty as charged) but this actually has a fully discrete class B amplifier, sure using TIP31 transistors, but it's the thought that counts. Anyway after 5 minutes this thing looks the same as anything else I touch for 5 minutes: in pieces and FUBARed.
Now comes the part that endears me so to my mother, where I take over an eating surface and cover it with live electricity and lead solder.
Note the champagne colored blowtorch in the middle of the table. It's just there to add to the "electricity poison fire" death mix and freak her out more. For all the heat-shrinking, I just used a hot air gun which provides much nicer results with less burning. Here's a close-up of the board during the final stages of testing.
The two regulators were getting hot on the crossover board (lower, parallel to the plate) so I put in my favorite kind of light-duty heatsink: two strips of aluminum flashing sandwiching the devices. And this is what the finished face of the amplifier looks like:
You'll notice that the text no longer corresponds exactly, but I did my best. The speaker level outputs are the left and right binding post pairs, with the red set for the right speaker. The volume control knob and phase control knobs still function, and the unmarked additional knob is a bass control. The line level inputs are new, but the power LED near the fuse functions as normally.

The plate amp only took a minute to screw into the subwoofer cabinet, and I was done. I ran it at high volume for a day (again, to the chagrin of my mother) to break in the drivers. A neat trick is to wire the satellites out-of-phase and face them inward, so the projected sound is greatly attenuated. Playing a mono organ music CD is the best way to hit the full spectrum, and the attenuation is maximal because of the mono signal. Note the impossibly cute iPod nano sourcing the tunes.
But you didn't come for just that photo. Here is the coup de grâce, the raison d'être for this speaker system...

HOW TO ENJOY YOUR JACK-IN-THE-BOX SPEAKERS

Appreciate the beauty and simplicity of only having to transport one cabinet. Once you've decided on a location to unpack...

Turn the box upside down to expose the four thumb screws holding the bottom panel.

Undo the four screws and remove the panel, exposing the inner compartment holding the satellites!

Remove the first satellite, and place it aside.

Remove the second satellite. Place the two straps for removing satellites back in the chamber, and reseal it with the thumb screws.

Enjoy!

THINGS THAT AREN'T PERFECT

Open-loop design for the amplifier PCB. I lost $100 for the PCB and about $25 in parts, just because I didn't do a careful design review. Don't ever ever design a circuit without testing at least something first. In simpler terms, don't be so arrogant as to think you can design everything right the first time. I've had professors that talk about "back in the day" when they'd do layout non stop with grid paper and pens, all the while subsisting on hot dogs and coke, and the chip would work the first time. There's a reason why they're the big shots, and that's definitely not me, and probably not you either.
Volume control noisy. The only dual-gang pots I could find are 100K, which when dialed to their midband position have a maximal Thevenin resistance looking into the amplifier. You can actually hear the speakers get noisier in the middle volume setting, due to the carbon resistor noise.
Adire Audio woofer. For all the hype surrounding its XBL2 motor technology, I'm pretty unimpressed with it's performance. I'm almost positive its sensitivity is several dB below advertised. When you factor in the price, it's definitely worth it to stick with Tang-Band drivers.

Other than these points, which admittedly aren't so minor, I'm very happy with the outcome. It's a project 7 months in the making, with all the design occurring at night after work in Pavia, and the assembly interspersed with each trip to the Bay Area. The fit and finish of the cabinet is very sturdy and presentable, and the snugness of the satellites in the woofer enclosure is perfect.

Sound wise I can't say definitively yet because the woofer is still breaking in, but the satellites are eminently musical. Thanks mostly to their physical footprint the stereo imaging is exceptional, and I find the highs present without being straining, which is my primary complaint with cheap tweeters. The lows are obviously absent, but assumed very smoothly by the subwoofer. I was hesitant to choose a fixed crossover point for the satellites, but I think I lucked out in that the listened response as well as the measured response is flat in the 100Hz region, after a bit of placement tinkering and phase adjustment.

Well, on to the next project!

New York Speakers Part 4: Drive Electronics

2007-07-10T09:35:00.000-04:00

UPDATE 12/31/07: I revised the crossover board to version 2.0. This includes a gain of 6dB in the subwoofer path, to compensate for a mis-calculation I made when I changed the subwoofer amplifier from a bridged amp to a normal amp (thus losing 6dB). I also changed the opamps to TL074 which is a mostly equivalent but lower noise version of the TL084.

If you're the type of person who doesn't recognize the validity of any circuit constructed using devices having more than three terminals (do you make your own water from hydrogen and oxygen too?) read no further. Otherwise, let's talk about subwoofer plate amplifiers. This plate amp's going to need a few things.

3 amplifiers, to drive stereo left right, and the subwoofer.
An electronic crossover.
A preamplifier, with separate bass control and phase reversal.

While I have built several amplifiers from discretes, both bipolar and MOSFET, class A, B and D, in this case space is a premium and the application isn't honest hi-fi. It's hi-fi given the size, which is why I chose to go with a monolithic amplifier chip, the LM4781. It's a 27-pin TO220 package with three channels of 35W each into 8 Ohms, from National Semiconductor's Overture series. Even with this chip though, I don't have enough room given the enclosure dimensions to fit everything on one board. So I'm going to have to split the circuit into two boards. One board will contain the rectifier/filter and power amplifier, and the other board will hold the preamplifier and crossover. The transformer is mounted inside the cabinet away from the electronics, to reduce EMI and gain space on the plate. The plate amp should look something like this:
The first thing to work on is the crossover board. I'm going to show the schematic, then explain it.
The input comes in through DC blocking capacitors, then goes directly to a header which connects to a dual-gang pot for volume control. It's then buffered and gained by 8dB, the output of which feeds the high-pass and low-pass filters. These are standard 40dB/decade second order filters, the high-pass filter is tuned to about 100Hz, and the low-pass filter is tuned to 250Hz. The reason for the asymmetry is the bandpass box is tuned for a high cutoff of 100Hz, so the low-pass filter is not strictly necessary; it's only there to prevent high-order vent resonances which sometimes happen in these alignments, when you rely entirely on the box geometry to do the filtering.

The subwoofer enclosure crossover is fourth order bandpass, the sattelite enclosures are second order sealed and the satellite electronic crossover is a second order which gives the combined response a fourth order shape as well.

In the subwoofer path there is a gain of 2 in the filter bank, this is to account for the lower sensitivity of the woofer. The gain is accomplished by making the filter values unequal; their values are chosen to maintain a filter Q of 0.5.

Following the low-pass filter is a standard inverter, which in conjunction with the non-inverting signal goes to a DPDT switch, allowing you to choose between 0 and 180 degrees of phase. I toyed briefly with the idea of implementing a continuous phase adjuster, but I've always been wary of the standard analog implementation of that concept (it basically takes the phase and puts it in a blender) so I ultimately decided against it. The output of the DPDT switch goes into a second dual-gang pot, which controls bass level. Resistors R2 and R3 are there to make the linear pot behave logarithmically, or at least somewhat logarithmically. (It's strictly a rational function that looks logarithmic at the volume where our ears are most sensitive.)

Next we have the main amplifier board, but this is uninteresting as it's just three copies of the datasheet suggested application.

You might wonder why I bothered to bring the bass information in as stereo, if I sum it right off the bat. Well, I have in mind to one day make a more powerful amplifier with two stereo amplifier chips, using one of them in bridge mode driven with the stereo bass signal, one in-phase one out-of-phase. By have in mind, I mean I made the PCB already but thanks to a stupid ground loop due to a misunderstanding of how the chip is wired internally, it's useless and I shelved that idea. Note that the 7.5 Ohm output resistors actually have parallel inductors that aren't shown on the schematic, this is just to quench high frequency oscillations although I feel they aren't strictly necessary as these chip amps are very well behaved.

This is pretty much it; the resistor values might look strange but I just chose the numbers I had on hand. You can make liberal substitutions save for perhaps the crossover values without degradation in performance. Just remember to use components rated for the load in the power section, and to use bipolar caps in the feedback resistor path...

Here is the PCB artwork for the amplifier board, and the crossover board. Just send this to AP circuits to get boards made.

New York Speakers Part 3: Satellite Design

2007-07-10T09:34:00.002-04:00

Thanks to the extra volume I got from making the sub a bit bigger, I could make real 2-way satellite speakers. I splurged a bit and chose a tried-and-true complement of Audax drivers, a 4" Shielded Aerogel Midbass and a 1" Textile Dome Tweeter. Although I've never used these drivers before, I've seen them paired in several designs so I felt it was a safe bet. Playing around in WinISD, the woofer seemed to work with both sealed and vented alignments, so due to personal preference I chose a sealed cabinet. The cabinet has an internal volume of 3L but with padding is equivalent to about 4L, for a roll-off around 90Hz to mate with the subwoofer.

Honestly designing a sealed box for a 4" woofer is not rocket science, whereas crossover design is always a challenge. Since I don't have access to commercial crossover design software, I decided to write my own Matlab program to auto-calculate crossover values, based on real response curves and a chosen topology. Here is the tweeter response:

And here is the woofer response:
From a quick visual observation, it looks like the tweeter is solid in the lows, while the woofer suffers some chunk around 2kHz, and then gets really squirrely around 5kHz. (Neither "chunk" nor "squirrely" are real engineering terms.) Even though I'm pairing a 1" tweeter with a 4" woofer, between which there is significant overlap, I don't want to crossover too low. Otherwise I'll risk damaging the tweeter or I'd have to use a higher-order crossover. In the interest of simplicity, I'll use a first order crossover for both the highs and lows, with a tweeter resonant peak cancellation and woofer inductive rise cancellation, and tweeter output attenuation to make it match the less efficient woofer. The script, downloadable as a .m file at the bottom of this post, also takes into account the offset acoustic centers of the two drivers when calculating the on-axis response. Due to a phase lag in the tweeter at crossover, the smoothest response was obtained when the tweeter's phase is reversed, something unconventional with this crossover topology. The final response expected is thus:
Since I was going all out for these speakers, I decided to have PCBs (silkscreened, no less!) made for the crossovers. I know, it's like half a dozen holes and four traces, and no one will ever see them. Whatever. I used my favorite--meaning, most fully featured yet affordable--schematic capture and layout tool, Eagle, to make the crossovers. The files can also be downloaded below.
The satellite SketchUp file is here, with a cutlist here.
The crossover matlab scripts including the driver response files are zipped up here.
The PCB art for the crossovers is here. The folder contains everything you need to place an order with AP Circuits.

New York Speakers Part 2: Subwoofer Design

2007-07-10T09:34:00.001-04:00

My starting point was the Cambridge Soundworks Ensemble IV, which is a passive 5.25" dual voice-coil driver in a 4th order band-pass box, whose longest dimension is around 12". Obviously a driver of that size has limited volume and extension, but it was tested to be "appreciably loud" down to 40Hz, not too shabby. However because my subwoofer was going to be powered, and because one of the chambers needed to be large enough to house two satellite speakers, I allowed myself a slightly larger cabinet and thus could expect greater volume and extension versus the Ensemble IV.

