Continue to Site

Welcome to EDAboard.com

Welcome to our site! EDAboard.com is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

Monte Carlo analysis for mismatching

Status
Not open for further replies.

zizbear

Newbie level 4
Joined
May 25, 2010
Messages
7
Helped
0
Reputation
0
Reaction score
0
Trophy points
1,281
Location
SGP
Activity points
1,356
Hello,

I use the cadence Monte carlo tool to do the MC simulation under mismatching-only and process-only option. Then I get two histograms for them, and histogram1 for the mismatching-only is not Guass Distribution, histogram2 for the process-only is Guass distribution.

What does this mean? --- is it meaning that the circuit is not design good in matching, but good in process corners?

What is this telling me? I mean if I want to improve the circuit, I need to improve for mismatching,right? Is there any one can recommend me a book (chapter related) or some papers?

Thank you very much !
 

zizbear said:
I mean if I want to improve the circuit, I need to improve for mismatching,right?
Better improve the matching! SCNR ;-)

zizbear said:
Is there any one can recommend me a book (chapter related) or some papers?
Study **broken link removed**
 
Last edited by a moderator:

Thank you very much !

But the link you gave was deleted?

And I do find some useful paper by search here :)
 

zizbear said:
But the link you gave was deleted?
My first link still works from my site. Just tried it successfully. The secondary links may have been deleted. Sorry, you came too late! :-(
 

Mismatch may improve analog performances.
Process may improve both analog and yield performances.
Corners improve yield performances and fasten simulation time.

Corners are not realistic, but if you pass them, your circuits will have good yields.
SS and FF are not enough think to temperature, supplies, FS, SF ... they are all independant !

For Monte carlo, a minimum number of simulation is 100, it gives you spread limits within a 10% accuracy :D
 

okguy
Is there a formula which tells you that if you run 100 sims, you get a value within +-10%.
I just run enough so that the value is not changing much from run to run.
 

It was calculated in a thesis. I only remember the result.
As far as I remember with 250 simulations, you can reach 5%accuracy on your 1 sigma spread :idea:
 

As far as I remember with 250 simulations, you can reach 5%accuracy on your 1 sigma spread :idea:
6% ;-) . As with all statistical methods, the (in-)accuracy decreases with 1/√N , if N is the number of sim runs.
 

Status
Not open for further replies.

Part and Inventory Search

Welcome to EDABoard.com

Sponsor

Back
Top