Including dspf file in Spectre and back-annotation of transistor's parameters and fingers

Hi Ronen,

The way you should include DSPF is via the DSPF fields on the Setup->Simulation Files form. You should not set that view to be a stopping view - so the netlist would include the schematic description. Doing it this way includes the file using the "dspf_include" construct in spectre, and the default behaviour is to use the port information from the subckt definition and map it to the pins of the DSPF - so this means you don't need to worry about port ordering (none of that messing around with CDF termOrder).

There's no back annotation performed here - it's simply using the devices (both designed and parasitics) fromm the DSPF. If there are parasitic resistances between the multiple fingers, APS will only combine them internally if the resistances are very small (i.e. below the threshold of shorting resistors anyway), so it's unlikely.

I assume you're asking about the "stitching" capability that spectre can do when you're talking about backannotation. In general that's not something we really promote and isn't on the forms in ADE (by default; it's hidden behind an environment issue).

If you include DSPF as a regular netlist, you have to take care of port ordering, plus there are some syntax challenges too - there's certain additional info in the DSPF which won't be honoured. So you shouldn't do it that way.

Regards,

Andrew.

Hi Andrew,

The dspf_include is a great solution, on one hand, it doesn't require messing with CDF and creating spectre view for the extracted block and on the other hand, it completely replaces the pre-layout circuit with the subckt in DSPF.

I have further question regarding dspf_include. We also use, in some cases, black-box flow where we do not extract one or more sub-circuits (digital sub-blocks). The documentation of blackbox option is not very clear, especially since we do not use QRC:

I understand that when you add the blackbox option to the dspf_include, Spectre uses the pin order in the black-box subckt inside the DSPF and the content from the original input.scs file. This is the exact behavior that we require.

However, our RC extractor does not output the  .subckt of the black-box. I know how to copy it from the cdl netlist, however, it is not clear where it should be. It would be very helpful if you can provide an example of such DSPF.

When I put the empty black-box in the DSPF after the .ends of the top extracted cell, Spectre output a message that the black-box subckt was replaced by the subckt in the DSPG file. When I put it inside the top-level subckt (before the .ends) it doesn't say anything about the black-box.

And lastly, how do I had multiple black box? I used blackbox="bb1 bb2". Is it correct?

Thanks,

Ronen

Hi Andrew,

The dspf_include is a great solution, on one hand, it doesn't require messing with CDF and creating spectre view for the extracted block and on the other hand, it completely replaces the pre-layout circuit with the subckt in DSPF.

I have further question regarding dspf_include. We also use, in some cases, black-box flow where we do not extract one or more sub-circuits (digital sub-blocks). The documentation of blackbox option is not very clear, especially since we do not use QRC:

I understand that when you add the blackbox option to the dspf_include, Spectre uses the pin order in the black-box subckt inside the DSPF and the content from the original input.scs file. This is the exact behavior that we require.

However, our RC extractor does not output the  .subckt of the black-box. I know how to copy it from the cdl netlist, however, it is not clear where it should be. It would be very helpful if you can provide an example of such DSPF.

When I put the empty black-box in the DSPF after the .ends of the top extracted cell, Spectre output a message that the black-box subckt was replaced by the subckt in the DSPG file. When I put it inside the top-level subckt (before the .ends) it doesn't say anything about the black-box.

And lastly, how do I had multiple black box? I used blackbox="bb1 bb2". Is it correct?

Thanks,

Ronen

Hi Ronen,

I've never used the blackbox option with dspf_include so I'd have to construct an example which uses this (and uses two blocks to check that too). I'm on vacation this week so won't have the time to experiment - so perhaps the best bet would be to contact customer support and follow it up that way?

If I have a moment when I return next week (if  you've not got an answer by then) I'll try it - but didn't want to keep you waiting.

Regards,

Andrew.