My first choice for woofers was the Tang-Band 6.5" woofer, whose Neodymium motor makes it lighter and have a smaller displacement, both major advantages for my system. It's designed specifically for these low volume multimedia "subwoofer" applications, which in reality are glorified woofers since bass below 40Hz is around zero. My first draft of the enclosure basically copied the Ensemble IV design, scaled up about 30%. It's got dual ports because the required cross sectional area needed to prevent vent noise would have necessitated a port longer than the enclosure itself. To get the response curves of the cabinet designs I use WinISD, a very neat free program.

In this design, the sealed chamber is the one that houses all the electronics. The vents pass through this camber into the adjacent one, something I needed to do because of the vent length requirement. This design supposedly rolled off at 45Hz, with a max SPL a bit under 100dBm. I actually had to resonate the cabinet a bit to get this volume, as this driver is pretty inefficient. The more I looked this design over, the more it nagged me. The chamber volume for the satellites was pretty small, so my satellite choices were either sealed cubes using 3" full-range Tang-band drivers, or a tiny two-way using a Hi-Vi 3" midrange and a Dayton 3/4" tweeter, neither of which had any low extension. I'd have to cross over around 150Hz, which would be high enough to make the sub pretty directional. I also did some playing with box volumes and decided I was well alway from the point of diminishing returns in terms of bass extension:box volume, so I should make the enclosure a tiny bit larger and get much better bass, as well as have room for real satellites.

Next I experimented with the idea of using two 5.25" Tang-Band woofers, in a more vertical cabinet. With parallel woofers, I can theoretically gain 6dB of SPL versus a single woofer, and these drivers are so compact anyway the box is barely any larger.
With the new enclosure I repeated the iterative process of satellite design. I've got to size the two chambers according to the equations for bandpass boxes, so that the upper-end roll-off mates well with the satellites, whose enclosure volumes depend on the volume of the sealed chamber. It's a darn over-constrained system, and on top of that it's non-linear! I decided to go with the same two-way system using the Hi-Vi/Dayton pair, but in larger sealed enclosures. My issue with this design is it's a bit inefficient, from a design point of view. Parallel woofers are used when greater extension and power handling are needed, and no single drivers exist which fit the bill. Given the relatively low-performance needs of my application, it seemed unreasonable to think that I couldn't find the perfect midbass driver.

At this point I remembered Adire Audio drivers. You might wonder why the link is broken. Well, it's not. No more then two months after I bought the driver, the company closed its doors. I'm an idiot. Anyway the Extremis 6" midwoofer is impressive; a long-throw design with 13mm Xmax, a stiff cone and cast basket with Neodymium motor. Adire was well known for their Shiva and Tempest woofers in the DIY subwoofer circles, and they applied their wide linear-range motor technology to a smaller driver. A single one of these has more output and power handling capability than a pair of the 5.25"s, so I went with this driver. Now I kinda wish I hadn't given the design will be irreproducible, but oh well it's unique. I'm sure with modifications a Tang-Band 6.5" can be used. Another advantage of this driver is the 8 ohm impedance, which means it will play nice even when bridge amplified with a chip amp, something I'll deal with later. Here's the final cabinet:

You can see the 3D modeling comes in handy here, as the vented cabinet is pretty crammed full of stuff, and it was good to see that the driver magnet has clearance from the plate amp components, something hard to visualize with pen and paper. Rather than listing the enclosure specifications, I'll just give the SketchUp files used to create it: the raw box with port, the cut list, and the driver model. The vented cabinet has about 30 cubic inches too much volume. I figured I'd just glue in some dampening blocks on the walls to consume this extra. Also, you can see the sealed cabinet, in which the satellites are stored, has a lip on the driver face. This is to protect the driver's rubber surround.

New York Speakers Part 1: Speakers in a Shoebox?

2007-07-06T11:19:00.000-04:00

For the past year I've been listening to all my music off Logitech mm50 speakers, one of those iPod based portable boom box thingies. For the size, I admit they sound pretty good and the built-in rechargeable battery is a huge convenience. Still, in the winter while visiting home I listened to full-sized speakers, and it reminded me just how much of the music I was missing. So I set about designing fancy floor-standing tower speakers which incorporated a Dolby Digital decoder I had lying around.

I wanted to try my hand at an all-in-one design, all the rage these days with boutique (but not necessarily good sounding at all) designers like Bose and Bang&Olufsen, that integrates as many independent channels as possible into as few boxes as possible. Designers cite purported advantages like easier room placement and higher SAF (spousal acceptance factor), but what they surreptitiously withhold is these boxes barely sound half as good as traditional separates, and the one remaining box they expect you to put in the audio and visual dead-center of your entertainment setup looks like the top of a naval carrier, with a drivers and sonic lenses protruding at every odd angle.

While I tried to avoid radar dishes on my speakers, I realized I would need to build cabinets with some non-right angles, and accurately drafting revisions of cabinets to determine things like box volume, driver cutouts and depth spacing quickly grows tedious. This prompted me to investigate good free CAD software for OSX, Linux and Windows. They all are terrible, and I mean absolutely terrible. I consider myself to be a CAD intermediate, having used a dozen or so Electrical and Mechanical CAD programs. Maybe I missed some good ones, but all the free ones are either too limited, utterly un-intuitive, or resource hogs. I guess for an actual good CAD tool, you need to pay CAD prices. Of course, you could pay top price (like, 200K a license) for Cadence and still get absolute shit.

In the end I settled upon Google SketchUp. For a free program, it's decently powerful, intuitive, and has some neat texture mapping features. You can also integrate your design into Google Earth, but seeing as I seriously doubt my living room is already part of the library, this feature doesn't serve me. My big gripe is its terrible interface with the back-end. It tries to be smart, in that it's got snapping links galore. It will try to figure out if you want a tangent, endpoint, midpoint, parallel, or orthogonal by how you approach the point you want with the cursor. For the most part it does pretty well, but what if you want to precisely orient something using parametric coordinates? That interface is practically non-existent. I can understand how maybe making those features an optional "expert mode" switch or something, but not implementing them at all is like a car company welding the hood shut so drivers never have to change the oil! It still drives me crazy sometimes. As you get used to the program, you start to figure out how the smart engine thinks, so to get the desired snapping point sometimes you have to zoom and re-orient the figure and then it works, but still it's a dumb workaround for a theoretically simple problem.

My other complaint with SketchUp is it's got an idiotic rendering engine. Everything seems to be stored as polygons; if you create a curved surface, one with 24 points will behave differently than the same one rendered with 48. Furthermore, because the curve is stored internally as an n-sided polygon, unless you find one of the corners of the polygon which lie on your curve, the smart engine basically cannot interact with your curve. For example, if you render a circle with an odd number of points, you most likely will be unable to create a tangential line which is also parallel to 2 out of 3 of the xy, yz, or xz planes. The corollary to all this is the engine gets easily bogged down when rendering the scene, because it's gotta render all those points. Maybe there's something I don't understand, but isn't a vector rendering engine the way to go? All I know is, none of the designs I'm doing should be considered at all "complex" and yet I need to hide half the layers before I can even rotate the axes. Stupid SketchUp. The more I think about it the more I hate Google; they put out their crappy terrible product, and then I use it all the same and even advertise it on their blog site. I hate them because they are geniuses.

Anyway in the end I got over myself and drafted out my new speakers. Each channel's got two chambers for the front and rear channels, plus a third side-firing subwoofer. With some heavy-duty digital filtering courtesy of Xilinx FPGAs, the dipole top speakers can direct sound in two lobes of our choice. Yes, directivity goes up with more elements, but it'd also like it to be a decent stereo speaker, so that takes precedence from a design perspective. Note the dimensions sketched on the diagram.

I was all set to build these over my visits back to Berkeley, and even started designing the control electronics (Xilinx offers a free WebPACK software which is great). Then, halfway through this dream I remembered that I'm moving to NY in the fall. After a short perusing of Craigslist, I realized I would be able to afford an apartment roughly the size of my coat closet at home. Immediately these speakers were out, because they need room large enough to take advantage of the radiating properties of the top drivers. This sent me down my current path.

My next thought turned to a 3 piece subwoofer/satellite system, which in my opinion are the best all-around choice for small rooms. I thought about just bringing the 3 piece system I made in MA, but recalling that the whole system together weighs 70lbs, I vetoed that idea. Shipping that system home was a nightmare, as the exposed drivers and delicate paint job required triple-boxing and still the paint got scuffed. So the main requirement, aside from decent sound of course, is easy transportability.

I've always been a fan of the old Cambridge Soundworks speakers, before they were bought out by Creative and now mostly sell plastic junk. Two of their systems which always appealed to me were the Ensemble IV and Model Twelve. The Ensemble IV was billed as an introductory home theater speaker set, 5 identical cube satellites and a "shoebox-sized subwoofer" with a sound that belied its size. It garnered rave reviews, especially when the price and size were taken into account. I actually got these speakers in 2001 as they began to sell out and clear their inventory of real speakers, for $150 which was a massive discount off the $500 regular price. I admit they didn't sound special in any way, and I would have actually preferred a good pair of two-way bookshelfs, but for a cheap home theater experience they were awesome.

The Model Twelve portable music system took the idea of a sub/sat system, but turned the subwoofer into a suitcase which carried the satellites when not in use. Thus the system could be carried around like any other rugged case, and when you wanted to set up you'd take the satellites out, re-close the case which was airtight, and connect the built-in amplifier to your audio source and the power supply, either 12V or 120V. It too gets consistently good reviews, and unlike the old Ensemble speakers which were just dumped altogether, this is still being sold albeit under the "Fleetwood Mac" name.

Thus my system is the combination of the two ideas. I like the bandpass subwoofer because it hides the woofer inside the box, and it's an efficient use of a woofer. With a fourth-order bandpass enclosure, one chamber is vented and the other is sealed. I designed the sealed chamber to have a removable panel, so you can stow the two satellites for shipping. The satellites are sealed two-way designs, using Audax drivers. Here's what the system should look like unpacked:

When it's packed up for transport, all you have to ship is one rectangular box (parallelepiped, for the geometrically inclined) with no protruding drivers, just a tough aluminum amplifier plate.
Over the next few posts, we'll look in detail at the various design steps involved.

Some Old Projects

2007-06-27T12:12:00.000-04:00

When I applied to grad schools two years ago someone asked me for some samples of personal projects, so I made a collection of pictures to post. I figured I'd integrate them into this site, since I'm going to reformat everything as I start writing more regularly again. Let's go through them chronologically.

