THD measurement - how to?

[Pjotr]

Originally posted by okapi

can anyone recommend a good, $350 or less soundcard for this type of work? preferably one that can do square waves. i have seen this page which has instructions on how to mod the M-audio delta for decent square wave reproduction

Uhm, that link doesn't work ???

Good PCI cards are the already mentioned M-Audio Audiophile 2496 and the Audiophile 192. The 192 has balanced inputs and outputs which can be an advantage when measuring amplifiers to avoid ground loops. Concerning distortion performance they are almost equal.

When it comes to USB the EMU 0202 and the EMU Tracker are a big bang for the buck. The Tracker is a more advanced version of the 0202 and has phantom power to power condenser microphones. The noise level is very low approaching the 24 bit theoretical level when using averaging. But despite they can sample at 192 kHz, they are only practical usable up to approx 50 kHz signal bandwidth.. Above 50 kHz the noise shaping mechanism of the converters starts rising the noise level sharply. This will you give wrong THD or THD+N figures. So practical sample rate is limited to 48 kHz then.

[okapi]

sorry about the bad link.

here is the new one.

[tritosine]

emu 1212m pci soundcard has the AKM flagship adc, converters are on seperate PCB not connected to PCI bus.Definately worth modding. Dont know about the m-audio.

[Pjotr]

Originally posted by tritosine
[B]emu 1212m pci soundcard has the AKM flagship adc, converters

Yes, as do the 0202 and the Tracker. But how does the 1212m perform at 192 kHz sample rate concerning noise at high frequencies? Do you have any loop- through measurements available?

[jackinnj]

Originally posted by okapi
i am writing a set of procedures for amp and speaker evaluation. I have a early version up and running but my THD measurement is higher than expected for a LM3875 based amp. This result made me wonder if my method for calculating THD was correct.

Now that the economy is hitting the wall, a lot of high quality analyzers are coming into the market. You'll also see some of the Tektronix units at knock down prices. The Boonton 1120/1121 and HP8903B are programmable via GPIB.

If you are starting out, you'll find that one number is going to give you quite a bit of information -- THD% at 1 or 2 kHz -- and you can do this quite well with an HP334 or HP339.

[Iain McNeill]

Originally posted by okapi
would you mind posting your matlab code here for the Farina method.

sure, I think this is OK as long as the credit to Angelo Farina is recognized.

This was the tricky part, actually making the ESS signal and it's counterpart, the inverse filter. So you'd run the m-file, use wavwrite to create the recording, play it through the DUT and record the output.

Extracting the impulse response is a simple convolution of the DUT recording with the inverse filter.

All the impulse responses (linear + all harmonics) get separated along the timeline of the impulse response. You can pick a particular impulse and do an FFT for freq response.

Automating the collection of the harmonic IR's is a bit tricky too. I can post a snippet of that if anyones interested. As I say, I haven't worked out the calibration on it yet so it's always dBr.

PS, you have to change the extension from .txt to .m
(or not)

[tritosine]

Originally posted by Pjotr


Yes, as do the 0202 and the Tracker. But how does the 1212m perform at 192 kHz sample rate concerning noise at high frequencies? Do you have any loop- through measurements available?

no i dont even have that soundcard, but a lynx aes16 (hehe) , but a buddy of mine modded one with diamond buffers & did a complete redo basically.

I doubt very much that the 0202 or the other one you mentioned has the 5394 ,cause another buddy of mine has 0404 usb that has an inferior chip for sure compared to the 1212m ( and btw that 0404 used to be praised for its "block transfer" based USB drivers, not because the flagship adc chip, what was a rather overused term when the 1212m came out), so it was bested by an old RME converter box , eguipted with an older AKm chip (538x) .

BTW there s also the PCM4222 ev-kit from TI, but that could do better in the jitter department ( 74lvc clock divider and buffering), mod-wise.

[Pjotr]

Originally posted by tritosine


I doubt very much that the 0202 or the other one you mentioned has the 5394 ...........................

No need for that, check E-MU's website: http://www.emu.com/products/category.asp?category=610 It has the AK5385 which performs also pretty good.

[joma123]

Originally Posted by Iain McNeill

Automating the collection of the harmonic IR's is a bit tricky too. I can post a snippet of that if anyones interested.

Hello everybody,

I've implemented a code snippet in C++ according to FARINA which collects the harmonic IR's.
Unfortunately the curves scatter. I suppose that I didn't choose the right length and starting point in the time domain of the according signal.

I would be glad about any help.

Greetings

joma