| Author |
Message |
cjupiter
Joined: 10 May 2004
Posts: 45
|
 28 Sep 2004 11:08 channel length modulation |
|
|
|
The channel length modulation coefficient lambda usually has a typical value between 0.001V-1 and 0.1V-1 for level 1 transistors.
Why in Bsim3 models, instead of lambda you have 'pclm' to represent channel length modulation, why does it have no units, and why can the value exceed 1.
Just curious about this because im following razavi's analog ic design book and he uses the level 1 values of lambda to calculate gain of opamps.
When i plug in my pclm values (which are 1.1 at the moment) into the same equations he uses, my gain is much much less. Im just wondering if im doing anything wrong.
Thanks
Last edited by cjupiter on 28 Sep 2004 14:59; edited 1 time in total |
|
| Back to top |
|
 |
pixel
Joined: 16 Sep 2004
Posts: 569
Helped: 59
|
| 28 Sep 2004 14:00 Re: channel length modulation |
|
|
|
maybe I did not understand the question, but:
gain is ~Rout
Rout~1/lambda*Id.
->When you have higher lambda then you should have lower gain, or I had made error?
|
|
| Back to top |
|
|
cjupiter
Joined: 10 May 2004
Posts: 45
|
| 28 Sep 2004 15:05 Re: channel length modulation |
|
|
|
I think i might just answer my own question here, anybody correct me if im wrong please. The equations I am using from Razavi's book are elementary equations. I am simulating BSIM3v3 transistors so the channel length modulation coefficient i use may not be suitable in the elementary equations.
Does anybody have any helpful hints when using BSIM3v3 models in OPAMP design?
|
|
| Back to top |
|
|
orbb
Joined: 30 Sep 2004
Posts: 15
|
| 30 Sep 2004 5:11 Re: channel length modulation |
|
|
|
first post 
I use bsim 3v3 too. You are right: The PCLM value does not correspond directly to lambda. If you have time, you can get bsim parameter definitions off http://www-device.eecs.berkeley.edu/~bsim3/get.html
The equation using PCLM is far too complex for first order hand calculations, which is probably what you want.
You could simulate the IDS .vs. VDS .vs. VGS curves for conditions similar to the ones you expect to place your MOS in. Extrapolate curves to the VDS axis and calculate an approximate value of lambda (for MOS, true lambda is this intercept value divided by the transistor length)
You can use this value as a rough guide for your calculations.
OR you can interest yourself in the EKV mos model...more complex then first order(not too much), but more accurate. Its what BSIM is based on
|
|
| Back to top |
|
|
