wildcards with "ic" statement

Dear greywanderer,

greywanderer said:

I have multiple instances of the same cell, named ucell0, ucell1, ucell2 etc, and want to set the initial conditions on the internal nodes of these instances.

I tried to set them using the ic statement as --> ic u0.u1.u2.ucell?.net1=0

But spectre does not like this and reports --> "Unexpected hierarchical name with wildcard" and then stops the simulation.

A couple of thoughts that might spur a thought if I may:

1. The wildcard syntax shown in your Forum post does not look like the wildcard syntax Cadence suggests in their RAK for the save statement. From that document on page 120 (Lab 13) it reads:

Scenario 1: Saves voltages of all nodes from level 2 to level 3 (where 1 is the top level) whose name starts with X and ends with 1 (for example, X1.in1). 

save X*.*1 depth=3 

This suggests to me that your save statement should be something like:

u0.u1.u2.ucell*.net1=0

I think the wildcard character of "?" only refers to a single character.

2. Are you using an extracted view to generate your netlist or a schematic view? The syntax you are using may not be the syntax used in the netlist. For example, the syntax may be 

us0/u1/u2/ucell0/net1

Shawn

greywanderer said:

I tried to set them using the ic statement as --> ic u0.u1.u2.ucell?.net1=0

But spectre does not like this and reports --> "Unexpected hierarchical name with wildcard" and then stops the simulation.

The syntax looks reasonable, and from an experiment that I made myself, it works fine; "?" is indeed the wildcard meta-character for a single character match.

Which Spectre version are you using? Wildcards are certainly supported with ic and nodeset (I answered a similar question here)

Andrew

A quick search on the Cadence Support website turned up support.cadence.com/.../ArticleAttachmentPortal

Frank Wiedmann said:

A quick search on the Cadence Support website turned up support.cadence.com/.../ArticleAttachmentPortal

Frank,

Well, that's embarrassing... I wrote that article! I should have done a quick search myself for the error message.

Andrew

Any chance that this "feature" will be fixed? It seems to have been known for many years.

Frank,

I've given the CCR I filed on this a push - hopefully that will help in getting it moving.

Andrew

The CCR is listed as fixed in Spectre 21.1 ISR20.

I recently tested the fix too - it does indeed work properly now (I have also updated the article).

Andrew