The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Maxsidebands parameter for SC amplifier (PNOISE)
https://designers-guide.org/forum/YaBB.pl?num=1562114931

Message started by vm511 on Jul 2^nd, 2019, 5:48pm

Title: Maxsidebands parameter for SC amplifier (PNOISE)
Post by vm511 on Jul 2^nd, 2019, 5:48pm

Hi,
I am trying to simulate a switched capacitor amplifier with correlated double sampling, and trying to do a PNOISE analysis on the same. In one of the Cadence documentations, it was mentioned that the method to obtain accurate maxsidebands is to keep increasing the value until the results are more or less constant.
I simulated the setup with maxsidebands till 2000, but the value keeps increasing. Moreover, I then tried the full spectrum option in the PNOISE analysis, and the answer is significantly lesser than what is obtained from maxsidebands 1000.
I'm unsure of what's the cause of this, and would be grateful if someone could help!
Thanks!

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by Ken Kundert on Jul 2^nd, 2019, 8:54pm

The total noise power will keep increasing as you increase maxsidebands, but at a decreasing rate. I recommend that you play with maxsidebands to get a sense of how the noise varies. Then back off of maxsidebands to a value that is reasonably practical. The result from PNoise analysis will then be an estimate of your noise, but you know it will be a little low. So the simulations need to out-perform your spec by a bit. You can estimate how much by seeing how much the total output noise varies with maxsidebands.

As I understand, fullspectrum is suppose approximate setting maxsidebands to infinity. However, I have never seen an explanation of how it works, so I always treat it with a bit of skepticism.

-Ken

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by vm511 on Jul 3^rd, 2019, 3:18am

Hi Ken,
As I increased the max sidebands from 1000 to 2000, the total noise power nearly doubled. Do the simulations require such a high number of sidebands?
Moreover, I had tested the PNOISE analysis on a simple RC circuit with a switch, and what I observed is that depending on the bandwidth of the circuit, there are cases where the PNOISE analysis significantly overestimates (double) as I keep increasing the max sidebands for a few cases. This happened especially in the case where the load capacitor was very small ~10fF. Whereas, for a few cases of the circuit bandwidth, the analysis does saturate to the theoretical value as mentioned in the documentations. Does the bandwidth of the circuit under test need to be higher for the PNoise analysis to work?
Thanks!

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by Frank Wiedmann on Jul 4^th, 2019, 1:09am

When you use the fullspectrum option, you must make sure that the value of the maxacfreq parameter of the pss analysis is large enough and covers the complete noise bandwidth of your circuit. You should increase its value until the result stabilizes (similar to the maxsidebands method). The fullspectrum option only calculates all white (frequency-independent) noise components, you still must set the maxsidebands parameter so that it covers all 1/f noise components. If your circuit has a large noise bandwidth and a low pss fundamental frequency, the required value of maxsidebands (without the fullspectrum option) can be very large.

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by vm511 on Jul 4^th, 2019, 6:28am

Hi Frank,
Thanks a lot. However, one thing I was concerned with was that in a simple RC track and hold, on increasing the max sidebands significantly for certain bandwidths, the integrated noise starts to become grossly overestimated (more than 2-3 times) from the theoretical value kT/C. I wasn't able to figure out why that is so. Moreover, how is number of harmonics to be decided correctly- as of now I just kept increasing it till the answer stabilised, similar to maxsidebands.
Thanks!

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by Frank Wiedmann on Jul 5^th, 2019, 4:52am

Without seeing your simulation setup, it's difficult to tell why the noise was overestimated. Perhaps the frequency steps in the pnoise analysis were too large or the maxacfreq value (or the number of harmonics) was not large enough. If you run a sampled (or jitter) pnoise analysis, you must always integrate from 0 to half of the pss fundamental frequency, otherwise you will get incorrect results.

The number of harmonics has a similar effect as the maxacfreq parameter: it controls the maximum timestep of the pss analysis. I recommend to use the maxacfreq parameter for this purpose because it defines the frequency range more clearly and does not create unnecessary results.

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by vm511 on Jul 5^th, 2019, 7:11am

Frank,
Thank you so much! I decreased the frequency step in the PNOISE analysis and the issue was solved. I wonder though, what is the reason of the noise being overestimated if the frequency step was too large. I had earlier done an automatic sweep, and the steps the simulator was taking were too large.

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by Frank Wiedmann on Jul 5^th, 2019, 6:50pm

If the frequency step is too large, the shape of the noise distribution over frequency will not be represented accurately. This will then result in an incorrect value of the integrated noise.

Title: Re: Maxsidebands parameter for SC amplifier (PNOISE)
Post by vm511 on Jul 7^th, 2019, 9:24am

Thank you for your help Frank!