How good of a measurement tool is a soundcard?

[Speakerguy]

Hey,

I need a quick and easy way of measuring the RMS value of a sinusoid. It will be in the range on 1-10kHz. It will be generated by a bench signal function generator.

The signal to be read is a sinusoid, at a constant frequency (yet to be chosen, but in the range given above). The sinusoid will be approximately 0.5Vrms. It will fluctuate about 30mV during the the experiment (so the signal will be a sinusoid, constant freq, 500mV-530mVrms).

The signal I'm interested in is the differential, ie the 30mV. Can I get better than 1mV resolution out of a good computer soundcard (like an M-Audio Audiophile 2496)? Can I get that kind of resolution out of a crap soundcard, like whatever is built into my HP pavillion laptop? (Nice laptop, but I don't assume the built in sound is too great).

I should add that I've tried using a Keithley 2100 bench top multimeter, but cannot get more than 10Hz readings on AC voltage. I can get 2KS/s on DC measurements, but not AC. I think it has to store a number of values (specifically 20) and then calculate the RMS value, and it reloads all new values (the next 20 it measured)and recalculates (as dropping the least recent value, including the most recent measured value, and then calculating the RMS voltage, which would then give you an effective 2Khz sample rate, if there is such a thing as AC measurement sample rate).

I may be able to increase the signal by a little, but certainly not any more than 50% (ie, 750mVrms).

This is for work, and I don't get paid very much, but I'll send you an e-cookie or something for any help

 

[Speakerguy]

nobody? anyone?

 

[Nigel Goodwin]

You want an AC millivoltmeter, which wouldn't usually measure RMS anyway, a multimeter is likely to be pretty useless for it.

A soundcard is probably better than your multimeter for the job!.

 

[Speakerguy]

I've been using a Keithley 2100 benchtop multimeter, and you are right - not the tool for the job.

It does 2KS/s on DC measurements, but no more than 10S/s (just 10) for RMS AC measurements.

The decision now is to go with the soundcard, or have the company buy a $500 NI ADC module. Which we could use on future projects, but if a soundcard will do us for now or even be better then I'd want to use that. The NI module has 0.3mv resolution at its stated bit depth, if it can actually meet it (16bits) instead of only getting 14 or 15 like a lot of products.

The plus with either of these methods (soundcard or NI module) is I can straight record the signal, then notch the excitation voltage, and just get the differential signal. I'm coding that up in Matlab right now while I waffle over the decision on how to measure it.

 

[Nigel Goodwin]

The obvious trouble with 16 bit precision is that it's likely to be below the noise level anyway.

 

[ericgibbs]

nobody? anyone?

hi,
I was interested in this 'ecg.zip' Visual Basic application, using the sound card.
[rainy day project!]

I havn't used it yet, but it may give you some pointers.