Typical threshold voltage variation- Monte Carlo

[adnan]

statistical, threshold voltage variations

My simulator doesn't support Monte Carlo analysis as Cadence. However it allows sweeping parameter values that are chosen based on statistical variations. For example, it can vary the threshold voltage VTH using random values chosen by probability distribution. The user provides the central value of VTH and its relative variation (as percentage of the central value).
My question is what is the typical relative variation of VTH for, let's say a good layout design? The technology I am using is 0.35 um.
Thanks

 

[kishore2k4]

threshold voltage sigma variation

My understanding is that Vth not only depends on the process corners but also on the MOSFET aspect ratio(neglecting body effect). To test out statistical variations, I think you can use the gaussian distribution with the the mean set to your Vth0 and a +Sigma,-Sigma set to their 0.34% of the mean.

NOTE: AFAIK, a good layout design can only closely match transistors(Vth,thereby) but not iron out statistical variations across the chip and wafers.

 

[adnan]

threshold voltage rel_variation

Thank you kishore2k4. Can you please elaborate more as I am not familiar with statistics? my tool command syntax is as follow:
.param VTH0=gauss(nominal_val, rel_variation, sigma)
if VTH0 nominal is 500 mV, and based on your reply, sigma=1.7mV (0.34% of 500 mV). what about rel_variation (relative variation)?