The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Again PSS for SC sampling circuits
https://designers-guide.org/forum/YaBB.pl?num=1131602905

Message started by johannes on Nov 9^th, 2005, 10:08pm

Title: Again PSS for SC sampling circuits
Post by johannes on Nov 9^th, 2005, 10:08pm

Hi everybody,

I read the Kundert paper on PSS/Pnoise but still like to ask for your help in some questions.

I am simulating a differential SC sampling circuit applying an OTA, which itself uses a SC-CMFB. Thus if running the transient analysis it takes some clock periods to be stable (until the CM output level is fixed). The DC-OP is not constant each cycle until it reaches its final value (though it works then considering settling time and error only).

1 Is it allowed to run a PSS over this? Or do I have to fix the circuit (e.g. by reading the final DC data of the trans first) before starting the PSS?

2 When doing the Pnoise then, can somebody tell me what spectrum of the output noise I am to expect? Now, it still looks like 1/f + thermal , but shouldn't it look periodically?

3 I am sampling frequencies up to 8MHz (if this matters) with 40 MSamples/s. What maximum sideband shall I use, and from where to where shall I integrate to get the noise power? Like 1Hz to fc/2 only?

Thanks a lot, I really appreciate your help!

Johannes.

Title: Re: Again PSS for SC sampling circuits
Post by chase.ng on Nov 15^th, 2005, 10:33pm

kindly take a look on the http://www.designers-guide.org/Analysis/sc-filters.pdf

there is a tstab option in pss analysis that u need to fill in the time ur circuit need to stabilize. if u leave out the tstab, most probably the pss will not converge. to confirm that the pss is valid, check out the pss result with result browser. it works for me so dun worry about the sc-cmfb

the pnoise spectrum is the the paper by kendurt, it looks like 1/f + thermal to me... y should it looks periodic?

according to the pnoise paper, number of sideband determines the accuracy of the result, kindly take a look at the paper for more detail guideline.

about the total noise power, it should depends on ur antialiasing filter bandwidth in front of the sampling circuit. note that u should intergrate starting from 0Hz, otherwise u missed out a lot of 1/f noise.

hope that help.

chase

Title: Re: Again PSS for SC sampling circuits
Post by johannes on Nov 15^th, 2005, 11:18pm

Thanks a lot, chase! The tstab was exactly want I needed. Now it works as expected with me, I got the noise spectrum I wanted.

But it does not look like 1/f + white noise. I don't have flicker noise at all anymore, because it is folded up to the sampling frequency. I thought this is the case for all SH circuits, but maybe it's only with the certain switching scheme I looked up. I am doing a pipeline ADC gain stage with offset cancellation and two clocks - Don't worry about that and thanks for your help!
Johannes