Back in 1998, I was in high school and I'd caught the HiFi bug from my dad and a few of his friends, so I started getting into speakers and amplifiers and 5.1 sound and even laserdiscs. (remember those?) I decided to build my own 5.1 amplifier. Unfortunately I knew practically nothing about real circuit design, so I used amplifier cards put out by Marchand. I designed the enclosure and power supply though, and had the monstrous (1.2kW@80% duty) power transformer custom built for me by Tortran. The father of a friend of mine ran a machine shop, so they both helped me build the case out of scrap aluminum. Here it is looking at it from the outside:

And the back:

And inside:

You'll notice there's room for one more amp, this is because when I was designing this the Dolby EX standard was just coming out (I remember Gladiator being one of the first DVDs to have this) that called for a middle-rear sixth channel. There's substantial physical contact with case plates and the heatsink, so the whole case is one huge heatsink and barely gets warm. Here are the supply filters, on the PC boards I etched in my backyard, then reinforced the copper plating with strips of copper sheet soldered on for more current handling. Yes, that would be 225,000uF per rail, quite the overkill but it didn't stop me. At least I'm not going to go Gaincard and say that LESS filtering produces better sound. (idiots)

It works subjectively great, since back then I didn't have any fancy measurement equipment it couldn't analyze it objectively. It puts out 5x100W continuously without breaking a sweat, and weighs about 50 pounds. It has since been retired because my parents have been using my room to watch movies and the complication of a separate preamp, amp, surround decoder, DVD player is just too much. Now I have a DVD player and an integrated surround amp. Fewer cables is nice, must say.

Toward the end of my time in high school I got into old-school video games, namely the Neo-Geo. When I was little I wasn't allowed any game machines, but I remembered there was one machine, the Neo-Geo that was way cooler than all the rest. Of course this coolness fetched a price premium, and still did when my interest surfaced. Therefore I went about "consolizing" a game board built for the arcades. There are whole legions of people out there who do this so it wasn't anything remarkable, although I do feel my version is particularly easy on the eyes.

I finished this project about two weeks before leaving for college, and as a result never got to fully enjoy playing until I already outgrew my appreciation for it. I still keep it around as I now have a more general appreciation for what I consider to be "aesthetically pleasing electronic design," a quality which I feel is fully embodied by the Neo-Geo hardware, designed in the early 80's.

My first year at the university I started to build a good collection of electronic tools, and I needed a bench supply. A quick look at prices compelled me to build my own, resulting in this cute box:

Actually, the background is more telling than the subject itself: we've got Windows Media player playing some probably downloaded music (I forget the dominant file sharing program of the time, I think it was still Napster), we've got the obligatory crappy blue plastic cup, some variant of which is purchased by all freshman students their first week of school, we've got the Palm V which sadly didn't get any use until it was painfully outdated, there's a corner of an Nsync poster (no excuse there), and coolest of all we've got a Sun mouse, since my secondary computer was an Ultra 5.

A quick peek inside shows the somewhat poorly laid out interior, including the circuitry to drive the variable supply and the current meter with output select.

My second year in school I was convinced I should be a digital designer, so I took the eponymous class. In the end I made a "digital turntable" by which you could control the playback pitch, forward/reverse of a CD as if it were a record, meaning scratching effects were possible. This was modeled after a product by American DJ which had just come out that year, following my vague interest in such endeavors. I think the coolest part of the whole thing is how the record speed is tracked, with a stripped optical mouse. Other than that, it was just a bunch of PALs and CPLDs. It should be noted that the whole thing could fit on one 10K gate FPGA, which we had available, but I declined to use as I enjoy routing pretty bus lines.
A year later I'd taken an E&M theory class, which effectively shattered my understanding of "current needs to circulate in a loop" as I understood it, since I never previously considered "thin air" as part of any closed loop. It should be noted that three RF chips later, it still blows my mind. Anyway, I took it upon myself to design a Tesla coil that would be: small, cheap, and pretty. These parameters satisfy my engineering frugality without conceding any aesthetic standards.

In the end I paid about $40 for the whole thing, the majority of the cost absorbed by the capacitor bank because I didn't find any good surplus caps at the moment. A pity that right after I found a few pounds of even better capacitors for a handful of singles, but that's life. The primary transformer came from surplus, the spark gap I fashioned out of two dorsoventrally drilled bolts and two pieces of tungsten welding electrode.

The capacitor bank is one of those series parallel things, the input filter I made from parts salvaged from speaker crossovers, and the secondary transformer is made from copper drain pipe, a pill bottle wound with #30 wire and a doorknob. It's a pity that I'll never see full performance out of this Tesla coil because of secondary to primary leakage, which I unsuccessfully attempted to mitigate with the plexiglass shield. Fortunately the secondary is being rewound, the story of which you already have half seen with the coil winder project... I'm also planning on using an Alpha CNC lathe to turn a toroid topload. In due time, in due time. All the same, it makes some nice sparks.

The next summer I took a short break from electronic things, and did some machining. Well, it's not completely free of electricity but the majority of the hurdles in this project involved a big lathe from pre-World War II and a milling machine. Here's the rotor of a 10KW generator:

and what it looks like as it's being assembled:
At this time I was living with people who loved to camp and do outdoorsy type things, and one of the toys they had around the house was a tin can stove made from internet instructions. I looked it over, and decided it would be fun to try my hand at it. I took advantage of unlimited machine shop access, and produced this:

It's turned out of a solid 3" bar of aluminum, and so while it's slower to start than the real tin can stove owing to its higher thermal mass, it can really get going:

I dub this an official failure, because compared with the real tin can stove which costs nothing if you lose it and weighs 10 grams, this would make me cry if I lost it and weighs 80 grams. That weight differential is like asking an ultralight camper to pack a sledgehammer, apparently.

Once the semester started (we're in the fall of 2004 right now) I had a brief rekindling of my interest in old-school video game hardware, thanks to a freelance job I got which required me to design a video decoder circuit. Once the job was done, I took one of my prototype video decoder boards and stuck it in a box along with a reconfigurable PIC-based controller decoder, and got a mini-JAMMA interface.

You can see the breakout cable for Capcom 6 button fighting games, mostly the only arcade games I played. It's got component video outputs, which as far as I know is the first true component decoding board that existed for this purpose (the Jrok board soon followed though). The top will accept neo-geo style 4 button controllers or the Sega Genesis 6 button controller, whose buttom remapping can be changed by pressing the "mode" button. Inside you see the two boards:
And once again disgusting wire layout. You can see the audio is unconnected on the right, because I built it primarily for CPS2/3 boards (which, by the way, were JUST cracked!) that all have the on-board Q-Sound RCA outputs. I have today it's fun to see the "Credits 0/0" on your screen, then press a switch a few times and see the number go up like a slot machine. Like free money, except it's free money so you can spend more time rotting your brain playing video games.

Right about then I discovered the MIT hobby shop, and a world of speaker building possibilities opened up. My HiFi tendencies had up to this point laid dormant, since living in cramped spaces with other people who many not share your taste in music tends to stifle sonic expression. Still, nothing could stop me from building speakers!

These are the satellites I made for one of my cousins' subwoofer/satellite systems. This cousin likes mostly eclectic jazz (I have no idea the formal genre name) and lots of vocals, not necessarily much at high volumes. So his speakers have very good midrange in exchange for bass clarity and output level. You can tell this is my first attempt at a gloss spray painted finish: although this picture is particularly unforgiving it's nothing special at any angle.

I'm a big fan of sub/sat systems. When done right, they give the sound of larger speakers, but with much greater placement flexibility and a much less intrusive visual presence. Better yet, it takes some more good engineering to get them right, which makes them a challenge. Finally, small satellite speakers are just darn cute.

Here are the satellite speakers for my other cousin; this one likes more traditional jazz and the occasional loud martial arts action movie, so it has better lower bass (for reproducing the low bass strings) and greater output level. You can tell I gave up on the spray finish, mostly because my makeshift spraybooth had to be taken down.

They get the same subs, though: 8" ported woofers. These drivers are monstrous yet still sound tight, due to a correspondingly stiff suspension to go with that giant magnet. They were a surplus win from Apex Jr.

Finally, I made myself a triplet. I wanted to go slightly nuts, so I got a Hi-Vi SP10 woofer, which was actually so heavy (30 lbs) it was shipped in its own wooden enclosure! To go with this monster I needed efficient speakers, so I chose MTM satellites. Here are the satellites before painting. What makes me sad is in the background you can see my custom shelf system I constructed for my desk. It had cubbies for my oscope, two different soldering irons, part trays, signal generator and my monitor. I had to leave it in MA when I left because it was too heavy. :(

Here is a speaker after painting, in its natural environment of my old apartment. You can tell I got a bit better at the spray painting.

Here's the sub on the floor, an 11" sealed cube. I'd be lying if I said I wasn't going for the Sunfire look, but honestly I doubt this sub performs a tenth as well. Bob Carver's Sunfire sub is engineering genius.

Now that I don't live in MA anymore, here are my speakers sitting on the floor in my dad's HiFi room. No use for them in Italy! We're having some work done on the house so the HiFi room is in a bit of disarray at the moment.

Here are my dad's reference speakers, the incredible Bella Voce Signatures, a really kind gift from the designer himself. Behind you see the old reference speaker, a Vandersteen Mk II. You also see the drop cloth everywhere.

I had a bunch of other projects, but they were all lost when my hard drive gave up the ghost. Thankfully I learned my lesson, and I'm doing daily incremental backups!

Steps, Waves, and Loops

2007-02-06T07:35:00.000-05:00

Don't worry, part two of the power converter series is coming. I recently had a conversation with a friend about step and frequency response, and I figure I should write it all down before I forget it again. I've taught analog circuits for four semesters, two undergraduate and two graduate levels. The first hurdle seems to be the idea of a small signal model, and linearization in general. The second is poles and zeros, especially zeros. There are lots of published works on how poles and zeros work with respect to complex analysis, but as I learned when I switched from math to engineering, Riemann mapping and Laurent series are as useful to intuition as a tree is to a cabinetmaker.

I'm going to try and explain how I look at poles, zeros, and feedback while invoking as little math as possible. After all, a circuit is a real thing you can hold in your hand and you shouldn't need anything imaginary at all to describe it. Unfortunately, like my old professor used to do to me whenever I thought I had struck upon some circuit epiphany, you can always say "Well what about this case..." and show some pathological circuit designed to break all generalizations. I think the best I can do is look for patterns that work 90% of the time, and be aware of the types of 10% where the pattern does not apply. After all, intuition is just that; no need for precise answers at first glance, that's what a simulator is for. It just helps to know beforehand if your results are supposed to go up or down or in circles.

Poles and Zeros

Suppose we have a real circuit, and we put it in a sealed box. Now physically, the net energy crossing the boundary of the box is zero; this box neither sinks nor generates energy. That's not to say all the electrical power we put in will come out as electricity; some of the power we put in will heat the box, radiate magnetically, and dissipate. But the point is this energy is accounted for when we say "crossing the boundary of the box" and so we are safe. It also doesn't mean that the box doesn't store any energy; during steady-state operation, a constant amount of energy is stored in its inductors and capacitors.

