Designer’s Guide Community

qiushidaren

Community Member

Offline

I am a lovely
MOSFET!

Posts: 86
Chengdu

Instance length does not fit the given lmax-lmin
Jun 18^th, 2008, 2:00am

Hi all,

I need to use some PMOSFETs which have a length of 40u, but in the .scs file provided by TSMC, the length can not be larger than 20u, so I modify the lmax=20u to 50u in the file. But, when I simulate the circuit in Spectre it still says: �IXX.MXXX: Instance length or width does not fit the given lmax-lmin or wmax-wmin range for any model in the �pch� group�, what should I do to solve this problem? Thank you in advance.

Best regards,
Terry

Back to top

IP Logged

ACWWong

Community Fellow

Offline

Posts: 539
Oxford, UK

Re: Instance length does not fit the given lmax-lmin
Reply #1 - Jun 23^rd, 2008, 4:42pm

Hi qiushidaren,

Generally its not good practice to hack foundary model files, especially for MOS models based on BSIM fitting. The maximum limits for the MOS length is set so the model is only applicable for devices within the limits, with no guarantees as to model performance/match to silicon beyond these limits.
Anyway If you want a 40um length, why not use 2 series 20um length devices. There have been a few discussions on this topic in this forum, so do a quick search for more details...
e.g http://www.designers-guide.org/Forum/YaBB.pl?num=1162979112/30#30

cheers
aw

Back to top

IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1999
Massachusetts, USA

Re: Instance length does not fit the given lmax-lmin
Reply #2 - Jul 1^st, 2008, 5:35am

Adjusting lmax should have worked ... assuming you got the right model card. Are you sure you got the PMOSFET model card (not nmos) and the right "type" (standard-vt, high-vt, etc.)? Also, obviously, you should check the wmin/wmax values.

And ACWWong is right about no guarantee that the model card will accurately represent silicon behavior beyond the foundry's lmax.

Back to top

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged

Miconos New Member Offline Posts: 7	Re: Instance length does not fit the given lmax-lmin Reply #3 - Jul 1^st, 2008, 5:56am Another possibility you have is to use the length of the transistor as a parameter and to sweep it. I thinh that way you avoid both the echematic entry and the simulator limitations.
Back to top	VCO design IP Logged

qiushidaren

Community Member

Offline

I am a lovely
MOSFET!

Posts: 86
Chengdu

Re: Instance length does not fit the given lmax-lmin
Reply #4 - Jul 20^th, 2008, 7:58pm

Geoffrey_Coram wrote on Jul 1^st, 2008, 5:35am:

Adjusting lmax should have worked ... assuming you got the right model card. �Are you sure you got the PMOSFET model card (not nmos) and the right "type" (standard-vt, high-vt, etc.)? �Also, obviously, you should check the wmin/wmax values.

And ACWWong is right about no guarantee that the model card will accurately represent silicon behavior beyond the foundry's lmax.

Yes, I got the right model and right type, and I changed both the lmin/lmax and the wmin/wmax values, but it still said that, I don't know why, maybe there are some other values that are relative with the wmin/wmax and lmin/lmax parameter?

Terry

Back to top

IP Logged

qiushidaren Community Member Offline I am a lovely MOSFET! Posts: 86 Chengdu	Re: Instance length does not fit the given lmax-lmin Reply #5 - Jul 20^th, 2008, 7:59pm Miconos wrote on Jul 1^st, 2008, 5:56am: Another possibility you have is to use the length of the transistor as a parameter and to sweep it. I thinh that way you avoid both the echematic entry and the simulator limitations. Hi Miconos, Then what will happen?
Back to top	IP Logged

Geoffrey_Coram

Senior Fellow

Offline

Posts: 1999
Massachusetts, USA

Re: Instance length does not fit the given lmax-lmin
Reply #6 - Jul 21^st, 2008, 9:07am

qiushidaren wrote on Jul 20^th, 2008, 7:58pm:

Yes, I got the right model and right type, and I changed both the lmin/lmax and the wmin/wmax values, but it still said that, I don't know why, maybe there are some other values that are relative with the wmin/wmax and lmin/lmax parameter?

How do you know you got the right model and type? Why did you change wmin/wmax if it was length that was out of range? I would have just changed lmax to 1 meter. (If you set lmax=40u, it could be that "DL" makes the device have a slightly longer channel so it doesn't fit the bin.)

Can you cut out a single instance of the model from the library file and run it? How complicated is the library -- I've seen some with "sections" and "includes" that make it quite difficult to find what's actually being simulated (eg, maybe you changed lmax in the fast corner rather than nominal).

Back to top

If at first you do succeed, STOP, raise your standards, and stop wasting your time.

IP Logged