A practical method for measuring signal to noise ratio

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

truzz

Well-known member
Joined
Oct 25, 2009
Messages
47
Location
www
Hi,

Please excuse my newbie question.
Already done a search with no luck.
What is a correct and practical method to measure S/N ratio of a device?

Thanks in advance.
 
This is a pretty ghetto way but.

Using a daw, look at the meters with nothing connected to the input (usb, fw, line etc) and write down the
level you see in terms of the residual noise level. It should be waaaay down.

Hook up the device to test to the daw input, short the device under test's inputs and repeat. Any extra noise is largely, but not entirely the noise added by the device under test.

There are better ways.
 
lemme see... 6am in the orning here, so this could be wrong but.

there's some software that'll do this for you. open source too.

google "rightmark" -- it uses your soundcard to do some tests.

bear in mind, if your soundcard is a POS, then your results may reflect that.


/R
 
Here's the most practical method I know.

1.) Crank it. 2.) put on some headphones. 3.) Decide if the noise level is acceptable or not.
 
You need a true RMS voltmeter.
I have a Tektronix TDS2000 digital scope, and a Hameg analog scope, but I'm not sure about their accuracy within this kind of measurement.
Much better something like an Agilent 3400A. Grab it if you can. It has 10Meg input impedance, this reduces the reading error a lot.
Probably there are better and more appropriate instruments to read this, with A-weighting et al.
Guys, here, please suggest some piece of gear!

Do not forget to terminate your DUT input with a resistor of value of your source expected impedance.

Step 1)Define SNR mathematically. The SNR is defined as SNR = Ps/Pn, where Ps is the average power of the desired signal, and Pn is the average power of the undesired noise. These power levels must measured at the same points and within the same bandwidth of the system.

Step 2)Study the root mean square (RMS). RMS is a means of measuring a varying quantity. It is especially helpful for measuring waveforms, such as electromagnetism or sound where the quantity changes in a statistically predictable manner.

Step 3)Calculate the SNR when the power of both the signal and noise is measured across the same impedance. Under this condition, the SNR may be calculated as SNR = Ps/Pn = (Rs/Rn)^2, where Rs is a measurement of an RMS amplitude (typically voltage) of the signal, and Rn is a measurement of an RMS amplitude of the noise.

Step 4)Examine the decibel (dB). A decibel is a measure of any quantity relative to a known reference level. It is a logarithmic unit, which allows it to easily represent very small or very large numbers.

Step 5)Express SNR in decibels. This is defined as SNR (dB) = 10log10 (Ps/Pn), so SNR (dB) = 10log10(Rs/Rn)^2 = 2(10)log10(Rs/Rn) = 20log10(Rs/Rn).


Hope it helps.

Respect,
Val
 
What a load of shit!
Step 1.  Buy a decent peice of test gear. Anything by HP.
Step 2. determine the clip point. Apply tone from a good oscillator. Wind it up till it clips.
Step 3. Take the tone off. Wind the meter down till it reads the noise.
 
radardoug said:
What a load of shit!
Step 1.  Buy a decent peice of test gear. Anything by HP.
Step 2. determine the clip point. Apply tone from a good oscillator. Wind it up till it clips.
Step 3. Take the tone off. Wind the meter down till it reads the noise.

What you describe is a method to measure dynamic range, not SNR.

JDB.
[in what bandwidth are you measuring? How exactly do you define the clip point?]
 

Latest posts

Back
Top