Designer’s Guide Community

ShanoShih

New Member

Offline

Posts: 2
Taiwan

Comparator Noise && Correlated-Double-Sampling
May 04^th, 2006, 6:44pm

When there is a comparator with autozeroing configuration,
how do I get the comparator input refered noise?
with autozeroing,the low frequency noise(1/f noise) will be reduced,
and the broadband noise(thermal noise) will be folded back to the baseband.
Therefore, the calculation is too complicated.
Does anyone have methods to get that?
maybe simulation methods?
Thanks a lot for any response

Back to top

IP Logged

huber

Community Member

Offline

GO BEARS!

Posts: 45
Los Angeles

Re: Comparator Noise && Correlated-Double-Sampling
Reply #1 - May 4^th, 2006, 7:04pm

If by "comparator" you are refering to a differencing amplifier, then you can easily obtain the input-referred noise with a pnoise simulation (spectreRF). Ken has a paper titled "Simulating switched-capacitor filters with SpectreRF" in the Analysis section that explains how to set this up. If you are refering to a latch then you might have to change the circuit topology slightly for the noise sim to kill the regeneration, but I'm not sure about this.
-Dan

Back to top

IP Logged

vivkr

Community Fellow

Offline

Posts: 780

Re: Comparator Noise && Correlated-Double-Sampling
Reply #2 - May 8^th, 2006, 2:02am

Hi,

If you have a decent bit of preamplification before your latch in the comparator, then your noise will be dominated by this.
The easiest method then would be to simulate the noise of the preamp alone. You can even do a rough estimation of the
increase in wideband noise due to autozeroing by hand.

You can write the noise transfer function seen by a noise source tied at the input of the comparator, in form of difference equations.
Then, the noise power gain is easily obtained using Parseval's theorem. If you use a simple noise analysis for your comparator now,
then you know the unaliased thermal noise. This multiplied by the noise gain will give you the net wideband noise.

The typical CDS scheme gives a noise power gain of 2 for wideband noise.

The whole process takes 5 minutes, and the results are quite accurate, and don't depend on any simulator settings

Regards
Vivek

Back to top

IP Logged

ShanoShih

New Member

Offline

Posts: 2
Taiwan

Re: Comparator Noise && Correlated-Double-Sampling
Reply #3 - May 14^th, 2006, 11:23pm

Dear huber:
Thank you for your suggestion.
I have read the paper "Simulating switched-capacitor filters with SpectreRF".
But I still have some questions.
My comparator is just a cascade of differencing amplifiers
with output-offset-storage in the first stage and input-offset-storage in the second.
The questions:
1.in the pnoise analysis,which net(or nets) should I treat as output nodes?
If I choose the inverter output, it is already a digital out,
and the amplitude noise is compressed.
2. in Ken's another paper "Noise in mixers, oscillators, samplers, and logic: an
introduction to cyclostationary noise". There is a mention that if the output of one stage caused
the subsequent stage to behave nonlinearity, the �noisetype=correlations� should be chosen.
Should I do that?

Thanks again for your kindly response
Any response is appreciated.

regards
Shano

Back to top

comparator.jpg

IP Logged