The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl Design >> Analog Design >> Distortion analysis for highly linear circuits https://designers-guide.org/forum/YaBB.pl?num=1119968143 Message started by Vivek Sharma on Jun 28th, 2005, 7:15am |
Title: Distortion analysis for highly linear circuits Post by Vivek Sharma on Jun 28th, 2005, 7:15am Hi, I am trying to simulate the linearity of an opamp in a 20-bit delta-sigma application. I would like to simulate it in Spectre. I am not sure how I can manage this. In order to see the harmonics, I need to achieve a noise floor of about -120 dBFS and this would require simulating the circuit for about 2^20 samples, which is out of question. Moreover, I read in the docs on this site that PSS-based analyses dont work for delta-sigma ADCs as they do not have periodic states, or atleast the period is usually too large. Is it possible to simulate the nonlinear distortion in some way? Thanks Vivek |
Title: Re: Distortion analysis for highly linear circuits Post by sheldon on Jun 28th, 2005, 8:34am Vivek, Should be no problem simulating the op-amp and getting the noise floor you are targeting. However, simulating the op amp by itself really is not meaningful and you will probably need to simulate the entire Delta-Sigma ADC to make the measurements you need. It should be possible to look at the distortion without running 2^20 points. Somewhere between 2048 and 16384 should be enough, look at the FFT plots in the following document, they use 65536 point FFT and have plenty of dynamic range. http://www.imse.cnm.es/esd-msd/WORKSHOPS/MIXMODEST/PRESENTATIONS/medeiro2.pdf Some other things to watch are: 1) You need to use windowing function when performing the FFT, for example, the hanning window. Unlike Pipeline ADC, windowing is required when performing an FFT on a Delta-Sigma ADC 2) You also need to control the simulation accuracy, try using Spectre's relref=pointlocal and method=gear2only BTW, if you do this simulation at transistor level it will take a while, filter the outputs to prevent filling up the disk. Also consider strobing the outputs, again to minimize the data size. Best Regards, Sheldon |
Title: Re: Distortion analysis for highly linear circuits Post by Vivek Sharma on Jun 29th, 2005, 6:50am Hi Sheldon, Thanks a lot. I am able to observe the harmonics very clearly now. I previously used coherent sampling and a rectangular window. Usually, this approach works quite well for me, but in this case, the window transfer function was getting in the way. I am sure that I will be able to test the THD in the overall system with this scheme. Regards Vivek |
The Designer's Guide Community Forum » Powered by YaBB 2.2.2! YaBB © 2000-2008. All Rights Reserved. |