Maker Pro
Maker Pro

MOSFET Characterization with SPICE

S

Spaceghost

Jan 1, 1970
0
Hi folks, I was using HSPICE today and did a .OP analysis, which gives a
bunch of small-signal parameters. Are there cases where this small-signal
parameters aren't going to make sense? I'm basically looking for parameters
such as un, Cox, Vt, and lambda, to use for rough hand analysis. I have a
modern short-channel device and the parameters .OP gives don't seem usable
for hand analysis.

The goal here is to take the parameters given by .OP, and use them for hand
analysis.

Here is an example from the web of what kind of thing I should get:

subckt
element 0:m1
model 0:nch
id 252.5184u
ibs 0.
ibd -24.7482f
vgs 1.4400
vds 2.4748
vbs 0.
vth 700.0000m
vdsat 740.0000m
beta 922.2733u
gam eff 500.0000m
gm 682.4823u
gds 2.4642u
gmb 220.2702u
cdtot 5.0000f
cgtot 23.9418f
cstot 18.5418f
cbtot 4.000e-16
cgs 18.5418f
cgd 5.0000f

I can fill in the classic sat eqn. ids=k*W/(2*L)*(Vgs-Vt)^2 and get a close
result for ids. Now in my case, with the short channel device, the
parameters are not giving anything close for ids to what HSPICE gave.

Above Vdsat=Vgs-Vt=1.44-0.7=0.740. My MOSFET results don't even make
sense...Vgs=0.9, Vth=0.6, but the tool is giving a Vdsat much lower than
0.3! I don't understand. :(

One book I have suggests extracting these from I-V curves, which I can do...
I just want to understand why the small signal parameters HSPICE is giving
don't work with the hand-analysis equations.

I'd be happy to explain this in more detail, but if you've done a .OP
analysis to get small signal MOSFET parameters I think this should be a good
start.

Thank you!

P.S. I took BETA to be un*Cox*W/L which seemed to work.
 
Top