Issue with importing csv file to the ADEXL outputs

Dear greywanderer,

greywanderer said:

But when I make similar modifications to the rest of the occurrences in the csv file and import it to the output of the adexl, I get the following error,

Have you inspected the spectre netlist to determine the name of the transistor and its syntax (not the calculator expression)?

I usually use the syntax from the spectre netlist since that is a valid syntax.

Shawn

The spectre netlist has "Mxx\<5\>". And I see the same error as mentioned above when I use this format.

Dear greywander,

greywanderer said:

The spectre netlist has "Mxx\<5\>". And I see the same error as mentioned above when I use this format.

Have you tried using the CDBA syntax checker nmp? I tried it using your syntax and your syntax appears illegal. I might suggest you try the alternative syntax I show which is legal:

$ nmp isLegalName CDBA  "Mxx\<5\>"
illegal
$ nmp isLegalName CDBA  "Mxx|5|"  
legal

Shawn

PS. This tool is detailed at the Cadence on-line support URL:

support.cadence.com/.../ArticleAttachmentPortal

value(pv("Mxx|5|" "ids" ?result "tranOpTimed") 5e-07)

This does not work either.

value(pv("Mxx|5|" "ids" ?result "tranOpTimed") 5e-07)

This does not work either.

Doing all this messing around with nmp or looking up netlist names really should not be necessary. Sorry, Shawn, but you first suggested using the names in the netlist, and then suggested checking for CDBA legal names (CDBA names are what appears on the schematic, not in the netlist). So that's conflicting advice. I would try to use schematic names wherever you can, so that you're not at the mercy of how names get mapped in the netlister (and most of the time this is a reasonable strategy unless you are trying to access something that doesn't correspond to a schematic element - e.g. within a subckt model).

If it's a schematic name (and using outputs(?result 'tranOpTimed) in the CIW should show you the schematic names) then you just need to start the name with a / and use / as the hierarchy separator. If the name ended up in the netlist as Mxx\<5\> (which I'd expect) then because backslash is an escape character in SKILL, to truly get it you need a double-backslash.

So using something like this in your CSV should do it:

Name,Type,Output,EvalType,Plot,Save,Spec
mxx5_gm,expr,value(pv("/Mxx<5>" "gm" ?result 'tranOpTimed) 2e-07),point,t,,
mxx5_gmnet,expr,value(pv("Mxx\\<5\\>" "gm" ?result 'tranOpTimed) 2e-07),point,t,,
mxx4_gm,expr,value(pv("/Mxx<4>" "gm" ?result 'tranOpTimed) 2e-07),point,t,,
mxx4_gmnet,expr,value(pv("Mxx\\<4\\>" "gm" ?result 'tranOpTimed) 2e-07),point,t,,

The _gm are schematic names, and the _gmnet are netlist names (note the double-backslash). Having a single backslash in the string would be a problem because SKILL will try to read the < or > as escaped, which doesn't make sense - hence the parser error.

By the way, you can also see the netlist names you could use by typing: outputs(?map nil ?result 'tranOpTimed) in the CIW.

Andrew

As I had mentioned in the parent message, I saw the double backslashes when I exported to csv. But it did not work when I imported after changes to other outputs.

For whatever reason, both your suggestions are working now. I dont know what I might have done wrong.

Thanks for being so free with your knowledge. I did not know about getting netlist names with outputs() among other things.

Hi Andrew, Shawn,

Sometimes when using netlist names, there is a problem with the import of the csv file and it misses the double back slashes giving the error below. I will use the schematic names for now. Just thought you might be interested.

"value(pv("Mxx\<
                        ^^
PARSER WARNING at line 1 column 28 of file *string*
*WARNING* (Parser): character found after backslash is not meaningful - '\<'"

Looks like you are recommending to use the schematic names wherever possible.

greywanderer said:

Looks like you are recommending to use the schematic names wherever possible

I would generally go with the schematic names out of choice. However, the failure you mention with netlist names - this should work. So please post the relevant line from the CSV and the IC sub-version you're using, and I can check if it's something wrong with the CSV or there's a bug.

Andrew

This is the line from the CSV

library:cell:test,Ids0p5u_Mnb5,expr,"value(pv(""ux1.ux2.Mnb\\<5\\>"" ""ids"" ?result ""tranOpTimed"") 5e-07)",t,,

The IC sub-version,



Just want to clarify that this works sometimes and doesn't work sometimes.

The challenge appears to be if the expression is in quotes (and you have the "escaped" quotes embedded within). If it's like this:

library:cell:test,Ids0p5u_Mnb5,expr,value(pv("ux1.ux2.Mnb\\<5\\>" "ids" ?result "tranOpTimed") 5e-07),t,,

then it works with no problems.

If you have the surrounding quotes in place, there's an extra layer of parsing, so that means you need to double-up the backslashes to (yes) four!

library:cell:test,Ids0p5u_Mnb5,expr,"value(pv(""ux1.ux2.Mnb\\\\<5\\\\>"" ""ids"" ?result ""tranOpTimed"") 5e-07)",t,,

The joys of escape characters!

Andrew

I see. The double quotes were placed there when I exported to a csv file and I assumed they needed to be so.

Thanks much!