Archive for the 'MIT' Category

Built Some Speakers (Courtesy of Bose!)

Last Friday, Jordan and I took the one-day speaker-building workshop offered during IAP by some Bose Engineers. We each built a set of some surprisingly good speakers!

(side point: for those unfamiliar with IAP, it’s MIT’s way of saying “do whatever the hell you want for a month and maybe even learn something if you want.” I’m TAing an introductory electronics class that focuses on building cool stuff like little EKGs that kids can take home)

Anyways, I thought I’d spend at least a day this month actually learning something. This class spent a few hours going over the theory of how speakers work and then we built some simple speaker “cabinets” to house the transducers that Bose uses in their 201-series desktop speakers. The cabinet? Well, it’s a piece of PVC pipe. 😛

(yes, I installed one of them upside down and yes it’s annoying me)

Looks pretty weird, right? Well, the sound’s surprisingly good! I didn’t expect it to be nearly that good for the materials we’re using. Basically, we installed the woofer and the tweeter onto the PVC and then ran wires to the back. Then, using a resistor and a cap, we created a damped first-order high-pass filter to prevent the tweeter from being driven by low frequencies. Finally, we threw some fluff into the PVC to dampen some resonances. Here’s a shot of the extremely advanced crossover and the fluff.

It wasn’t very technically difficult to build these speakers. Anybody could really do it so long as they had a drill and some PVC; you don’t need much more than that. I was hoping that we’d actually build the speaker cones, but those were all pre-built (they had a pre-assembled one to show how they work).

These cones can be driven with up to 100W or so (though the Bose site claims that the speakers that they’re usually in can be driven to 120). I’ve got a second set of outputs on my stereo amplifier, so I might try to hook them up to those outputs. They also gave us a simple 15W amplifier kit based on the TPA3122D2 chip from TI.

When Jordan and I were testing them out, we managed to find a really strong resonant frequency of the PVC pipe. When a certain note hits, the pipe resonates and makes a really gross sound, louder than everything else. Good cabinet design is really crucial to making things sound good. Except for that frequency, we didn’t find any other glaringly bad parts of the speakers’ sound.

All this made me wish all the more that MIT had an acoustics class. Bose himself used to teach MIT’s acoustics class, but nowadays, there’s nothing. I’ve always had a sort of fascination with sound, but never really got motivated to build acoustics-related projects. Maybe I’ll work on modifying these speakers to sound better and see where things go from there.

End of Semester, 6.111 Final Project

Well, the semester’s finally over. This one kept me especially busy even though I was only taking the average 4 classes: 6.012 (Microelectronic Devices and Circuits), 6.034 (Artificial Intelligence), 6.111 (Introductory Digital Systems Laboratory), and 21H.207 (The Energy Crisis). I enjoyed all of these classes. In addition, Next Make has finally gotten underway; we’ve received quite a bit of money from my dorm to build a lighting system. Next semester, I’ll be involved in a variety of activities and this January is definitely going to be exciting. In addition to working on the Next House lighting system, I’ll also be finishing my own lighting system and my door sign. I’m also TAing an introductory electronics class that I took in Freshman year.

By far, my busiest class was 6.111, “digital death lab.” This class taught the foundations of digital design with FPGAs. They had their own custom-designed FPGA boards: the 6.111 Labkit, complete with a 6-million gate Virtex 2 FPGA and a variety of peripheral chips. The largest contribution to the grade for the course was the final project, where students created some kind of design.

I decided to work on an idea that I came up with last winter, a parametric equalizer. Parametric equalizers are devices that effectively allow users to create arbitrary filters and apply them to an audio signal. This device was fairly simple; it took a “gain curve” — a mapping containing some divisor gain for each frequency bin of an FFT — and applied it to the output of a 1024-point FFT (a frequency domain signal). Then, it took an IFFT of the data to transform the signal back into the time domain and fed it to the audio output.

I’ve updated my verilog code, design report, and presentation slides. The design is small enough that it should be implementable one a Spartan 3 FPGA (such as the one on my Nexys2). The system only uses a mono signal right now; I simply need to duplicate it to generate a stereo sample. You can see a video of my demo here.