The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> Is Partial MC mismatch analysis Possible ?!
https://designers-guide.org/forum/YaBB.pl?num=1213316814

Message started by Kalimero on Jun 12^th, 2008, 5:26pm

Title: Is Partial MC mismatch analysis Possible ?!
Post by Kalimero on Jun 12^th, 2008, 5:26pm

My understanding is that a Device L/W mismatch Monte Carlo analysis in spectre includes all the active devices in the circuit on which the simulation is run. My issue , that I was hoping for a bit of guidance on, is that my circuit includes a model of a SPAD (Single Photon Avalnche Diode) which I designed separately. It is included into the circuit as a symbol. I do not want the active devices within the SPAD symbol to be involved in the MC analysis I am running on the circuit. In other words, I need the Lenght and Width vary statements in my statistic block to leave the L and W of the active devices within the SPAD model alone and only act on the active devices outside the symbol. I do not have that option of running the simulation on the circuit without including the SPAD Model, as my results would be meaningless in such event. Is there way a to make this distinction in spectre... Am I right in assuming that the devices within my SPAD (Model) symbol will also be included in the MC analysis .. Thank you

Title: Re: Is Partial MC mismatch analysis Possible ?!
Post by ywguo on Jun 13^th, 2008, 6:00am

Hi Kalimero,

I think that you designed the SPAD device and the corresponding model. Right? If what I think is right, why do you worry about the Monte Carlo simulation? You define the W, L or any parameters in the SPAD device, just make them don't vary randomly. If what I think is wrong, would you like to present how do you model the SPAD?

Best regards,
Yawei

Title: Re: Is Partial MC mismatch analysis Possible ?!
Post by Kalimero on Jun 13^th, 2008, 1:42pm

hmmm .. That's interesting. Let me see if I am understanding you right. l and w are default device parameters. In order to perform a MC mismatch analysis I have to include a Statistics block and within it a vary statement in the device model file. The Parameter of the vary statement have to be valid parameters of the device.. i.e l and w. This will then vary statistically the' l' and ' w' of all the devices within the circuit under simulation. I was asking to see if there was a way for me to exclude certain devices in the circuit from the EFFECT of the 'vary' statements i have included in the model file... bit I think what you are suggesting is for me to edit the 'l' and 'w' parameters definition and define them in terms of another variable ... say 'l_vary'..somethign like this : ( l = l*iPar(l_vary) ) Butr to only do this for the devices I need included in the mismatch analysis. Then I can change the vary statement in the Model file to ( vary l_vary dist=gauss std= ......) instead of ( vary l dist ... ). Am i understanding you correctly? .. oh , and BTW - Thank you for your response.

Title: Re: Is Partial MC mismatch analysis Possible ?!
Post by Geoffrey_Coram on Jul 1^st, 2008, 5:39am

I haven't worked with the Statistics block myself, but: surely the 'l' and 'w' are specific to some model card, that is, the variations you define for MOS are not the same ones used for MIM capacitors, since the variations are due to different processing steps.

So, it would seem there is some association of the statistics block with the model card of the device being simulated, and if you don't have a statistics block associated with your SPAD, then there will be no MC variation applied to it.