Cheap Speaker Measuring

A Cheap Acoustic Measurement System

It simply blows my mind how many people build speakers without even trying to work with the real frequency response and impedance of the drivers they buy in the boxes they build. The tools to do so are easily available, and while learning to use them can take some time, I feel it is completely worth it. I would not even bother to build a speaker system if I didn't have a measurement system to work with. It would be an exercise in futility - the odds against coming up with the best system using the parts you have would be exceedingly slim, or else you'd have to devote so much time to the project that you wouldn't get to do much listening (which is supposed to be the whole point!).

Okay, ranting over - here is what you need:

A computer with a soundcard capable of full-duplex operation. Full-duplex means the sound card can play and record at the same time. I have used a SB Live successfully. Right now I have an SB Audigy 2. You don't need anything fancy - you are not trying to set up a calibrated lab-grade system (unless you are, in which case you should not need to be reading this).
Speaker Workshop, a really cool FREE bit of software that will let you measure such things as the impulse response, frequency and phase response, impedance and thus Thiel / Small parameters, and distortion (although I do not like the way it does distortion measurements - I don't think it corrects for the distortion in the soundcard). SW is a noise-based measurement system, like MLS. A point of note is that SW runs as a two channel system. This means you can measure 'real' phase. You don't have to rely on a Hilbert transform and the assumption that your speaker's minimum phase which makes me feel warm and fuzzy. You can also design passive crossovers with SW. It includes some basic goal seeking functions so you can optimize your designs. No, it's not as powerful as something like LEAP5, but it also costs significantly less. You may note some oddities in the software depending on precise setup, but really, can you complain when it's free?
Some cables to connect your sound card to your sound system and connect your mic to your sound card. This typically requires some 1/8" stereo to RCA converter cables and some RCA female-to-female plugs. Radioshack usually has this sort of stuff for outrageous prices.
A microphone - one option would be to buy a Behringer ECM8000 and suitable preamp that provides the phantom power it needs. You could probably do this for around $100, but in my opinion this violates the spirit of a cheap but good measurement system. I mention because some people may get squeamish about what I do recommend - building your own mic. Well, not really building your own - you buy a Panasonic mic capsule. A WM-61A and a WM-60AY are the usual suspects. One has lower noise but less sensitivity which would probably work fine. I think I have the other, but I'll leave it to you to read the data sheets. Anyway, these are cheap, like in the $3 range from Digikey, and less if you buy enough to hit the first price break. Now you may use this as is, connecting to the two terminals on the back of the (tiny) capsule, or you may go all out (in the diy spirit) and modify your mic capsule for lower distortion / higher SPL capability as outlined by Sigfried Linkwitz on his webpage, or alternately, on this mirror of his page.
Typically, you will also need a preamp for your mic. I built a preamp using the Texas Instruments INA103 chip. This is a glorified op-amp, so if you've done any work with those, you will probably not have a lot of trouble building something similar if you just follow the recommended circuit in the data sheet. Once again, you could also buy any number of cheap pro-sound oriented preamps, but what fun is that?
If you want to measure impedance (and T/S parameters), you will also need an impedance jig. You can make this for maybe $20 or so in parts that are typically available at Radioshack. See Eric Wallin's webpage for a nice tutorial. He also has some other useful stuff relevant to this topic.

And that's about it. Using SW, your computer generates a signal on the left and right output channels. One gets looped back to an input channel. The other one drives your audio system, whatever particular device you're wanting to test. Then you use your mic to pick up the sound generated by the speaker you're testing, and that gets recorded by the other input channel on your computer. Then the data recorded from the mic is compared to the data recorded from the output of the computer. Then lots of mathematical things go on and you can process your data in all sorts of useful ways in order to better tune your speakers. I would urge anyone contemplating getting into serious speaker building to try this type of approach before spending more money on fancy equipment and / or software. What if you build one set of speakers and never touch the stuff again? At least you haven't dumped hundreds or thousands of dollars on the greatest software out there. Even if you build speakers for the rest of your life, using a setup like this for a while will really let you / make you learn a lot, and you'll be better informed if / when you decide to get some more advanced equipment.