The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Design >> Analog Design >> Monte carlo miamatch calculation
https://designers-guide.org/forum/YaBB.pl?num=1450399042

Message started by robotist on Dec 17^th, 2015, 4:37pm

Title: Monte carlo miamatch calculation
Post by robotist on Dec 17^th, 2015, 4:37pm

Hi my name is kim.
I have a question about mismatch simulation.
I just want to set resistor size by calculating standard deviation.

The attatched file describes my question and hand calculation
I use propagation error formular with assumption that there is no correlation.

Surely, there are no parameter related correlation in model file.

The result in simulation is quiet different with my hand calculation.
Maybe I missed something.

With same sheet resistor, how can I calculate resistor divider mismatch correctly?

Thank you for reading.

Title: Re: Monte carlo miamatch calculation
Post by ULPAnalog on Dec 17^th, 2015, 7:01pm

Did you check your PDK documentation for what the Pelgrom coefficient (A_R) is?

Title: Re: Monte carlo miamatch calculation
Post by robotist on Dec 17^th, 2015, 8:13pm

R0 and R0+R0 are simulation results.(Monte carlo simulation)
So, R0 and R0+R0 are determined by Ar, W and L.
What I want to know is R0/(R0+R0) standard deviation.
Thanks for reply.

Title: Re: Monte carlo miamatch calculation
Post by raja.cedt on Dec 18^th, 2015, 2:55am

Problem with your hand-calculation.
1. How do you know Sigma of Ro from PDK or a simple resister calculation? 2. assume R1 and r2 in the potential divide and differentiate individually and finally substitute same value in the place of R1&R2.

Best Regards,
Raj.

Title: Re: Monte carlo miamatch calculation
Post by ULPAnalog on Dec 18^th, 2015, 1:31pm

I overlooked that aspect in your hand calculation (One pointed out by Raja). You should have used 2*(20)² instead of 28 in your calculation, but it does not change the outcome significantly as the variance is dominated by the other term (The term that is erroneous is one half of the other).

Title: Re: Monte carlo miamatch calculation
Post by robotist on Dec 18^th, 2015, 5:17pm

Thank you for reply.
To raja
1. Sigma of R0 is monte carlo simulation result. Surely, it does not make sense that R0 has mismatch itself. But the sheet resistance of R0 varies with gaussian in simulation.
2. I am sorry but I can not understand your comment.

To Analog
Would you explain why I have to use 2×(20)2 in detail?

Thank you again.

Title: Re: Monte carlo miamatch calculation
Post by ULPAnalog on Dec 18^th, 2015, 6:16pm

Hello

Attached is the response to your query. It is exactly same as what Raja commented. Two resistors of equal value have variations which are iid and should be treated as such. That iss where the stddev of the resistor with twice the value, you expressed in your original post becomes irrelevant.

Title: Re: Monte carlo miamatch calculation
Post by robotist on Dec 21^st, 2015, 6:43am

Thank you very much ULPAnalog.