The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl Simulators >> Circuit Simulators >> Using dcmatch analysis in Spectre https://designers-guide.org/forum/YaBB.pl?num=1124468279 Message started by vivkr on Aug 19th, 2005, 6:34am |
Title: Using dcmatch analysis in Spectre Post by vivkr on Aug 19th, 2005, 6:34am Hi, Can anyone suggest a good link which documents how to use the dcmatch analysis available in new versions of Spectre. This is supposed to allow mismatch analysis without recourse to Monte-Carlo analysis. I looked up the Spectre reference but the explanation there is rather vague and not helpful. I dont see how it can be used efficiently. The menu allows sweeping a design variable or other quantity which is helpful in a limited way but to analyse the effect of mismatch, one needs to allow random variation in all parameters of all transistors. The only such option that I find is one where you can specify the model which needs to be varied, and then specify which parameter needs to be varied and in what range. As this allows variation of only 1 parameter at a time and the user must know the realistic range for variation, it seems to be more of a sensitivity analysis than a real mismatch estimation. Also, there is a parameter called "threshold" whose definition is not entirely clear. Has anyone used this already? If so, then could you please offer some suggestions? Thanks Vivek |
Title: Re: Using dcmatch analysis in Spectre Post by sheldon on Aug 22nd, 2005, 4:58pm Vivek, There is an app note that includes a detailed discussion of dc mismatch and information on how to derive mismatch models. Some clarifications: 1) You can think of dc mismatch analysis as the equivalent of corner analysis for device mismatch. It introduces a 3-sigma mismatch between matched devices and allows designers to determine whether the design passes or fails with mismatch. 2) DC mismatch reports the effect of the device mismatch, for example, the offset voltage of the device. 3) In addition, it reports the "root cause" of the problem so that designers can improve the design. This is the purpose of the sensitivity outputs. The report also includes some insight into the source of the mismatch, for example, Vt, Beta, ... --> Information is helpful in understanding how to improve the design. 4) The "threshold" parameter is used to allow designers to ignore pairs that don't make a big contribrution to the output. This is helpful when analyzing large designs. 5) Designers do not vary the parameters, models including dc mismatch need to be provided by the foundry. In the app note there is a description of how to derive the required model parameters. 6) You can save the dc operating and use it with the small signal analysis. So you can calculate the effect of mismatch on common-mode rejection and power supply rejection. DC mismatch analysis is being actively enhanced. Recently gummel-poon bipolars and the EKV model were added to the analysis. DC mismatch is also very useful for verification and synthesis. Best Regards, Scheldon |
Title: Re: Using dcmatch analysis in Spectre Post by IvoryDesigner on Oct 21st, 2011, 10:50am Hi Sheldon, Do you please have the link of that app that talk about dcmatch? It will be really appreciated. Thanks Ivory |
Title: Re: Using dcmatch analysis in Spectre Post by analog_design on May 30th, 2012, 3:45am Hi Sheldon, Thank you very much for such nice description. I am also looking for document which can describe about how to use DC Match Analysis ? Can you do me a favor ? I do not have access to Sourcelink. If possible then, Can you post app not here. If it is not violating any regulations. Thank You, Girish |
Title: Re: Using dcmatch analysis in Spectre Post by wave on May 30th, 2012, 11:29am I'm curious how many FAB processes have models that support DC match. I used to work for the company that developed, it prior to the tool being absorbed by Cadence. If designers don't know about it, it's unlikely to be supported by models. ;) WAVE |
The Designer's Guide Community Forum » Powered by YaBB 2.2.2! YaBB © 2000-2008. All Rights Reserved. |