The Designer's Guide Community Forum

The Designer's Guide Community Forum
https://designers-guide.org/forum/YaBB.pl
Simulators >> Circuit Simulators >> How to plot I-V C-V curve of MOS using Spectre
https://designers-guide.org/forum/YaBB.pl?num=1175070667

Message started by ywguo on Mar 28^th, 2007, 1:31am

Title: How to plot I-V C-V curve of MOS using Spectre
Post by ywguo on Mar 28^th, 2007, 1:31am

Hi, Guy,

I have a fool question, how to plot I-V C-V curve of MOS transistor using Spectre? I used to run simulation using HSPICE. :(

BTW, I think a good manual is important for engineers. Would you please recommend any manual or books about Spectre?

Thanks
Yawei

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by Geoffrey_Coram on Mar 28^th, 2007, 5:13am

Put in a "save" statement to get the capacitances:
save m1:all

You can also run an ac analysis with a swept dc bias
ac1 ac freq=0.1591549 dev=vd start=0 stop=1.8
which sets the frequency to 1/(2 pi) such that imaginary currents are exactly equal to the capacitances. (You do have to look out for internal resistors, though, which will distort the results.)

For *using* Spectre, the built-in help (spectre -help) is good enough for me; Ken did write the Designer's Guide to Spice and Spectre if you want to know more about the internal workings.

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by John O Donovan on Mar 28^th, 2007, 9:51am

Hi,

Off topic, but Spectre has a set of builtin mathematical and physical constants that can be used in expressions. Run "spectre -h constants" to get the list.

 (1) % spectre -h constants
M_E             2.7182818284590452354           exp(1) = e
M_LOG2E         1.4426950408889634074           log2(e)
M_LOG10E        0.43429448190325182765          log10(e)
M_LN2           0.69314718055994530942          ln(2)
M_LN10          2.30258509299404568402          ln(10)
M_PI            3.14159265358979323846          pi
...

So you could use "freq=1/(2*M_PI)" in the ac analysis.

Regards,
John

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by imd1 on Mar 29^th, 2007, 1:41am

Do you know if it's possible to use similar built-in constants in ELDO or HSPICE ?
I find this rather useful, and I've been looking for a way to automatically obtain the thermal voltage during a simulation.

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by Croaker on Mar 30^th, 2007, 9:07am

Geoffrey_Coram wrote on Mar 28^th, 2007, 5:13am:

Yes, this is a very good tip, assuming the resistance is low. Also, you can use the CAPTAB option to get the node caps at an operating point. Is there any way to do a DC sweep and plot the CAPTAB results for a node, similar to the 'save m1:all' tip?

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by ywguo on Apr 4^th, 2007, 1:36am

Hi, Geoffrey,

I use save m1:all. However, may I save them in analog design environment? Furthermore, how to plot them in wavescan?

Thanks
yawei

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by Geoffrey_Coram on Apr 4^th, 2007, 7:39am

I don't use the analog design environment, nor wavescan, actually. But I know other threads on this forum have covered plotting "gm" in wavescan, and I would expect the capacitances to work the same way. You might have to look out for problems like "m1" being a subcircuit, so that the actual device that has the operating-point info is "m1.m"

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by Croaker on Apr 4^th, 2007, 9:53am

Yes, you can pull up the results browser and put the results into the calculator to plot the dc op stuff vs. whatever.

Title: Re: How to plot I-V C-V curve of MOS using Spectre
Post by Andrew Beckett on Apr 11^th, 2007, 7:53am

To answer the question about saving them in ADE, you need to create a text file and put the save statement(s) in there, and then include this file in the simulation. The simplest way of doing this is to use Setup->Model Libraries and reference the file there (Model Libraries are effectively any spectre syntax file - they don't necessarily have to be models).

Regards,

Andrew.