Help with gmro vs. VDS curve

[sjamil02]

I did simulation for a diode-connected NMOS to get the curve ft vs. gm/Id, Id/(W/L) vs. gm/Id and gm/Id vs. Vov. The results look ok.

However I get a strange curve when I did simulation for gmro vs. VDS curve.
Here's what I did:
I ran a simple NMOS dc simulation (fixed Vg=Vt+200mV and sweep VDS 0V-1.8V). Simulation was done in PSPICE.

To get gmro vs. VDS curve:
(1) Slope 1/ro is the derivative of Id vs. VDS curve.
(2) gm0=Id/(vgs-vt), in my simulation vgs-vt=200mV
gm=gmo*(1+lamda*VDS) where lamda=1/ro*Id

Please have a look at the graph and the data in the attachment. Anyone experienced can help me to shed some light?

cheers
sj

 

[holddreams]

In the attachment, page 7, answer d, there is gmro vs vds curve, which is just like the waveform you pasted.

With spectre simulator, I can measure ro=1/gds, gm respectively.
I do not know wheter pspice can do it or not.
You can also refer ee214 lecture which give some spice netlist for simulation gm/id related issues.

 

[sjamil02]

Holddreams,

Thanks for sharing the data. Ok, my simulation looks correct and ee240 notes convinced that.

In Pspice I have to export the simulation data into spreadsheet and do the post processing to generate the chart. For some reason, method used in HSpice or Cadence did not work in PSpice.

Another question I have,

To determine Ft = gm/cgs+cgd, I got cgs and cgd value from dc operating analysis (from the dc output file, I picked cgs=cgsov=0.5f and cgd=cgdov=0.5f for L=0.18um and W=1um). Is this method correct?

cheers
sj

 

[holddreams]

Hi,

In ee214 lecture, ft=1/(2*pi)*gm/Cgg, where Cgg=Cgs+Cgd+Cgb.
You can search ee214 lecture to review it.

Best Regards,
Xianguang

Holddreams,

Thanks for sharing the data. Ok, my simulation looks correct and ee240 notes convinced that.

In Pspice I have to export the simulation data into spreadsheet and do the post processing to generate the chart. For some reason, method used in HSpice or Cadence did not work in PSpice.

Another question I have,

To determine Ft = gm/cgs+cgd, I got cgs and cgd value from dc operating analysis (from the dc output file, I picked cgs=cgsov=0.5f and cgd=cgdov=0.5f for L=0.18um and W=1um). Is this method correct?

cheers
sj

 

[Anusha Chavva]

Hi sjamil,

I need some help in simulating ft vs. gm/Id, Id/(W/L) vs. gm/Id and gm/Id vs. Vov, gmro vs. VDS curve. Could you kindly attach the netlist of these simulations.

Thanks,
Anu.