Now let's pick a pair of terminals, label one the input (voltage, current, power) x and the other the output (voltage, current, power) y. So usually when we do two-port theory we have to issue disclaimers right about now such as x is not directly connected to y, or y is not connected directly to a dependent generator and nothing else. But (a) this is a real circuit, so even if x and y were directly connected there would be a tiny bit of parasitic resistance in the wire and (b) we're not going to nitpick. It turns out, again thanks to physics, that the ratio of |y/x| must get smaller as we turn up the frequency of x. If this isn't true, then the world would end. Actually, for fun one day I'd like to come up with a logically continuous set of statements that begins with "If the world didn't have a low-pass roll-off, then..." and end with something absurd like "...I would be able to eat Mars, Venus and Neptune for breakfast and still jump 200 feet in the air." So with this depressing perspective that everything will eventually go wrong, a pole is easy to swallow. It just tells us at what frequency this begins to happen. A zero, however is still puzzling. A zero means the output is fighting the natural roll-off of the universe; energy is coming from somewhere. But our circuit's in a sealed box, and if all the inputs are kept constant, is the box generating energy? Nope. A true (authentic) zero just doesn't exist in real life.

Well what about an inductor you say, it's got an impedance which rises with frequency so if we drive it with a current source then the voltage across it will just keep going up with frequency. Well a real inductor's got a self-resonant frequency above which it looks like a capacitor. Bummer. Of course this example is slightly screwy, because the power of the input source is going up as well so nothing is explicitly being violated. The point is nothing gets infinitely large.

So then where do zeros in circuits come from? Well consider this example.

We've got two simple single pole paths whose outputs add to produce the overall output. This is a completely natural situation that arises all the time; in any circuit the signal has multiple paths to the output. For example an op-amp wired in summing configuration, or if you want to get fancy a gm-C stage with a mis-matched input pair. Let's compute the output value.

Wow! There's a zero! Why did this happen? Well, suppose a₁=a₂=1, and τ₁ >> τ₂. At DC, the output gain is 2, with equal contribution from the two paths. Past the first pole frequency the first stage contribution suffers, and so does the output UNTIL the gain in the first path has dropped below one, (at twice the pole frequency) then the output is entirely from the second path. That is the zero location. Obviously I judiciously chose τ₁, τ₂, a₁ and a₂ to make the example work but you get the point. Here's whats going on pictorially.
So a zero in a circuit is not some magical amount of gain that suddenly makes the output rise, it's just one path of a circuit taking over for another as frequency increases.

Invoking partial fractions for a moment, it turns out that any transfer function for a real-world circuit--the denominator has a higher order than numerator--can be disassembled into the sum of low-pass and band-pass components. Now these components may have no physical meaning with respect to the circuit, but it just shows that all the zeros in the transfer function arise from re-direction of the signal inside. If you have all these components and want to get a rough overall response, you plot each component on the same Bode plot then follow the maximum of all the plots superimposed. This is what one of my teachers called "the Bode obstacle course".

Feedback, Step and Frequency Response

It seems to me that the first feedback gurus were all mechanical engineers and physicists, before the days of the diode, triode, pentode, compactron, MOSFET, JFET, MESFET, MODFET, HEMFET, IGBT, blah. An early crude form of feedback existed for the regulation of steam engine output speed, called a governor.

Then starting in the 20's feedback became cool for electrical engineers, so you had a lot of brilliant math and physics guys come up with all these neat theorems. They used silly terms like "electrical network" and "driving point" and "cps", but that just adds to the authenticity for me. This lasted until the 80's, after which analog fell out of vogue and digital became cool. From what I've learned in classes, it seems nowadays analog design has become a lot of topology and pattern recognition, and most of the innovation is with mixed signals rather than at the transistor level. The mechanical engineering undergraduates I've met know a lot more about feedback than most electrical graduates, so I'm lucky to have had an adviser who is a huge proponent of all things feedback.

The characteristics of step responses from which we can infer frequency response information depend on the general topology of the circuit. For example an RF power amplifier may have a qualitatively similar looking step response to an audio power amplifier, but because the audio amp has a strong global feedback loop around it means we know about its loop gain, something which is meaningless for the RF PA. In particular, without a global feedback loop the relationship between bandwidth and gain is lost, instead trading bandwidth for delay.

I've always thought about distributed gain circuits in terms of pipelining. This is the only time I use a digital analogy to describe an analog phenomenon. With pipelining, you break a complex task into a series of simple tasks, and so even though from input to output there's a longer latency once you get going the throughput is higher. Anyway in analog if you have a lot of low gain high bandwidth amplifiers in series, then the total bandwidth of the system is higher than a single amplifier of the same total gain--if you built your amplifiers properly--but now you have a disproportionately long delay from input to output. That's about all we'll say about non-feedback systems for now.

What's a Step Response?

So suppose there's a circuit, just chillin. All of a sudden we run up and kick it!! Then we watch what it does. Ideally we'd want to measure its impulse response, but there are these problems with generating impulses like infinite-amplitude, zero-width, and finite-energy. The Fourier components of a step are a bunch of cosines, whose amplitudes are the inverse of their frequency. One way to think of a step response is to treat the system like a filter bank. We take the step, split it into all its cosines, send each one through its own filter, then recombine them at the output.
Each component gets scaled by an amplitude and shifted by a phase. This is nothing new, it's a somewhat digital way to think of things, but it's how I do it.

The Timewidth-Bandwidth Property

If you've never heard of this, don't worry as of three days ago neither had I. I've always been wary of applying a fancy name to a simple concept since it makes me feel like I'm reading the math section of a psychology book, and I have enough fancy names to remember already. Try CMOS065NLVTLP, for "low power 65nm N-type metal oxide field effect transistor with low threshold voltage" WTF! Besides, "timewidth" isn't even a word! The point is simple:
Frequency and wavelength are inverses. If we look at very narrow slices of time, we can only see high frequency behavior. To see low frequency behavior, we need to look at longer slices of time. So the high frequency response controls the initial step response behavior, and the low frequency response controls the steady state behavior. Mathematically these are called the initial and final value theorems. What's interesting about feedback is just that; at any point in the circuit the signal will travel around a loop and a portion will return to the same point.

But how long does it take for the signal to come full circle? Well, if you look at the phase delay of the system, we can get an idea of the time delay in the feedback loop.

Here we see that in-band, the system has a constant delay equal to one period of the pole frequency. If the pole is at 1kHz, then the delay is 1ms, and so on. We know that past the unity gain frequency the feedback is negligible, so the minimum delay of the feedback loop is 1/f_u. That means that between t=0, and t=1/f_u, there is no feedback going on and the system is behaving open-loop. After this time, the feedback gets stronger as the loop closes. As the loop closes and the loop gain increases, the output gets closer to its final value.

At least that's how I see it.

These concepts are encapsulated nicely in Blackman's theorem. With this theorem, you decompose a circuit into two transfer functions, one which is valid when the feedback is completely operational (low frequency) and one valid when the loop is completely open. (high frequency) Then you determine a third transfer function which "interpolates" between the two transfer functions.

Gain-Bandwidth Product

This is one of those things that seemed reasonable when I learned it, having a pair of parents who throughout my life beat me with the wisdom that "you can't have your cake and eat it too" but I nonetheless didn't quite grasp until later. I mean it's weird right, you can take a huge electric motor that takes full seconds to reach top speed, wrap an opamp and power transistor around it, and all of a sudden it's doing backflips.

Let's forget about gain, bandwidth, and frequency response for a second to ask, what does it mean to say the speed of a circuit? Well, it's how fast the output changes, dv/dt. We want to know what's the maximum derivative of the output of the "plant" we're trying to control. Suppose we have a plant with a gain of A₀ and bandwidth of ω₀. In this case the maximum derivative is

Now let's consider what happens when we wrap a loop around this plant. Let's suppose we want the loop to have a gain of A, and we want to determine the new bandwidth. Well it's still the same plant, so we should expect the same maximum derivative.
Hence, the gain-bandwidth product. The interesting corollary of this is everything being kept equal, if we turn the amplitude of the input to a feedback system up, it's actually more taxing on the plant. That's obvious when you think of the plant being a motor or something mechanical, but perhaps less obvious when the plant is an op-amp. This is also why back in the day when we had ±15V supplies, the unity gain bandwidth of the amplifier also came with a peak-peak sinewave spec; often it was something much lower than rail-rail.

Second Order Systems and Ringing

Second order systems are ubiquitous for two reasons. One, they're very easy to solve. Two, turns out lots of phenomena are either exactly second order systems or closely modeled by them. I remember being asked why in class, and my wrong mathematical response was "Well sin''x=-sinx..." And everyone else in the class was like, no because F=ma and F=-kx'', dummy.

Anyway in this same class I saw a really neat proof of some second order system properties. Before this I never remembered why magnitude peaking occurs at 45°, or why the resonance peak is not exactly at the natural harmonic frequency. There are crazy formulas for them involving ζ and ω_n but I could never remember those either. I don't know who came up with this proof, but it wasn't me. The best I can do is remember it three years later.

Let's begin by normalizing our system by ω_n, which means |p|=1. Then we can write our system as:
Which is graphically represented by this:

The value of our transfer function is one over the product of the two solid line vectors.

The questions we want to answer are: (1) for what values of p do we see peaking in the frequency response, (2) at what frequency does the maximum occur, and (3) what is the magnitude of the peak?

To solve (1), we have to see if |(s-p)(s-p*)| has a local minimum less than one. Referring to the dotted line in the unit circle above, we see that |s-p*|=|s-(-p)|=|s+p| which means |(s-p)(s-p*)|=|(s-p)(s+p)|=s²-p². But we know s=jω, so s²-p²=-p²-ω². How does this help us? Well, -ω² lies on the negative real axis, and p² is somewhere in complex space. Note that Arg{p²}=2Arg{p}, which means that if Arg{p}>135°, then p² lies in the right half plane, and therefore does not have a local minimum distance to any point on the negative real axis. This is illustrated below.

Arg{p}>135° corresponds exactly to the case where p is more complex than real, i.e. the angle it makes with the negative real axis is greater than 45°. Also note that since the radius of the circle is one, any local minima which occur have value less than one. Hence, there is magnitude peaking.

Once we have this, parts (2) and (3) are easy. From the figure above, the local minimum clearly occurs at the projection of p² onto the negative real axis, and the magnitude at the minimum is the distance from the real axis to p². These are respectively

If you've been following along, you'll notice that I dropped a minus sign. The thing is we've been defining θ as mathematicians do, but in systems they define θ from the negative x axis, hence the missing sign. It's easy to show that these definitions match with the classic control theory definitions, as follows:

Certainly we can all agree the previous expressions are easier and more intuitive than these, but then again some of us are arguing that π should have been defined as 2π so let's not get pedantic. Just to drive the point home on how neat this proof is, I quote the corresponding proof from an LTI System Review paper: "The peak gain will occur when <second order rational function> is a minimum. After simple differentiation, we find the minimum ω_m is given by <formula given>. The corresponding peak value M_p is given by <formula given>." In engineering/math parlance, calling a task "simple" or "trivial" is synonymous with "this sucks" and "I don't want to do it".

