Multiplier("m-factor") not working in IC5.1 Analog Artist spectre direct interface

I am using UMC 0.13 design kit under IC5.1 and analog artist spectre
direct interface. I found that the multiplier (m) parameter is ignored
by the netlister and does not appear in the resulting netlist. I
searched through the internet and found that the flat netlister does
not support m-factor while hierarchical netlister does. However, the
"Environment Options" dialog window does not allow the choice between
flat and hierarchical netlist for spectre direct interface. I then
searched through the Cadence documentation in which I found that I can
set an variable
useMfactorToIterateInstances
to t to make m-factor work. I tried this but the result is the same, no
"m=..." in the netlist.

I was just simulating the same schematic that used to work with
IC5.0.33.
Please help Andrew.

 

Spectre always uses hierarchical netlisting.

For your m factor issue check the cells
component description format (CDF) and there
the simulation information for spectre netlisting.
The 'instParameters' are the ones which are taken by the
netlister.
I assume m is missing there.

Bernd

 

I opened the "Edit Component CDF" dialog box from CIW tools menu, and I
specified the library (umc13mmrf) and a cell whose m-factor was ignored
by the netlister. I suppose this is what you meant by checking the
cells CDF?

Under "Component Parameters", there is the parameter "m" with default
value of 1. There is another parameter called "simM" with default value
of
iPar("m")*iPar("fingers")

With analog artist in IC5.0.33 (the environment in which I originally
developed the schematic) the generated netlist contained text such as
m=(4)*(1)
when the multiplier is specified as 4, for example.

 

According to dirsimmodel.pdf (page 31 for version 5.0) using iPar() in
CDF is discuraged by Cadence. There are several Foundries in Far East
that seems to ignore the advice of the tool vendor. Even the name simM
seems to be a common usage among these foundries.

I haven't had the problem that you describe, but I have had other due
to this "unsupported" use of iPar().

I think the proper way must be to ask one of the Application Engineers
of the foundry to tell you how to solve your problem, and don't forget
to ask for the motivation to use iPar() in such an unnsuported way.

This was not much help, but at least you are not alone ...

 

You also should have a look under "Simulation Information" -> Edit
Choose Simulator == spectre.

Then have a look a the 'instParameters', there must be some kind of
m-factor in the list of parameters.

Bernd

 

I'm confused, because there is no flat netlister with spectre direct.
There was a choice of flat or hierarchical with socket based netlisters
such as spectreS.

If you're using spectre direct, the only reason m wouldn't be supported is if
it was omitted from the parameters netlisted in the CDF. You can check that.

However, Svenn's advice about contacting the foundry is the right thing to do
here.

There's also no reason why it should work in IC5033, but not in IC5141, if
the interface used and design kit used is the same.

Regards,

Andrew.