Monte Carlo anlaysis to draw scatter plot of vth against leakeage current in SRAM

[Ravinder487]

Hi everyone, Now I am working on process variation SRAM cell design. I need to draw histograms and scatter plot for vth. But i can't have them in the prameter list of histogram windows.I figured it out the problem partially we should give vth as process varible in static block of .scs file. When i am doing so simulator is showing the error message mc1: Attempt to run Monte Carlo analysis with process variations, process variations were specified in static block. please help me.

 

[erikl]

Process variables belong to the parameters block; variation parameters belong to the statistics (process or mismatch) block.

 

[Ravinder487]

Thank you very much for u r help. could u please tell me typical values of standard deviation of vth and tox for both process and mismatch. I am using 90nm node.

 

[erikl]

could u please tell me typical values of standard deviation of vth and tox for both process and mismatch. I am using 90nm node.

These values are absolutely process- and fab-dependent, even the typical ones, and available only with an NDA. Ask your foundry!

For a foundry-independent example (180nm process) s. this page from a C@dence MC tutorial: View attachment Cadence_Monte_Carlo_simulation_tutorial_p36.pdf .

 

[Ravinder487]

Thank you for your help. When I am trying to include both process and mismatches in my model file(.scs) I saw only mismatch block I couldn't find any process block and also I am confused with the statements in model files(in subckt section): varvt = .004 // 1 sigma Vt mismatch variation in unit of v-um + mm_delvt = varvt * geo_fac * mm_pvt_mc and finally in model section vth0 = s1v_vth0_ne + mm_delvt. mm_pvt_mc is parameter defined in static block for mismatches only with a gaussian distribution of std =1/1. What does that mean? From above I could understand that I didn't have process block in my model files. So I added a new parameter pv_pvt_mc to included process variations. and I modified the vtho as vth0 = s1v_vth0_ne + mm_delvt+pv_pvt_mc. Could u please tell me whether my correction is wrong or right? and also one more thing i couldn't able to draw histograms for leakeage current for SRAM .

 

[erikl]

... mm_pvt_mc is parameter defined in static block for mismatches only with a gaussian distribution of std =1/1. What does that mean?

I suppose: 1σ (sigma).

From above I could understand that I didn't have process block in my model files. So I added a new parameter pv_pvt_mc to included process variations. and I modified the vtho as vth0 = s1v_vth0_ne + mm_delvt+pv_pvt_mc. Could u please tell me whether my correction is wrong or right?

Should be correct; just try it!

i couldn't ... draw histograms for leakage current for SRAM .

I attach the full C@dence MC tutorial. See p. 26! View attachment Cadence_Monte_Carlo_simulation_tutorial.pdf