Anyway the rest of that system review covers second order systems nicely, so we'll leave that out. What's interesting is when a system rings but doesn't look second order. What about a transient response like this:

It looks like a perfectly well behaved step response except it's got some annoying high frequency junk on it. What can happen is an emitter follower is driving a capacitive load, and since the output of an emitter follower appears inductive (a one transistor gyrator!) it's possible that we get an unwanted resonant tank. Thing is these parasitics are often much to small to appear as dominant effects in hand analysis, so the system looked well behaved on paper. This is the frequency response of the loop which resulted in the above transient response.

As you can see the system's got a 90° phase margin but the resonant tank out of band pushes the gain back up at a high frequency, which is what's riding on our first order response.

Zeros and Over/Undershoot

What sort of circuit would produce step responses like these?

This is actually a direct result from the discussion of multiple paths above. Suppose we have two paths to the output, one fast, one slow. The fast path dominates the initial response, and the slow path dominates the final response. If the two paths have different polarities, then the initial response can have different polarity than the final response. We call that undershoot. Undershoot means there's an odd number of right-half-plane zeros. Again, this comes from the Initial and Final Value Theorems.

With overshoot, it's tempting in the simulator to try and "tune" component values so that the output shoots up and lands squarely at the final value, meaning a super fast rise-time and no ringing. This is a bad idea because in practice values will be different and you'll be stuck with the settling time of the slow path, which probably sucked to begin with. It's better to find the source of the overshoot and eliminate it, which will probably speed up the dominant path in the process.

In practice, unless you know for sure your circuit should behave like this (power supply zeros, for example) the initial behavior is mostly likely caused by some parasitic capacitive path to the output. This looks low impedance at high frequencies hence the initial response but block DC, so the expected behavior takes over after a short amount of time. If you're experiencing undershoot, look for capacitive paths around inverting stages.

Doublets

When I first learned about doublets I tried to find a simple answer for why they cause slow settling transient responses. One way to do it is to brute force calculate the inverse transform of a transfer function with a doublet and partial fraction decompose the answer to isolate the exponential with a slow time constant. This is incredibly unsatisfying. For now I've come to the conclusion that it's all about delay. The key point is that phase starts to fall before gain does. So even if the pole and zero are fairly close together, a factor of two apart, you still get a phase bump of 16°. What does a 16° phase bump mean? Well, it's assumed that the doublet is occurring at pretty low frequencies (we'll deal with compensation later) either inadvertently or for canceling an unavoidable low frequency pole. A phase bump at low frequencies can cause a significant delay to low frequencies, which is the source of the long settling time; for all the frequencies below the doublet frequency, the phase delay of the system behaves as one with a bandwidth of the doublet frequency!

Here is a Bode plot of three systems, the blue curve is a system with a gain of 100 and a bandwidth of 1MHz, hence it crosses over at 100MHz. The green curve is the same system but with a pole at 1KHz and a zero at 2KHz; this also has a pole at 1MHz, but crosses over at 50MHz due to the doublet. To show that crossover is not the culprit, the red curve is a system with a gain of 100 but a pole at 500KHz, so it crosses over at 50MHz same as the system with a doublet.

The resulting phase delay from these three systems tells the story. Note that I had to use a logarithmic y-axis to get the three curves to fit nicely.

The original system has a 1us delay; the slower system has 2us. The doublet, however, has a full 0.5ms! Thinking again about how feedback is a loop, it means all the frequencies below 2KHz are running open loop until 0.5ms later, which is the source of the slow settling time.

That's about as long as I want this to get, it's already pretty incoherent and adding more would just make it harder to read. Please if you spot any errors let me know. (not grammatical errors, I make those on purpose)

What Really Matters: Power

2007-02-01T02:00:00.000-05:00

This blog was supposed to be about engineering, but there hasn't been a shred of anything remotely scientific here for months. So with this realization-- suspiciously coincidental with the turning of the new year, but let's not make this a Hallmark moment--I have decided to stop talking about my feelings and buckle down to business.

Living in another country and having to adjust to the wall supply of nearly twice the voltage and 10 cycles per second less, I've been thinking about compact efficient power converters. This may be foreign to the younger crowd who's grown up on the 100-240V universal power supplies that power nearly every electronic device these days (notably excepted are those by Nintendo and Microsoft, who were willing to accept the monetary losses of region-specific hardware in order to make our lives more difficult) but if you ask a parent who's used an electric razor overseas--try your dad first--he will probably tell you he bought an over-sized plug which served the dual purpose of physically changing the interface as well as make the razor run instead of smoke.

Phase Control Converters

What's going on inside one of these doohickeys is probably a simple TRIAC circuit. The circuit is set so that it will periodically (mathematically speaking) chop off a portion of the sinusoidal 220V RMS input signal such that the resulting waveform will have a 120V RMS voltage.

In the figure above, the RMS power of the red curve is identical to that of the pure blue sine wave, even though its maximum amplitude is 85% of the red curve. The angle of the delay when the red curve turns on is called the conduction angle α, and this type of power controller falls into the aptly-named class of phase control circuits. The waveform is not a sinusoid anymore, we say it exhibits harmonic distortion, which means the spectrum of this wave has components at frequencies other than 50 or 60Hz.
I've plotted the first nine harmonics of the red waveform above referenced to the power of the first harmonic. The sum of the powers of all the harmonics higher than the first as compared to the power of the first is called total harmonic distortion, or THD.

When the mains line has a lot of harmonic distortion, we colloquially say the power is dirty. Not only is the harmonic distortion a problem, the stock version of this circuit which constitutes all the cheapo power converters out there also has a problem driving certain types of loads which we will discuss in a second.

RFI

We've got so many electronic gizmos these days we have to make sure they don't interfere with each other; specifically, the device must avoid emitting Radio Frequency Interference. I'm not sure if there's a concrete difference between RFI and Electro Magnetic Interference, they seem to be used interchangeably. The FCC--the happy branch of government that's currently slicing and dicing our mm-wave band so we'll be able to stream HD video to our TVs wirelessly--has strict requirements on how much junk a device can radiate into the air. Turn anything electronic over (I just turned my iPaq over and it says FCC ID:NM8BG) and you'll see some sort of obligatory certification from the FCC. This is also why we're not going to see an iPhone for like, forever. That thing is stuffed with so many chips and antennas, the events in Contact will be precipitated by someone posting some video she just edited on her iPhone to YouTube while her iTunes song paused so she could accept an incoming call with her Bluetooth headset, all the while sending an email. I'm pretty sure Jules Verne didn't call that one, at least not to the extent to which we're taking it.

So the problem with chopped sine waves is the sharp transitions mean substantial high frequency content. This high frequency content is on wires and devices that were not originally designed to carry it, thus are not shielded and so the junk radiates off into space, hungry to make your garage door open. (or stranger things, Google it)

Not only is harmonic distortion bad for the environment, it's not good for the device it's powering. For non-electronic appliances, the surges in current which follow the steps in voltage put stress on the wiring and components. This causes losses due to wire heating and the device itself may not last as long. For devices with transformers, not only does the circuit require slightly more advanced phase control (see below) the harmonics do little other than heat the windings and possibly saturate the core, which leads to greatly reduced efficiency.

Electronic devices are also sensitive to dirty power. Switching power supplies often have compensation called Power Factor Correction circuitry at the mains input, which try to make the supply look as much like a resistor as possible. This circuitry, which is FCC-mandatory for supplies over 75W, tries to compensate for the fact that the bridge rectifier in a Buck converter draws current in pulses rather than continuously. A passive PFC circuit is similar to an RF impedance matching circuit. In these passive filters there are capacitors, which have ripple or pulse ratings that are different than DC maximum voltage ratings in that they specify how well the capacitor tolerates transients. When you supply dirty power you are directly challenging this rating, and the caps might just pop.

An active PFC is actually a Boost converter placed right after the bridge rectifier, which dynamically adjusts its output voltage based on the input voltage so that the mains current will be drawn proportionally. This is also why we have universal power supplies: with a single Buck converter we are unable to accommodate a 100-220 input range, but if we use a Boost-Buck topology, we have enough added flexibility with the intermediate voltage that the input range is widened. In any case the bridge rectifier doesn't care about whatever kind of voltage waveform we give it, but the Boost converter does, and at the very least it will run less efficiently.

Inductive Loads

Another problem with phase control supplies is the load has to be well-behaved. Ever notice how the converters explicitly say "non-inductive loads only"? The OK list includes electric razors, water heaters, light fixtures, hair dryers, and coffee makers whereas it's not OK to attach an electric fan, blender, or basically anything with a big motor in it. This is due to the reluctive (do we say that? As in, a reluctive impedance has a positive real part as opposed to a reactive impedance, which has a negative real part? Let me know) impedance of an inductor. Turns out when you apply a sinusoidal voltage to an inductor, current will also be sinusoidal but delayed, the exact time of the delay depending on the inductance and frequency.
The TRIAC is a voltage triggered switch, because the purpose of the circuit is to generate an RMS voltage, not current. On the other hand the TRIAC turns off when the conducting current drops below a certain value. So we have two statements:

The "ON" period of the TRIAC starts when the mains voltage reaches a trigger value, and ends when the mains current drops below a certain value.
With inductive loads, there is a variable amount of phase shift between voltage and current.

This results in some whacky behavior when the TRIAC circuit wasn't designed to account for this variable phase shift.

As much as I don't like these TRIAC and SCR circuits from an aesthetic perspective, they aren't going anywhere. Turns out we can make SCRs capable of handling heinous amounts of power, many many many many kW. For industrial heating applications like furnaces and welders, it's the most efficient way to regulate their power. Moreover even for big motors, there are ways to make sure the switches operate properly with a bit of control circuitry.

The Problem

As it just so happened, I got an American Xbox 360 while I was in Europe, and the jerks at Microsoft make the power brick 100-120V only. (Japan and US/Canada) It continuously draws 180W, which is more than my Mac Mini AND laptop AND monitor draw put together. Given that thing is big enough to backstop as a tire block you'd think they'd make it multinational. Jerks.

A phase control converter just isn't going to cut it. First, the radiated EMI is going to be huge. Just like light dimmers hum when the lamp is not set to full brightness, every piece of audio equipment nearby--and possibly the power cord connecting the Xbox--will hum loudly. Furthermore, the Xbox power supply is notoriously sensitive to dirty power, and when it's not happy the brick shuts down and you'll lose that 3.2GHz triple-core of awesome computing power.

