From excellent series of articles by Stuart Yaniger "Practical Test & Measurement" I've learned about Peter Millet's Soundcard Interface device. In short, this is an analog adapter for transforming voltages in the range from 20 millivolts to 200 volts into the range accepted by soundcard's line input—about 1 volt. The device also offers protection for the soundcard inputs from accidental high voltages.
The application of this device is for testing audio amplifiers. Below is a general connection scheme:
The amplifier under test is connected to a load. The load can be either real speaker or a dummy load—a resistor, for example. The soundcard interface is connected in parallel to the load. Since the interface's input impedance is about 100 kOhm, it draws only little current by itself (for a comparison, typical oscilloscope input has 1 MOhm impedance.) The output from the soundcard interface goes into a soundcard, which is connected to a PC (or Mac) running measurement software.
The input to the amplifier is provided directly from the soundcard's line output. Though if it's a mobile phone with an analog output or a portable audio player, the test signal can be played from the device itself.
What is so good about the Mr. Millet's device? First, you must build it yourself, which is fun. Second, it's cheap compared to "pro" measurement frontends. Let's see:
- QuantAsylum QA401 Audio Analyzer costs $450;
- Avermetrics Averlab costs $3000;
- Prism Sound dScope Series III can be found used for about $8000.
I've spent $265 (shipping excluded) on the parts for the Soundcard Interface. If you are in electronics DIY hobby seriously, you may have some parts already or buy them in larger quantities, which will take the price even lower. Sure, the Millet's device plus a soundcard may not deliver the same precision of measurements as those frontends, but for home projects it's enough.
The third good thing about this interface is that it doesn't require any specialized software. All the "pro" interfaces mentioned above come with dedicated software which you may or may not find easy to use. But for soundcard measurements, there are lots of audio analyzer software tools, some of them are even available for free. So it's possible to compare and choose what better suits your needs.
Speaking about alternatives, I must also mention recently emerged Jan Didden's L|A Autoranger which from its description seems to be similar to the Millet's interface, and has a comparable price. One great thing about the autoranger is that it automatically adjusts itself to the input signal level, like "pro" measurement interfaces do. Sounds interesting, I will perhaps try it some time later.
So far, I have only soldered up the device and calibrated its zero and full scale levels.
One really important issue that must be addressed is the level of noise produced by the device itself. While having the PCB lying open on my table, I couldn't get the "zero" level lower than 0.9 mV (P. Millet is saying that "zero" on his unit is 0.3 mV.)
Folks from the diyAudio forum report that putting the board into a shielded enclosure helps to reduce the noise. Since the enclosure suggested by Mr. Millet is made of plastic, I've ordered a couple of MuMetal Ultraperm Permalloy sheets to cover the box from inside. In fact, this is the same approach that E-MU was using for their line of audio interfaces, e.g. 0202 and 0404. Their bodies are made of plastic, but on the inside of the boxes there is isolating coating. Hopefully, this will help.
After I finish with shielding the enclosure and assembling the device I will start measuring the parameters of the device itself, and will try it on some headphone amplifiers.