ELC do not recognizing spectre format model file well

[willvlsi]

Hi, all

I am doing std cell characterazation using ELC. And I use a model file from TSMC, it defines multiple sections for different processes and corner. At each section, it defines many parameters which has unique values to that process and/or corner, then it include the mos model which uses these parameter to describe its characters. The ELC could direct into the right section and include the mos model well, but it seems it does not recognize the parameter definition. It reports errors when I do "db_spice". Reported errors like:

Error found by spectre during hierarchy flattening. ERROR (SFE-1999): "/home/abc/works/elc/foo.ipdb/NCH.device/simulate/model" 4: Model `nch.1': parameter `wmin': Unknown parameter name `dxw' found in expression.

This model file works well with spectre in virtuoso, but not with ELC.

Anybody could help me with this? Thanks in advance!

ps. In the elcUG dorcument, under the SPICE Input directory, it said "An optional parameter library file, which defines the device parameters for different process corners" but no more detials following. Anyone know how to write this file and how to include it in ELC project?

 

[graphene]

did you check with the sample files provided by default by the foundry and compare the generated file by ELC??

 

[willvlsi]

Thanks for your advice.But the model file is one the inputs and it is provided by the foundry, ELC can not recognize it so I get nothing right now.
what you said is what I want to do

 

[graphene]

well.. I faced similiar issues... I had to develope SCL library and did cell characterization...

I used Liberate (Altos Liberate) again from Cadence instead of ELC for a similar problem.

I am not exactly understandiny your problem now... So please also see if your organization has license for Liberate...

- - - Updated - - -

also another suggestion would be that

1) take your lib file copy and remove all cell definitions except for the NOT gate for ex. but ensure you include every line before your first cell definiton available and feed that it... it shall steam out the nuisnace parameters if any...

2) you may be asked to include atleast one 2 input gate, one clocked cell (DFF) or so.... if asked include only those cells asked for,, that too the least potent cell... i mean a single 1 inp NAND gate, the simplest DFF or so...

if these dont work.. can you please be very eloborate in explaining your problem for us to understand it....


wishes..

 

[willvlsi]

emmm, I think it's not the problem of the simulation setting. My problem is that elc do not recognize the parameter definition command in my model file. it declares some parameters which are used to describe different process corners. So if elc could not recognize them, the simulation wont start up.

Hope this express my problem clearly.
And anyway, thanks for your willing to help me!