The simple recourse is to buy a step-down transformer. This is the no-brainer brute force solution in every possible sense; you take two inductors, couple them magnetically, and thanks to a lot of theory a sine wave should pass losslessly from one to the other. It's also monstrous, heavy, and in fact does waste a bit of power. As an added bonus now you've got very clean but nonetheless strong 60Hz EMI radiating from that little beast and probably some mechanical noise from the windings vibrating, unless you pay big bucks for a nicely shielded and mechanically dampened transformer. As a side note, feeding a transformer dirty power is a great way to make it sing in all sorts of funny ways.

The Solution

Here comes the awesomeness of electronics, it's very exciting. Let's start with a basic Buck converter. It takes a DC input voltage, and produces a DC output voltage of lower potential. What if instead of fixing the output voltage, we tell it to produce 220/120=0.545 times the input voltage? Then if we get a 220V RMS sine wave in, we'll get a 120V RMS sine wave out. This isn't the optimal solution because the output integrity of the sine wave is entirely dependent on the input, but knowing that this will only be plugged into AC mains we can overlook this. If we were cooler, we would take the input voltage, rectify it to some intermediate filtered DC voltage, then reconstruct a 120V RMS pure sine wave at the output. But we'll leave that to the UPS guys. No, not that UPS, this UPS.

There are a few implementation hurdles to overcome. First, we'll need to make sure the feedback control of the converter works properly with a 60Hz AC reference, which actually shouldn't be a problem since DC converters normally have a high enough bandwidth to reject AC line ripple. The other is to make a bipolar switch to accommodate an AC input. A generic buck converter looks like this:

The problem with this converter is it can only convert higher potentials to lower ones, i.e. it has unidirectional power flow. During half the sine wave that's not true. Either we will have to do some fancy switching at the input of the converter, or we will need to make a bidirectional buck converter. Here is a slight modification to the topology that allows this:
Our input to the switches is full-wave rectified AC, broken into two alternating phases.
Then, we drive the switches like this:

Phase1 ϕ1:d ϕ2:0 p:1
Phase2 ϕ1:0 ϕ2:d p:0

In phase one, the upper left MOSFET and lower left diode form the Buck converter, and the lower right MOSFET is the return path. In the second phase, the upper right MOSFET and lower right diode form the Buck converter, and the lower left MOSFET is the return path. By switching between phase one and two, the load sees an alternating polarity.

Even this simple circuit isn't a piece of cake. Because this device will be working off 220V mains, the sine wave has a peak voltage of 310V. For safety margins, the absolute minimum voltage rating of all the devices should be 500V but no commercial product would use less than 600V. Turns out 600V is high as mainstream MOSFETs go, and they get expensive. Although working at a cocaine plant might be a different story, when you work for a large electronics company you get a few perks, one being free samples. So getting the switches for this circuit shouldn't be a problem. Now I just have to design the drive electronics.

To be continued... feedback controllers, snubbers, resonant conversion and novel power supplies.

As an aside if I ever got into a position of extreme power at the IEEE, for fun one year I would mandate that all papers submitted for the ISSCC (International Solid State Circuits Conference) may not contain any instances of the following words/phrases:

new
novel
adaptive
high-performance
mitigate
investigate
any restatement of Moore's Law
ultra (unless the word immediately following is "sucky", because let's face it sometimes breaking the monotony is fun)

This is nowhere complete, but with just these changes I would watch a normally 800+ page publication shrink to the size of a Newsweek. Just to clarify: if you're shooting to get your work published in the world's premiere circuits journal, there is absolutely no reason to include the phrase "high-performance" in the TITLE! Do you think they're going to let you publish An All-Digital PLL with Narrow Lock-Range, Above-Average Performance and Middling Power Consumption?

A Trip Through Europe

2007-01-17T14:59:00.000-05:00

Since I didn't bring my computer with me, I did my best to capture notable experiences on scraps of paper which all ended up at the bottom of a very dirty duffel bag.

From Milan to Geneva by Train
As European train stations go, Milan Central is a decrepit shack. Its only saving feature is right now there is a five story high poster of Gisele Bündchen for Dolce & Gabbana The One fragrance. There's no better way to pass time between trains than find the least uncomfortable stone bench and gawk. Simon and I arrived at the station an hour and a half before our departure as we had accompanied a friend to the airport who had an early flight. By the time we boarded our 8AM Eurostar to Geneva, I had a neck cramp. Recalling the four torturous hours spent with the world's most dysfunctional couple my last trip to Geneva, I was relieved to find us in a cabin with a kindly older Swiss man from Lausanne and a girl our age. I stuck up a conversation with the girl, Murielle, who it turned out works in fashion designing the advertising campaign for purses. Originally from Paris, she now works in Rome and was going to Sion, Switzerland to pass Christmas with her family. At some point Simon fell asleep leaving me alone to muse the possibility of stripes re-appearing in 2008 French fashion. At least I now know a lot of the back story behind The Devil Wears Prada. The finer workings of fashion to me are like tort law: yes it's very important, subtle, complex, and someone's gotta do it, but I don't care.

An hour after our arrival, Catherine and family left to spend Christmas Eve with Jean-Blaise's father's family leaving Simon and me alone in a huge house. Before they left, Jean-Blaise gave us two bottles of wine which he absolutely insisted we drink, as well as a bottle of champagne. Since we're not alcoholics, we only drank one bottle of wine and didn't touch the champagne. That day began a two week straight state of inebriation. We passed the night by baking and drawing pictures for the kids.
Although the kids love my banana bread I was beginning to feel like a one-hit wonder, especially since banana bread is so easy to make. In an effort to prove there is depth to my culinary repertoire, I baked an orange-lemon chiffon cake with bitter chocolate frosting. There was no cake flour so the chiffon wasn't fluffy enough, but Catherine saved it by soaking the whole thing in rum. The youngest daughter Clémence fancied an alligator in a castle, so Simon obliged in water color. The pen drawing was just Simon doodling trying to figure out what alligators looked like, and if you observe closely you can see the telltale signs of Simon's twisted mind.

The next three days went by as one huge marathon meal. I've never had so much good wine. Our glasses were perpetually full, as Jean-Blaise has this knack of silently sidling up and refilling. One day I'm going to learn how to entertain like that. At this house, simply being awake one ran the risk of being offered piles and piles of impossible to decline delicious food. We ate foie gras three times in two days, which was homemade by Catherine's sister Eliane. Simon was almost shot at the table when he asked if foie gras is the same as pâte.
Here is Catherine, Jean-Blaise and Emilia, the little girls' au pair from El Salvador. This was right before the presents were opened on Christmas night.

The girls got a foosball table for Christmas. Unbeknownst to us at the time, this would serve as a prelude to our official introduction to foosball in Prague.
Catherine's mother is center, with Emilia and Eliane.
Two of the three girls, Oregane and Marine. Oregane will definitely grow up to be an actor or some sort of performer. She enjoyed clasping her arms together over her head and being carried like a basket. She's 8 and she can also do a really clean aerial. Talented little kids make me jealous.

After all this binge eating and drinking, we really couldn't bear another rich meal so Emilia offered to take us out one evening and show us around Geneva. Oddly enough, we ended up eating and drinking.

After dinner we went to a German bar that advertised a wide variety of beers on tap. Simon asked a nearby older man who was sitting by himself reading a paper if he could snap a photo. The guy obliges, then spends a minute trying to figure out how to use my camera, which is the digital point-and-shoot type. First he asks how to turn it on, then holding it backward he stares into the lens while fumbling to find the button. The three of us are treated to a big confused looking eye on the camera's screen facing us. Anyway it was funny to meet someone even more technically incompetent that either of my parents.

Geneva to Prague by EasyJet
EasyJet is the European equivalent to Southwest or Jetblue, a classless cattle-herding style of airline. During take off I was flipping through the in-flight magazine when lo and behold, there is the same perfume ad with Gisele Bündchen. Simon, being an artist, offers to draw her for me so I don't have to dismantle the magazine. As I wake up during landing, I've got a perfect picture of Gisele Bündchen. Awesome.

Prague at night is a resplendent city. Problem is, it's skin deep. It is an old city with rich history and architecture, and would be perfectly presentable in its age and understandable dilapidation. Yet all the old buildings we came across were being remodeled with some weird hybrid of new materials and old archtecture, sort of like a mother deciding to wear the clothes of her teenage daughter. Eeew. By the way if that really bothers you never visit LA.
This is a clock tower in the center of the old town. One good thing Prague had was sausage vendors every street corner. In an attempt to become more disgusting than I ever thought before possible, I ate three giant sausages, two fried ring pastries, a pile of sliced roast pork, and washed it down with extra-strong Budweiser. Budweiser is after all Czech.

This new gluttony record would stand for about a week. Best of all, all this cost about $6 thanks to the money conversion. But let's be honest, how could I not eat everything in sight, especially when I'm given so many choices?
Finally, here is the sparkling borderline gaudy Christmas tree in the middle of the town square.
At night we went to bars that were so dense with smoke I couldn't see from corner to corner. Everyone always says the smoke in Europe and China is really suffocating, but I never minded it. I even feel a bit cheated if a bar doesn't have that stale smoke flavor. In Prague though I just wished there was some sort of air circulation. The other funny thing about the bars is they all had nice foosball tables. Amazingly, it wasn't just fat old men playing but young attractive people as well! Simon and I figure, well when in Rome... (Simon abused that pun liberally, usually to justify doing something idiotic) Problem is, we got killed by the opposing team and no one came over to talk. Apparently the bottom line is not what type of competition it is, it's whether or not you win. After that we lost interest in foosball.

All in all it was worth it to go to Prague and see the sights. Aside from stuffing our faces we visited the national museum of anthropology, and we also saw an art exhibit with paintings by Alfons Mucha and the photography of Jan Saudek; both socially redeeming activities, even if they were just to pass the time between feedings.

Prague to Wroclaw by Train
The whole time we had been traveling, the most common complaint was too many people. People taking up legroom, bags everywhere, people with small bladders who insisted on sitting at the window, people who shouldn't be married to each other, little people who cry, all these factors raise blood pressure. This time though, we yearned for traveling companions. All alone in a rickety train barreling into a barren snowy land as the sun goes down, it was unnerving that the whole car was practically empty. For a distraction we decided to play chess. Unfortunately, between the time we weren't playing chess and the moment we decided we should, a chess set didn't magically appear. Fortunately Simon put his abundant artistic skills to use again.

Poland was a load of fun. Two general observations which really only apply to the city of Wroclaw, around New Years: (1) Polish people love foreigners, especially Americans, and (2) most people speak 3-4 languages.

