Parameters Sweep in ADE-XL results in "netl err"

This is almost certainly a specific initialisation problem with the PDK you're using. It might be that the formInitProc is not being called and it needs to be for the callback to work. 

I would suggest you contact customer support so that we can investigate further to confirm that is the case - I believe there is a way around this (I don't know the details off the top of my head - I'm checking though).

Regards,

Andrew

There is indeed a way of doing this (if it's the formInitProc issue that I think it is) but for various reasons the mechanism to fix it isn't documented (probably we want to be aware of anyone who needs it in case there are issues). So please contact customer support as I mentioned, and suggest to the application engineer that they consult internal article 11831138 for more details. If that's not clear, they can contact me internally.

Thanks,

Andrew.

Thank you, Andrew. I will follow-up with Customer Support as you suggested.

Hi Andrew,

Perhaps you are already in the loop on this, but according to Customer Support (case# 46233633), ADE-XL does not support pcell instance parameter sweeps.

Did you have a work-around to allow ADE-XL to the pcell instance parameter sweep (that can be shared)?

Or, is there a way to "fix" the pcell code to allow sweeps of pcell instance parameters?

No, I wasn't in the loop. I just had a look at the case, but nowhere does it say you're using a pcell other than in the original description from you. Are you sure it's a pcell? The error to me looks as if it's just the formInitProc that isn't being called - not that the component is a PCell. A PCell is a specific kind of device where there is a schematic view (or symbol view) where the connectivity changes based on a parameter; in this case I think you have a normal component, probably - unless you know for certain otherwise.

Rather than going back and forth to support any more, please can you try putting this in your .cdsinit :

defun(nlBeforeInstanceCallbackRun (instid cdfgData cdfgForm)
  when(cdfgData->formInitProc
    funcall(concat(cdfgData->formInitProc) cdfgData)
  )
)

Then restart virtuoso and try doing the parameter sweep.

Regards,

Andrew.

Thank you! Much appreciated. ADE-XL now completes the parameter sweep correctly.

To answer your question: Yes, the resistor is a schematic pcell (I guess with callbacks, since pcell instance parameters change based on other parameters).

Would you be able to explain your answer for my knowledge?

Thank you! Much appreciated. ADE-XL now completes the parameter sweep correctly.

To answer your question: Yes, the resistor is a schematic pcell (I guess with callbacks, since pcell instance parameters change based on other parameters).

Would you be able to explain your answer for my knowledge?

The CCR that was filed for you was related to sweeping parameters on pcells which affect the structure of the underlying schematic pcell. If you have a schematic pcell, then each pcell parameter produces a different subckt in the netlist - this is for parameters that affect the connectivity - so for a resistor, it might be the number of segments in a series connected resistor. Other parameters might just get passed into the subckt and not affect the connectivity - and hence may work fine. 

Because ADE creates the netlist once and then changes the parameter values for each point in the sweep, anything that requires a new subckt to be generated won't work - that's what the CCR is related to. Presumably you're not trying to sweep any parameter that requires it to netlist a different subckt - if you were, it wouldn't work.

In your case it was failing because the CDF callback gets called for each point in the sweep, and it looks as if the CDF callback needs some additional data to be initialised in the formInitProc - this is something that normally gets called when you bring up the edit properties form on the component, and then the callbacks get called as each parameter is changed. The callback mechanism in ADE XL doesn't call the formInitProc, and so the code I provided is a special trigger that is called before the callback, and I've defined it so that it checks to see if there's a formInitProc and if so, it invokes it - hence providing the initialisation that was otherwise missing.

Hope that helps.

Regards,

Andrew.

Yes, definitely helps.Thanks!

Sorry about the misunderstanding with the CCR - my issue was simpler. However, it would be nice to keep this CCR as it's possible I (or my fellow engineers) may sweep a parameter, such as "segments", that may affect the subcircuit netlist structure.