The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Mixed-Signal Design >> PNOISE analysis for switched-capacitor circuits
https://designers-guide.org/forum/YaBB.pl?num=1078362606

Message started by cwyeh on Mar 3^rd, 2004, 5:10pm

Title: PNOISE analysis for switched-capacitor circuits
Post by cwyeh on Mar 3^rd, 2004, 5:10pm

Hi,

For SC circuits, an OFF switch should not contribute noise theoretically. But I have recently found that it contribute more noise than an ON switch from SpectreRF PNOISE analysis. I think that the simulation result doesn't make any sense. Do I miss something? Please answer. Thanks.

By the way, I specify tdpoints (at the ends of sample and hold) very carefully.

- Chingwen

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by August West on Mar 3^rd, 2004, 6:23pm

Are sure the switch was off when it contributed the noise? The noise could have been generated when the switch was on, stored as charge on a capacitor, and is affecting the output at the time you make your sample.

-August

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by Chingwen on Mar 3^rd, 2004, 8:47pm

Yes, I am sure that it's "off" when contributing considerable noise. I know that the simulation result doesn't make any sense. Anyway, I'll try to get some help from Cadence.

- Chingwen

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by August West on Mar 4^th, 2004, 12:32am

Can I ask how it is that you are sure that the noise you are seeing was not produced when the switch was on? Is it that the switch is never on? If it is occasionally on, is there no chance for the noise produced while it is on to be stored on a capacitor? If it can be stored on a capacitor, is there no possibility that it is stored long enough for the effect of the noise to be observed at the output?

-August

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by cwyeh on Mar 4^th, 2004, 9:25am

Hi August,

They are good points. I thought of all of them before.

1) I performed PSS analysis first. It created steady-state results for one period (including hold and sample). Based on the results of the PSS analysis, I specified tdpoints at the end of hold and sample very carefully. Then I performed PNoise analysis. SpectreRF allowed me to check the noise contribution from each device at any specific timepoint.

2) It was found that the "sample" switches, which were off at the hold time and on at the sample time, contributed considerable amounts of noise at the end of hold. I am sure that the noise does not come from the previous sample. It's because, if so (noise accumulation), the noise amounts of the switches at the end of hold should be larger than that at the end of sample. But the simulation results showed that the former was smaller than the latter.

- Chingwen

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by August West on Mar 4^th, 2004, 10:41am

Let me see if I understand what you are saying. Can we assume a circuit like the one shown in Figure 1 in http://designers-guide.com/Analysis/sc-filters.pdf, and further assume that the switch is opening and closing periodically?

What I heard you say is that you are performing a strobed noise analysis using SpectreRF's tdnoise option and that you ran two noise simulations, one that samples the noise when the switch is open and one that samples when the switch is closed. You found that the noise you measure when the switch is closed (sample) is larger than when the switch is open (hold). Is this correct? Is it substantially larger?

If this summary is correct, I am surprised. I would expect the noise levels to be basically the same. But at the same time, I don't understand your reasoning as to why this indicates that the switch is injecting noise when it's off. Could you explain that more fully?

Could you also tell me if you have tried disabling the clock and performing the same noise analysis. Does the switch still generate significant noise when it is always off?

One last question, is the anomalous noise you are seeing due to thermal noise sources or flicker noise sources?

-August

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by cwyeh on Mar 4^th, 2004, 11:42am

Hi August,

Thanks for your attention once again. I think that you have pretty understood my points. I did see what you were surprised at.

For a SC S/H circuit, one can add noise PSD at sample and hold together to get the total noise PSD. From my simulation results, the noise PSD value at sample (kT/C simply) was very accurate while that at hold (kT/C plus opamp noise) was somewhat larger than expected.

Since the "sample" switch has more noise when sampled (sample mode), it's assumed that the noise at sample would not be added to that at hold in SpectreRF simulation. It's why this "implied" that the "sample" switch injected noise at hold when it's off in simulation results. As a matter of fact, I am not quite sure if the noise at sample was added to the noise at hold or not in SpectreRF simulation.

Besides, I am talking about thermal noise only here.

- Chingwen

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by August West on Mar 10^th, 2004, 8:23am

How much larger is the noise when in sample mode?

You say there is an opamp in the circuit? Can you describe the circuit?

-August

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by alice_alice_nc on Mar 17^th, 2004, 2:30am

I also want to simulate the sc filter ,but I don't know the sc filter is simulate by the netlist or by the schematic ,
and whether the ideal switch in the analogLib can be use as the switch or not?

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by alice_alice_nc on Mar 17^th, 2004, 2:38am

I studyed the netlist and found there is a sample and hold block (used sh.va file) ,why is there a sample and hold ? what is the performance of the sample and hold ?
and I created a symbol from the sh.va ,and I added it to the schematic , setup the parameters in the symbol,but it can not work ! why?

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by August West on Mar 17^th, 2004, 11:48pm

The sample and hold is part of the test bench. It is present because the output of most switched-capacitor circuits is sampled. The sample-and-hold assures that the output is observed only at the sample points. Without the sample-and-hold SpectreRF observed the entire waveform. The frequency response, noise, and distortion are all much different when you observe the entire waveform as opposed to just the sampled waveform.

-August

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by ethan on Jul 7^th, 2005, 9:56pm

August West wrote on Mar 4^th, 2004, 10:41am:

.........
What I heard you say is that you are performing a strobed noise analysis using SpectreRF's tdnoise option
......

-August

cwyeh wrote on Mar 3^rd, 2004, 5:10pm:

Hi,
.......................
By the way, I specify tdpoints (at the ends of sample and hold) very carefully.

- Chingwen

Does anybody know where I can specify "tdnoise" and "tdpoints" two options? Thanks a lot.

Title: Re: PNOISE analysis for switched-capacitor circuit
Post by Ken Kundert on Jul 8^th, 2005, 12:19am

In the Spectre netlist you use the noisetimepoints parameter of the pnoise analysis. In Artist, you go to the PNoise analysis form, go to the bottom and specify the Noise Type as being time domain, set the number of points and then click on Add Specific Points.

-Ken