We stayed with a friend of Simon's from acting school, so we had a guide for the area. I'm pretty sure we started partying within two hours of arriving, and didn't let up much until we left. The first night was Saturday, December 30th so everywhere we went was packed. The smoke was apparent but perfectly bearable, and beverages were cheap by American standards. People were approaching us whether we like it or not. It was the sort of situation I always envision when I hear the term alcoholics anonymous; lots of random strangers getting drunk and hugging each other. That's about all I remember of that night. The next day we helped our host prepare for his combined birthday/New Year's party, which started at 6PM in the evening, and ended at 6AM in the morning. Telling some Polish people about it later, they scoffed at my amazement over a party that lasted 12 hours. Perhaps "endurance partying" is one of those talents whose pursuit sacrifices much and yields nothing meaningful, but it sure sounds fun once in a blue moon.
This is the typical scene at the party. The guy at front is Giuseppe, a mutual friend through our host, who comes from Torino. I have no idea why that girl is standing on a chair. The benefit of starting a party at 6PM is if you do decide to punch in early, there's a good chance you've already been at it 5 hours. When midnight finally rolled around after an eternity, we went outside and watched the fireworks whilst avoiding the projectile champagne bottles.

Although it seemed the best use of the daylight hours was to sleep off the previous night, against better judgment we visited some sights in Wroclaw. There was a massive painting, the Panorama Raclawicka which depicted a scene in the Polish struggle for independence from Russians. It was duly impressive, but more artistically pleasing (to me) was a World War II memorial nearby.
Of course it's not a happy sculpture, it's two people facing the angel of death, not shown).All this not sleeping during the day required us to recharge, so we gorged on rich Polish food. It was not as cheap as Prague, but I bet fewer rat products are put into the sausages.

The last night we were there, we carried on the by then tradition of drinking excessively. There should be a warning on beer bottles, "May cause you to choose dance partners indiscriminately."

Then the next day right before we got on the plane to England, a few new friends insisted on drinking Tequila. The dozen remaining brain cells (between the two of us) wisely declined, but we met up for a group photo anyway.

Wroclaw to Liverpool by RyanAir
If EasyJet is the JetBlue of Europe, RyanAir has no American equivalent. It's not to say that RyanAir doesn't get you from point A to point B, or that points A and B differ from expectation. It just happens at some arbitrary hour after the time posted, with the most discord and chaos as possible. If SouthWest boarding is like herding cattle, boarding RyanAir is how trying to get on Noah's Arc must have felt. I also for the first time experienced the situation where two native speakers of the same language have no idea what the other person is saying. RyanAir is based in Ireland, and as such has cheery Irish flight attendants all decorated in visually assaulting colors. They spoke English, although I have no way of verifying this.

In any case I wasn't distracted too much by the flight attendants or the turbulence or the cabin lights brighter than a stadium's. It seemed as if the two weeks of drinking every day was finally catching up, as I had a dreadful stomach ache. I barely even noticed the landing, which was rough enough that the whole cabin applauded just because we were amazed we still could.

When we finally got into Liverpool at 12:30AM, we were both groggy and sick of carrying all our bags. At customs we were greeted by a surly surly agent. First she thrust customs forms into our hands because our flight attendants didn't do their jobs. We had to fill in our travel plans and personal passport data. Then she grilled me for several minutes, asking me each question multiple times. She couldn't even pronounce the city Wroclaw, (pronounced vroteswaf) a sin of which I was also guilty before visiting, but she's a custom's agent for the love of God! If it weren't for the fact that getting fresh with a customs agent in a foreign country is probably up there with vacationing in Bagdad as one of the worst ideas ever, I would have been more visibly annoyed.

After I got done saying the words "American", "vacation" and "Pavia" ten times over, it was Simon's turn. I was walking away adjusting my coat when I heard the agent practically scream at Simon in a trenchant English accent, "WHITE is NOT a nationality!!! You're AMERICAN!!!"

I didn't stop laughing for 30 minutes, the time it took for Simon's handbag to be separated from Noah's other animals in the underbelly of the plane and thrown onto the luggage conveyor.

There's a reason why there are no pictures of Liverpool. Simply put, Liverpool is a dump. The weather is miserable. Our hostel room felt like one big soggy sock, and my bed smelled like rotting carpet. It didn't take us long to figure out that the only food available at a decent price was fried. For breakfast we had fries with salt and vinegar. For lunch, we had fried chicken sandwiches. Around 4 we got hungry again so we went to an authentic looking pub and tried the hamburgers. It would just be insult on injury if I happen to get mad cow, because the patty tasted like dishwater and had the consistency of an airplane meal chicken breast; although it was visibly made from ground meat, it had a completely uniform texture. Maybe that's why it's called Spongiform disease, your brain turns into an English beef patty. Sick of all the disgusting food, we went to see Casino Royale. What would have been funnier is if the movie were dubbed in English, but with über strong accents. The coolest part of the movie was when I recognized the interior of the Prague national museum, which was supposed to be a hotel in the movie. Just kidding, the coolest part of the movie is when the stuff is blowing up and James Bond is beating people and Chris Cornell is singing "You Know My Name" and the scantily clad women just are.

Liverpool to Shannon to Paris by RyanAir
Turns out it was cheaper to pass through Shannon, Ireland than fly directly to Paris. Both pilots showed no better grasp of landing than the first one from Poland. As we landed in Shannon, the left wing dipped so close to the ground there were audible gasps from the cabin. Then our flight to Paris was one hour late, and as we were taxiing in for "landing" a little Irish boy started repeatedly screaming at the top of his lungs, "We're all going to dieeeee!!" which elicited nervous laughter from some people. I maintain that if he didn't have such an absurdly Irish accent more people would have taken him seriously. I just wished his parents would knock out his teeth.

There's nothing exciting to say about Paris since we were determined to make it the relaxing part of our trip. If anything we should have tried to see if some bleeding-edge French chef would buy our fat and alcohol soaked livers, but instead we bought a four day museum pass and one night had a really great meal at Le Precope, an ancient restaurant in the 6th. Simon tried snails, and we both tried Absinthe.

The decision to try absinthe was not arbitrary. We had been eager to do so in Prague because at Catherine's house she had a large book full of art and history of absinthe. In the end we couldn't stand the whole cliché that is Prague and held out until Paris. It was fun to see them bring out this giant apparatus (amazing the ingenuity that goes into getting trashed) but the liquid itself is sufficiently vile that we have no desire to do it again.

Simon took me on guided tours of museums. It was exciting to go to a museum with someone who although had never set foot inside already knew exactly which paintings he wanted to see. Surprisingly, I got less tired discussing the art all day than times before when I'd just go myself. Simon made a good observation though, which was that no matter how knowledgeable you are or how poignant your commentary, if you're in a museum critiquing a painting you unavoidably look like a douchebag to everyone within earshot. "Yes, I especially appreciate the artist's use of shadow, it really lends a sense of burden to the mournful figures in the--" See what we mean?

Other times, the art pretty much spoke for itself:

We were in Paris a week, and by the end we both wanted to avoid speaking French, especially to Parisians. Since American movies in Paris are usually subtitled rather than dubbed, we went to see Scoop, the new Woody Allen movie. We misread the start time, so we spent 40 minutes at a café talking to a lady from Iran.

Well, one exciting thing happened. We made some friends and wound up at a party back at their apartment. At one point, Simon came to me and asked where the toilet was. I told him it was just back around the corner. He said, "I know the bathroom's back there, but all I see is the bidet. I don't see a toilet!" I insisted it's right next to it, so with some trepidation he walked back. I sat there a minute when I realized that he wasn't used to old European buildings which have the toilet proper in a separate closet from the sink, bidet and shower. Five minutes later Simon reappears and says, "I swear I just peed in a bidet."

Paris to Milan by Train
Everyone likes to joke, but it's sad to see it's true: a train can depart from some country that's not Italy, and remain on-time as long as it doesn't cross into Italy. For once that happens, all bets are off. We arrived an hour after we were supposed to, which made us stress out because we were supposed to meet Simon's brother at the train station upon our arrival. Fortunately, there was a total mix-up with the flight times and he really didn't arrive until the next morning.

Milan to Rome by Train
We arrived at our hostel late Thursday night. Because were were wiped out and I had just spent a week speaking French, I was fumbling around with the Italian, saying all sorts of stupid things. The surprise was that our hostel was run by Chinese people, who insisted on speaking Chinese. Good for me, less good for Simon. But he was a good sport and did his best. Although our hostel hosts were nice, one of our suite mates had the most foot smelling feet imaginable. It smelled like a whole family of vermin crawled into his shoes and died. At least I hope it was his feet which smelled of feet. He also wasn't friendly, which justifies this complaint.

One thing that shocked all of us about Rome was we probably heard more English spoken on the streets than Italian. At one bar almost everyone including the bartenders was a native English speaker, and this was not advertised on the windows. Of course this did not mean there were any fewer communication problems. (c.f. RyanAir flight attendants)

Rome to Florence by Train
Florence was the last stop on our trip, since it was on the way back north from Rome to Milan. Thanks to the recommendation of our friend in Paris, and our collective weariness toward staying in yet another hostel, (which may place you but feet from someone who has the most odoriferous feet I has ever been my displeasure to endure) we splurged and stayed in a very nice hotel built in an old castle-building in the center of town. Not having to sleep with socks stuffed in your nose really makes for a better disposition the next day.

The first day there we had a nice lunch at a café, the sort of which has hand-written menus and non-matching plates. While we were waiting for our food, the three of us almost gagged (well, Simon did gag) because a smell worse than death wafted by. We looked around and the first thing we saw was a father nervously rocking his baby in a stroller, which made the blame naturally fall on him. We agreed that if any of us were to have children who could do that we would disown them immediately. This event made a funny joke for a few hours until we smelt it again, and this time we realized the culprit was the sewer ventilation grates. As it turns out, this smell pervaded Florence when we were there. Pretty city, but man it stank.

So that's pretty much it, 26 days of traveling and goofing off.

Christmas Time in Pavia

2007-01-17T08:28:00.000-05:00

At the beginning of December I turned 23, which was less stressful than in years past because no one knew and I could conveniently ignore the fact that I am officially not college age anymore. Jon Goler stopped by on his way to Israel, so we had a nice dinner in Pavia.

Right after Jon left, my mother came to visit me and some friends in Switzerland for a few days. We found a restaurant hidden off to the side of the church in Pavia, run by a husband and wife. The wife cooked and the husband was the server. The food was so good my mother insisted on taking a photo with them.

By the way, in December and through the new year many cities in Europe go nuts with the Christmas lights.

Right before my mother and I left to spend the weekend in Geneva, Enrico and Silvia invited us over for dinner. They heard through the grapevine that I liked lardo--i.e. I wouldn't shut up about it for weeks--so here they are presenting the first course of lardo.

Next week the University term ended. That evening, the main historic dorm threw its annual giant Christmas party. Since Jill was visiting, I took her.

A day or two later, Simon arrived. Before embarking on our grand tour of Europe, we went to Milan to do some clothes shopping. Here Jill is making herself properly useful.

