192KHz input and output system

Hi,
I am an electronic engineers but don’t know much about audio specifically.

For testing purposes (audio and non-audio) I need to setup a system that:
a) creates waveforms and/or noise at 192KHz
b) outputs that waveform at 192KHz (i.e NOT down-sample it or low-pass filter it)
c) acquire/records to file a feedback (“result”) waveform coming back from the output of the system under test

Few questions:

  1. what is a relatively low cost way to generate true 192KHz output? I have come across some DAC on Amazon (less than 80 USD) that claim 192KHz bandwidth. But I assume I need to use a software that can output that frequency on the PC. I read the PC/Windows itself does not output more than 44KHz freq. Can you please suggest a solution/setup? Both software and hardware that are relatively low cost (<100USD)

  2. same to acquire the feedback signal. I assume a good quality sound card. But would I need special software? Or can I use Audacity?

  3. what format can I export the file into from audacity (that supports playback at full 192KHz)?

Thank you :slight_smile:

I dunno.… That’s tricky for a couple of reasons. You probably should be using a signal generator and oscilloscope.

Audacity is intended for audio so it’s not always appropriate for scientific/engineering applications. The same goes for soundcards & audio interfaces.

a) creates waveforms and/or noise at 192KHz

Audacity can do that… Maybe… The sample rate has to be twice the signal frequency (Nyquist sampling theory… at-least one sample for the top-half of the cycle and one sample for the bottom half). You can set the project rate to 384kHz, but if you generate a 196kHz sine wave all of the samples line-up with the zero crossings and you get a flat-line. A square wave will work, or you could use Nyquist to generate a cosine wave (where the samples line-up with the peaks. (With one sample per half-cycle at the Nyquist limit there’s no difference between a digitized square wave and sine wave.)

I’m not sure how you are defining “noise”, but I assume white noise generated by Audacity (which you could optionally filter) will go up to the Nyquist limit. I’m pretty sure white noise is just a random number generator.

Or, I MATLAB (or one of the MATLAB clones) can do anything that’s mathematically possible.

b) outputs that waveform at 192KHz. (i.e NOT down-sample it or low-pass filter it)

Most soundcards/interfaces don’t specify the audio (signal) limits. There’s almost always a (unspecified) low-pass “smoothing” filter and that filter might not automatically adjust to the sample rate… It might be fixed somewhere above 20kHz so the full audio range gets through… I once had a cheap soundcard that I connected to an oscilloscope and I was surprised to see an un-filtered stair-stepped waveform. But if I remember correctly it was running at a fixed sample rate of 44.1 or 48kHz.

c) acquire/records to file a feedback (“result”) waveform coming back from the output of the system under test

Typically, a soundcard/interface can record whatever it can put-out, but the ADC requires a low-pass anti-aliasing filter, and again that filter might be limited to the audio range even at high sample rates.

I have come across some DAC on Amazon (less than 80 USD) that claim 192KHz bandwidth.

Probably not-exactly true. Any soundcard can play a 192kHz file (192hKz sample rate) because the driver will re-sample as needed to work with whatever hardware you have.

If you get ASIO hardware (hardware with ASIO drivers) and an ASIO application (Audacity is not ASIO “as shipped”) it’s not supposed to resample.

Are you referring to 192 kHz “sample rate”?

There are many sound cards that support 192 kHz sample rate, but I’d expect most of them to roll off frequencies above 20 kHz (low-pass filter) to avoid interference in the audio band.