For HSPICE analog users, tell me what is wrong !

[superluminal]

uo=350 phi=0.9 vto=0.7 mj=0.45

Well, it was my first time trying pole zero analysis using HSPICE. then I select a very simple "academic " CMOS CS amplifier. Even I used LEVEL 1 model to make things no easier. My netlist was :

PZ trail : CS
.OPTIONS LIST NODE POST
.OP
.AC DEC 100 1K 100G
.PZ V(VOUT) VS
.PRINT AC
VDD VDD 0 DC=3V
VS VS 0 DC=0.867V AC=0.707mV
M1 VOUT VS 0 0 NMOD L=2u W=30.64u
M2 VOUT VG2 VDD VDD PMOD L=4u W=13.94u
M3 VG2 VG2 VDD VDD PMOD L=4u W=13.94u
IREF VG2 0 DC=37.65uA
.MODEL NMOD NMOS
+LEVEL=1 VTO=0.7 GAMMA=0 PHI=0.9 NSUB=9.0e+14 LD=0.08e-6 UO=350 LAMBDA=0.14 TOX=9e-9
+PB=0.9 CJ=0 CJSW=0 MJ=0.45 MJSW=0.2 CGDO=0.4e-9 CGSO=0.4e-9 JS=1e-8
.MODEL PMOD PMOS
+LEVEL=1 VTO=-0.8 GAMMA=0 PHI=0.8 NSUB=5.0e+14 LD=0.09e-6 UO=100 LAMBDA=0.07 TOX=9e-9
+PB=0.9 CJ=0 CJSW=0 MJ=0.5 MJSW=0.3 CGDO=0.3e-9 CGSO=0.3e-9 JS=0.5e-8
.ALTER
R1 VS VG1 180K
M1 VOUT VG1 0 0 NMOD L=2u W=30.64u
.END

For the first case ,R1=0, I should get only a pole at f=65.18 M & a zero at f=5.82 G (by hand).

But HSPICE gave me :
poles (rad/sec) poles ( hertz)
**********************************************************************
real imag real imag
-234.2986x 0. -37.2898x 0.
-647.3471x 0. -103.0285x 0.

zeros (rad/sec) zeros ( hertz)
**********************************************************************
real imag real imag
-386.5422x 0. -61.5201x 0.
35.5751g 0. 5.6620g 0.

& for 2nd case , I should have 2 poles : 0.89 M & 65.18 M & a zero at 5.82 G. But I got

poles (rad/sec) poles ( hertz)
**********************************************************************
real imag real imag
-5.3817x 0. -856.5288k 0.
-369.3320x 0. -58.7810x 0.
-2.6527g 0. -422.1972x 0.
zeros (rad/sec) zeros ( hertz)
**********************************************************************
real imag real imag
-386.5422x 0. -61.5201x 0.
35.5751g 0. 5.6620g 0.

Well, if you run this & see the frequency response (mag & phase) you will see only the poles and zeros you excepct by hand caculations.

If anyone can help, I will be so glad.

 

[adsl]

You can try to replace the MOSFETs with equivalent Rs and Cs and run the simulation again to see what happens. I can say now is that the unexpected answers might result from either the model or the way you calculated.

 

[nathan]

A couple of changes you may want to apply:

1) set AC=1 (this is in phasor domain therefore you can choose the easiest amplitude you want)

2) set R1 in the first part of the netlist as well then, change the value. I'm not sure what happens if you change the connectivity on the fly.

nathan

 

[superluminal]

replacing the mos with its equivelant R & C is not good ! why I use a simulator then ? hand calculations already do this job & what is missing is to see how the real design differs from what I excepct.

Actually, if you copy this netlist in your HSPICE & run , you will see the response you know from your caculations & the phase response shows no extra poles or zeros that that I calculate by hand.

Also I set the AC= 0.707 ( note that HSPICE takes rms, not peak) & the gain was as I expect. The problem here that I got my response I want either by hand or simulation, but the output file of simulation did not reflect that.