Extreme English Part II

2006-12-22T06:17:00.000-05:00

A real conversation that took place between a coworker and me yesterday:

Me: Alright, have a good dinner see you tomorrow.
(we do one of those fancy prolonged handshakes I just taught him)
Coworker: Who's your daddy.
Me: No no no, you only say that when you're right about something or you want to exalt joy.
Him: Word.
Me: When you say goodbye, you say things like "Peace."
Him: But I'm not a Catholic!

Hopefully he will study harder over break.

Will it Blend?

2006-12-15T17:54:00.000-05:00

This is the best thing ever. It reminds me of the old metal box fan we used to have. I would drop bits of food into the spinning blades to watch it get diced up and fly off.

Stupid People on Stupid Trains

2006-12-12T04:38:00.000-05:00

My mom came to Pavia to see my crib last week. In spite of the uninviting weather I think she had a good time and most importantly we ate at several amazing restaurants. Friday we planned on taking the morning train to Milan, exploring the city a bit, then taking a train to Geneva where I would spend the weekend and my mother would go back to the states. When we got to the train station that morning we discovered that due to a strike ALL trains from Pavia to Milan were canceled. Busliners were carrying soggy people to Milan, but there was a 90 minute delay. This would have put us in Milan right after a train left for Geneva and three hours before the next one. Unexcited by the prospect of exploring Milan in pouring rain, we decided to cab it. In retrospect, our first clue that the ride would be pricey was the BIG Mercedes cab that we had to order from the hotel, one of the only ones who would drive the ~50km to Milan. The good news is if you have a car like that you can go 160+ kph and not notice it. 30 minutes and 80€ poorer, we were in the ticket line at Milan Central. More good news: both the outbound and inbound trains were full and the only availability was first class.

The people in our first class cabin were funny. For the first hour of the four hour trip there was a bored poseur skater kid wearing Kermit the Frog neon green shoes blasting his iPod. It didn't make sense for him to take first class, so I chalked him up to be the sort who burns all the nice polos and button downs his well-to-do family buys for him and goes shopping at the local goodwill, for enhanced "street cred". Then as he got up to leave he stretched and I realized he was wearing Versace jeans and a D&G belt, and the tail of the shit poking out from under his Anti-Establishment Slogan bearing sweatshirt read Valentino. I wanted to jump up and shake him and scream at him, "You spent $800 and all you got out of it was a blaring self-proclamation as the world's biggest douchebag!" But I just watched him leave, poor posture and all, bland American rock in his wake.

This left us in the cabin with a couple who belonged in one of those yachting magazines. If you ever go to the newsstand and pick one up, you will notice two things immediately, (a) boats are outrageously expensive, and (b) every picture containing water or sky will invariably feature one or more attractive women in the background. (never in the foreground, unless they're selling something) These magazines cater to the wealthy aging man who needs reassurance that no matter how unpleasant or physically repulsive he is, sheer wealth guarantees him the company of some pretty 20something who wears too little and talks too much. I'm exaggerating, the mister wasn't disgusting and the missus wasn't dumb, but I'd rather chew aluminum foil than listen to them argue for another three hours. Background information now. French white male, early to mid 40's, lives in Paris, worked many years for a multinational waste management company. About 10 pounds overweight. Ukranian white female, mid 20's, moved to Paris 6 months ago, long blond hair, perfect complexion and 6'1" tall. Body weight around 45kg. Most likely a model of some sort.

Let's examine how they interact with some sample conversation topics.

"How come you won't speak French to me? How am I supposed to learn if you don't help me practice?" 3 minutes of arguing, followed by 20 minutes of pouting and glaring and icy voices.

"You can't do math. We were in Italy for 6 days. If you can't figure that out, you are an idiot. Here, let me show you..." he proceeded to count, as if to a 5 year old, the days on his fingers to show her just how wrong she was. It's an honest mistake to miscount a day, especially when time zone changes are involved, but he was really laying into her. What I ought to have done is bludgeoned him over the head with a massive integral and left him at the next stop bleeding Taylor expansions from his ears. This put her in the mood to take a 45 minute nap, the accompanying silence almost making up for the awful row that preceded it.

Finally, not content to sit in stony silence for the rest of the ride, they started up again when the mister's grand views of unifying Europe collided with the missus' views on eastern European politics. My mom was having a conversation with the mister when the missus apparently couldn't take it anymore and started to interject. It seemed like she actually had interesting things to say, but her English didn't come out that fast and he just bullied her around until she shut up. That wasn't awkward either. In retaliation, she took his iPod, which he requested back no less than 5 times, and turned it up louder than the label whore who was in the cabin earlier.

In summary, if you question your ability to display rudimentary etiquette for the duration of your train ride you are advised to stay home or walk. Otherwise, someone not so nice will hide plans for a WMD in your passport so that you are seized by border control and deported to the United States.

Fun in Rome

2006-11-28T08:28:00.000-05:00

There are lots of creative ways to ensure a lump of coal in your stocking for Christmas. For example,

stealing candy from a baby
being a lawyer for the NRA
cutting 200 people in line to the Vatican City

This weekend I was in Rome with Nancy, one of my analog circuits buddies from MIT. We got up nice and early Saturday morning to beat the inevitable queue into Vatican City. On the way there we discovered a 24 hour bakery which sold gargantuan focaccia and cream puff pastries filled with heavy cream, rum and nutmeg. That's the best pastry I've ever had. It was dusted with powdered sugar, and after I finished it I realized I looked like (rightly so) I had an addiction. We get in the already sizable line and waited, just like everyone else. After 20 minutes the line had extended behind us a long block and around a corner, which was easily 200 people. All of a sudden, a shifty little Frenchman in a sharp blue coat edged in sideways a few people up from us. Initially everyone behind him looked at one another, a universal "WTF?" Then came the hemming and hawing and throat clearing, which the stubby Frenchman ignored. Then came the outbursts in several languages (I counted English, French, Italian and Spanish at least)[on that note, I bet there are dorks who insist on "speaking" Latin when they visit] whereupon finally the stumpy Frenchman turned, issued a few self-righteous dismissive "Oui"s and didn't budge. As we rounded a corner, the Frenchman was pushed back a few places to us. I don't typically do well with these public confrontation things. Nancy though, in a stroke of brilliance and slight of hand, managed to untie the louse's shoe without him noticing. We pushed ahead of him, and a few minutes later we saw him get savagely trampled as he stooped to tie it, which was the most compelling evidence of transcendental power we saw all morning, moaning people in the Sistine Chapel included.

My MO for exploring new cities is simple. Avoid unnecessary forms of transportation, bring a dictionary, and eat indiscriminately. For better or worse the randomness of our food exploration was colored (guided) by Nancy's printouts from the New York Times, but we did decently. Aside from stumbling upon the 24 hour bakery in the morning, we had fluffy gelato from a small (but apparently NYT visited, a fact proudly displayed on a barely legible photocopy of the exact article we were carrying pasted to the window) but creative gelateria. There were also rich sweets and fresh focaccia from a bustling bakery, and a very nice dinner in the Trastevere.

An in-depth analysis of focaccia follows. To my dismay, Roman (as made in Rome) focaccia is too much like Napolean (as made in Napoli, opposed to Napoleon focaccia, the small pieces of focaccia which try and take over the bakery) pizza, in that the bread is baked a little longer until more brittle. This mitigates one of the best properties of focaccia from the north: incredible olive oil retainment. My favorite focaccia bianca is plain focaccia which is then sprinkled with rock salt and soaked in a tray of olive oil until the bubbles of the bread form a lattice of translucent threads suspended in a sea of olive goodness. A 6"x6" square can weigh 200g, and has to be eaten in a plastic bag because the oil squeezes out when you bite into it. In Rome, focaccia bianca is more like thin chewy bread brushed with olive oil right out of the oven. It's good sure, but demanding critics call for more oil.

The fountain of Trevi retains sharply in my mind. From an aesthetic perspective it's striking in a way even most of the other Roman relics can't compare, in that it is integrated into the side of a building. Rome is architecturally exciting because ancient structures and new buildings intermingle, but the fountain is carved right out of the front of a palace. Well, the real reason I remember it so well is because we walked past it four times. The first moment of our random walk, if I may. By the end of our day at midnight, or 16 hours since we started, we conservatively estimated our walking at 12 miles. Consumption between us: one giant slice of focaccia, a cream puff, a slice of tiramisu, an apple tart, two cups of gelato, four slices of thinner focaccia (two plain and two with tomatoes) two berry tart pastries, an apple strudel, and a three course dinner. Not stuffing, but satisfying.

Another great use of ancient ruins: make an abandoned KITTY shelter. Can you spot them all? Mmmmm kitty.

Rome is very pretty at night.

I feel I should clarify, for those who don't like bad humor, that the lout was not actually trampled. He buddied up with some other French folks about 10 places behind us and stayed there. Nevertheless, if He really wanted to sell the product (to borrow the words of Tom Lehrer, while introducing his modest example of modern vernacular music The Vatican Rag) I feel like we should have witnessed instantaneous, merciless punishment. Not a good sign when one can't keep his or her own house in order.

Colin teaches English With Unexpected Results

2006-11-16T14:34:00.000-05:00

If you can get over yourself for a minute and stop making fun of me, you'll appreciate the humor potential.

Last night, I went to dinner with my coworkers at a fancy restaurant. Our office is in the process of moving into a new larger building (true to their word, they put card key access on the server room door) and as good Italians, we used this event as an excuse to eat and drink. A brief recap of the meal: 4 appetizers, 4 main courses, a cheese course, and a desert. Total time: 4.25 hours. Total number of slices of lardo consumed: 6. Lardo is not the IT guy's nickname, it's cured pork similar to prosciutto except it's made entirely of fat. Imagine albino bacon. It's FANTASTIC. The whole dinner was good, but lardo made it special.

At one point during our dinner conversation a coworker teased me for mispronouncing a word. I hate the Italian triple vowel combinations, they're a full-mouth workout. In retaliation I decided to teach her some simple English. The two phrases I taught her were (1) "What are you up to?" pronounced, "Wha'cha up to?" and (2) "What is up, dog?" pronounced "'tsup, dawg?" It should be noted that they all speak English well enough to understand the phrase if ennuciated slowly.

This had unexpected results. Imagine walking into an office in the morning and hearing some guy go, "What's up dog." with zero intonation. Or, while we were packing some boxes today, I experienced this conversation: (in Italian)
me: What are you doing?
her: Cleaning the keyboard.
me: Don't you think you should turn off the computer first?
random guy passing by: "What's up dog"
me: What did you just say!?
her: What's with this "dog" thing! I totally don't get it, what's the point of putting that at the end of the sentence!? "Dog"!! Sheesh!

What's also funny about it is whenever one person says it, another guy nearby will want to be let in on the secret, and then what follows is a long winded explanation of "Cosa succede, cane?" explaining how the "dog" is an integral part of the sentence yet each person can't for the life of them explain what it actually means. I guess to them it would be like saying, "What's kickin', lampshade?"