Designer’s Guide Community

luoshuer

New Member

Offline

Posts: 5

Spectrum
Nov 15^th, 2012, 3:06pm

Hi Guys

I am new to signal to noise ratio and may have really stupid question, which confused me a lot. I am try to interpret attached �DAC output spectrum.

(a)
If I want to calculate the rms noise voltage from DC to 40MHz, I can do square-root( integral ( noise power �from DC to 40MHz)), the noise floor is around -40dB.

so �VnoiseRMS= squareroot( �( 1e-4)^2 �* 40e6) = sqaureroot(4e-1)=0.632Vrms. �It it correct, I feel this is too HIGH, as it is close the the fundamental signal???

(b)
If I look at the fundamental signal, this is 25MHz,-3dB at the peak.
Do I immediately get Vsignalrms=0.5Vrms as 10log(0.5)=-3dB?

Or I have to integrate the signal power from 24MHz to 26MHz, then take squareroot to get Vsignalrms?

Many thanks,
shuer

[img][/img]

Back to top

SNR_1.JPG

IP Logged

wave

Senior Member

Offline

Posts: 117
Silicon Valley

Re: Spectrum
Reply #1 - Nov 15^th, 2012, 4:08pm

It's been a long while since I thought of this, so I'll be interested what others post.

I think you need to pull out the fundamental and harmonics, as they are not strictly noise. (and raise your noise summation!)

As for the signal, most tests do go slightly below full scale to avoid clipping and distortion. So your peak signal is rarely 0dB, but you are only looking for a ratio.

Look up some Data Converter mfgr datasheets and app notes for details. Good luck!

Back to top

IP Logged

aaron_do

Senior Fellow

Offline

Posts: 1398

Re: Spectrum
Reply #2 - Nov 15^th, 2012, 4:18pm

Hi,

You didn't take into account the resolution bandwidth. That's why your noise analysis is wrong. Secondly, if the Y-axis is dBV, then to convert to magnitude, you should take 10^(dBV/20). That will be the peak voltage. VRMS for a sinusoid is then 1/√2*10^(dBV/20).

-3dBV = 1/√2*10^(-3/20) = 0.5. But I think you arrived at the correct answer by coincidence. i.e. -3dBV is the only value for which your calculation would yield the correct answer.

Feel free to correct me if I'm wrong anywhere.

regards,
Aaron

EDIT: BTW, you could integrate from 24-26MHz, but the answer would not change much.

Back to top

there is no energy in matter other than that received from the environment - Nikola Tesla

IP Logged

luoshuer

New Member

Offline

Posts: 5

Re: Spectrum
Reply #3 - Nov 15^th, 2012, 5:46pm

Hi Aaron

Thanks for resolution bandwidth correction and indeed the Y-axis is dBV.

1) So for signal, indeed we need apply 1/√2 to translate the "peak" to "rms", 1/√2*10^(dBV/20)

How about the noise? Is it "peak" or "rms" when we look at the noise floor? Do we need apply same translation 1/√2?

2) Why there is no difference between "signal peak derived from -3dBV" and "sqareroot(integration from 24MHz-26MHz)"?

Best regards,
shuer

Back to top

IP Logged

ywguo Community Fellow Offline Posts: 943 Shanghai, PRC	Re: Spectrum Reply #4 - Nov 17^th, 2012, 1:30am Hi, Interesting. luoshuer, is that spectrum a simulation result? Aaron, why do you think the result does not change much when the integration limit changes? Best Regards, Yawei
Back to top	IP Logged