Cadence IC5.1.41 libInit.il

Probably. I'd have to suck it and see, just as you can do. This API is obsolete so I've not really used it for a few years (the whole corners mechanism changed in IC61).

Regards,

Andrew.

Hi Andrew,

 I have another query wrt the probing currents in cadence.

To probe the current i click on the pin (device or Instance symbol) , which shows a circle around it. The problem that i am facing is that this works perfectly when my simulator is Spectre.

But when my simulator is hspice probing a device pin gives a warning :

 Element xpm1 in print, plot, probe or measure statements is invalid. Output is ignored.

And the currents are not plotted.

I am trying to probe the drain currents of PMOS in a simple current mirror circuit.

When i check the netlist created for hspice, there exists a definition for xpm1 and the probe statement for it is I1(xpm1). For the same simulation i am able to plot the transient node voltages as well current at the plus terminal of an ideal current source or a instance symbol pin.

Another difference that i noticed between the two input files (.scs and .ckt) is that:

In input.ckt (hspice) : the devices are defined as XPM or XNM instead of MPM or MNM

In input.scs (spectre) : the devices are defined as PM or NM.

The model files for both spectre and hspice contain the definition of mosfets as subckt.. 

Could you please suggest what could be the reason for this? 

I was going to split this off as a separate topic, because it's wandering all over the place... anyway, I see that you have created a new thread (so I won't answer this one).

Hi,

I am facing the exact same problem currently, and did not yet find an answer - too bad that Andrew did not answer this question, it would have been very helpful.

Is there a way that I can keep my customized path to the modelfiles ?

One workaround I found is, to load first the "libInit.il" in .cdsinit BEFORE setting my custom path. By that the Virtuoso environment does not reload the "libInit.il" when a library is accessed the first time via schematic/layout/ADE.

But this workaround creates some warnings when loading "libInit.il" via .cdsinit, and I am not sure why and if it is okay to call it during .cdsinit.

Any better way to do it ?

If you allow the libInit.il to load by "accessing" the library it would be much cleaner, for example: ddGetObj("library_name") in your .cdsinit before your own code for setting the model paths. I have employed this dummy access for this exact reason - I can't control the foundry code but I can force their code to load before my own settings.  A bit of a kludge but it works and does not require changes from the foundry to accommodate my requirements.

You can't disable the loading of a libInit.il. There are two approaches that could be used - one is to just use:

ddGetObj("theLibName")

in the .cdsinit - that will force the libInit.il to be loaded in the standard fashion because it's loaded from the firstAccessLib trigger of the l library.

The second (a bit more complicated) is to define your own firstAccessLib ddTrigger (using ddRegTrigger) to do something after each library is first accessed - it could then augment or set the model files. However, this is a bit more complicated. I'd probably just use the ddGetObj approach.

Regards,

Andrew.

Hi guys,

thanks for kick'in in so quick !

My apologies - I first did not see the complete thread for some reason, just the last comment from Andrew, now I see the quite lengthy discussion.

I actually do typically use ddGetObj(), but somehow in the last PDK install it did not work. I have to debug again, and see what the problem is.

But now I know the reason why I used ddGetObj() always !

Thnx!

I didn't notice the full thread either! This is an annoying feature that it collapses the threads sometimes so it didn't look as if there was any history. At least I appear to be consistent in my responses!

Andrew.