printing VTH of a device

[Haritha Echampati]

Hello,

I point to the model files of an open source PDK. It is a BSIM level 72 model but has been manually changed to level 110  so that Xyce can be used to simulate with these model files.

I run a DC simulation and use the .print dc N(MM0:VTH) but the printed VTH value is ~ 250mV which is far higher than the expected ~ 75mV (which is known from previously published work on the PDK).

Can you please clarify if :

1. The statement to print VTH is right?

2. Is changing the model level from 72 to 100 contributing to error?

Thanks,

Haritha 

[Cliff Wiener]

Hi Haritha

What is the definition of threshold voltage in a FET?

If you are making a pacemaker, your threshold might be less than 100nA, because you are trying to keep currents down.

if you making a motor controller, speaker, etc, perhaps your threshold is 5mA, because less than that will give you insufficient current.

You may want to put your desired current into the FET with a cascode clamping the drain at your desired Vds  voltage, then measure Vgs. That would be your threshold.

I gotta believe that Vth is in the eyes of the modeler. I woudn't take the exact value too seriously. Do any of you FET modelers out there disagree with me?

Cliff
First order Xyce user

--
You received this message because you are subscribed to the Google Groups "xyce-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to xyce-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/xyce-users/553e9971-2344-4c4e-8848-a39624af4a7bn%40googlegroups.com.

[xyce-users]

Your syntax for printing the VTH output variable of the BSIM-CMG 110 model is correct.

The level 72 MOSFET in HSPICE is the BSIM-CMG, but I do not know what version of the model it refers to.  Xyce's level 110 MOSFET is the BSIM-CMG version 110.  If HSPICE's level 72 model is the BSIM-CMG 110, then no, switching the level number should not be the cause of this difference.  It is possible that different versions of the BSIM-CMG might give very different VTH values, but I think it unlikely.

To use other versions of BSIM-CMG in Xyce, you have to use a different level number (level 107 for BSIM-CMG 107, 108 for BSIM-CMG 108) --- unlike HSPICE, Xyce is unable at this time to support multiple versions of a model using the same level number by specifying a "VERSION=" model parameter.

When you say "which is known from previously published work on the PDK,"  have you actually attempted to run identical netlists in HSPICE and Xyce and observed a significant difference in the VTH value?

[Haritha Eachempatti]

Thanks, I haven't run the dc sweep on the PDK with hspice, which is the tool that was used by the original publishers.

here's a plot of Vth vs. Vgs

here's a plot of gm & Id vs. Vgs 

The Idsat seems to closely match the originally published value ~ 77uA.

Thanks

Haritha 

--
You received this message because you are subscribed to a topic in the Google Groups "xyce-users" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/xyce-users/FQSysLSpcQQ/unsubscribe.
To unsubscribe from this group and all its topics, send an email to xyce-users+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/xyce-users/f40a81c4-1e35-4b68-be29-753353ad4bfcn%40googlegroups.com.

[Haritha Eachempatti]

Hi 

To make it simpler to replicate I have attached my test bench and the model file.

Also attached the originally published sim results.

Thanks

Haritha