AES Conference Papers Forum Comments
http://www.aes.org/forum/?ID=1024
Copyright 2023 Audio Engineering Society Inc. All rights reserved.AES Conference Papers Forum: Comment on "Discrete-time Modeling of Continuous-time Pulse Width Modulator Loops"en-usMon, 25 Sep 2023 16:24:47 -0400Sat, 21 May 2022 21:47:09 -0400AES Webmasterhttp://cyber.law.harvard.edu/rss/rss.htmlAES Conference Papers Forum: Comment by Rajdeep Mukhopadhyay on "Discrete-time Modeling of Continuous-time Pulse Width Modulator Loops"
https://secure.aes.org/forum/pubs/conferences/?ID=1024#6517
Title: Discrete-time Modeling of Continuous-time Pulse Width Modulator Loops<br/>Paper 3-5; AES Conference: 27th International Conference: Efficient Audio Power Amplification; September 2005<br/>Comment by: Rajdeep Mukhopadhyay<br/><br/><p>Hi All, I have recently become a member of AES and the reason was this paper and some other by Lars Risbo. I am trying to develop a fast simulation set up to estimate/simulate class-D output noise. Right now, what I am finding is that AC analysis with average PWM model is optimistic, PSS/PNOISE is closer to silicon but still not quite there, and most frustatingly time to time suffers from convergence and/or accuracy issues. Add to that long simulation time of the latter. The main purpose of the PSS/PNOISE was to estimate the inband mixing due to high frequency components which is getting sampled by the PWM comparator. This paper seems to provide a method to model the entire feedback loop which does not use the average PWM model. But I stuggle to use the model outlined in this paper for my purposes. I have found a relevant paper from the same author which is easier to understand e.g. "A Versatile Discrete-Time Approach for Modeling Swich-Mode Controllers". That paper provides an equation for Ks(f) at 50% dutyc cycle and proposes a model (Figure 5). But Ks(f) contains z-parameters, so its not entirely a s-domain parameter which the paper claims.to be. I have got the following queries. </p>
<div>1) Can this model of Figure 5 (from "A Versatile Discrete-Time Approach for Modeling Swich-Mode Controllers"). be used for noise analysis using Cadence AC analysis? Or is Matlab better?</div>
<div> </div>
<div>2) If Matlab is better then how can we include components' (resistor, opamp noise) noise?</div>
<div> </div>
<div>3) Why the paper claim that Ks(f) is a s-domain function only but it clealy incudes z domain Hz(z)?</div><br/>AES Conference Papers ForumSat, 21 May 2022 21:47:09 -0400https://secure.aes.org/forum/pubs/conferences/?ID=1024#6517