SNR measurement audio amplifier

[AllTheGearNoIdea]

Hi all

I have been looking into how to do a  diy SNR measurement on an old audio amplifier.  I have done the usual google search and pretty much everything i find just say using the "audio precision" test bench or look at the spectrum analyser and see difference between noise floor and peak undistorted output.

Now assuming I don't have a magic black box that I plug the amplifier into how can I measure basic signal to noise ratio.

It's the noise measurement part that's confusing me as it typically specified over a bandwidth and I think weighting filter are applied. From what I have read most of the articles quick go down the rabbit hole and mostly disagree with each other anyway.

I assume the key to this is measuring the noise voltage I can calculate the  noise power as I know the load impedance.  I assume portable DVM won't measure a noise voltage as it does not have the bandwidth, I assume a scope has the bandwidth but cannot trigger on random noise events.  I am excluding FFT scopes even though I have a good one and Also no sound cards.

Could anyone give me a simple test procedure using  low tech test gear if this is at all possible.  I'm guessing maybe in the past these  measurement  used a calorific type measurement. I wondered if I could measure temperature rise in the load resistors etc.


Many  Thanks
Regards Chris






So how was this measurement done in the old days.

[TimFox]

Measuring the noise voltage requires two basic objects:
1. A filter to define the bandwidth.  Ignoring hum for the time being, a simple R-C low-pass filter suffices.  Note that the equivalent noise bandwidth for a single-pole filter is 1.57 x (bandwidth at -3dB).  see https://www.k-state.edu/edl/docs/pubs/technical-resources/Technote1.pdf
2. An AC voltmeter, preferably true-rms.  If you use an average-responding, rms-calibrated meter, there is a further correction factor required, since the rms-calibration assumes a sine wave.  This is about 0.89:  see http://hpmemoryproject.org/an/pdf/an_60.pdf for a classic treatment of the different styles of voltmeter and their response to non-sinusoidal waveforms (noise on page 6).
A basic treatment can be found at  https://pdfs.semanticscholar.org/d770/d93b92abf067aab6e469ba816da0d205777e.pdf

Weighting filters are normally used with acoustic noise measurements, to simulate the response of the human ear.  For amplifier signal-to-noise ratio, you measure the noise in the relevant bandwidth (probably 20 kHz for an audio amplifier) and compare it to the maximum amplifier output or to the typical amplifier output, as required.

[Kleinstein]

A good scope might be suitable to measure the noise. It just depends on the voltage level and on the way data a stored and used for the FFT.

Similar the soundcard is a reasonable good way to get a noise measurement - it just needs a calibration signal and suitable software to do the calibration.

Today using an FFT is the more obvious way instead of an analog Filter.
Especially for the usual A-weighted noise the filter is kind of complicated in the analog domain.

[AllTheGearNoIdea]

Many thanks for the replies i will have look at those links provided. It looks like I'm going to have to do some work to figure this one out.  I have a HP8903 and I'm building a sound card interface at the moment but I want to have ago at trying to do this old school.  I have seen some of the weighted RMS volt meter around on eBay from time to time but don't own one yet.   

Thanks again

Regards Chris

[AllTheGearNoIdea]

That link to the HP volt meter guide is fantastic

Thanks again

[f5r5e5d]

http://audio.rightmark.org/products/rmaa.shtml free version is pretty standard - just google rmma and any sound card name

of course PC motherboard chipsets and interface parts may not be up to measuring actually good analog circuits

then you would need at least one of the better usb/PC sound cards

with input protection for the sound card you can look at noise floor with gain - then even motherboard chipset performance may be adequate

attenuation lets you make the other standard rmma measurements of a audio power amp

[AllTheGearNoIdea]

Thanks I already have the rightmark software and have been building a interface today. But I want to figure out how to make these basic measurement without any fancy test  gear  - kind of back to basics and I think I will learn more.

I have have a fairly good 24 bit sound card but annoying that it appears to have been designed to roll off the input at 20 kHz.  I was using this to do a frequency sweep on the amplifier and if done long hand using a signal generator and scope the frequency response is much wider than the sound card would have you believe.

But thanks for taking the time to comment. Best regards Chris