Sweep the real inductor value in virtuoso

Dear Detailer,

Detailer said:

I have a question regarding real inductors. In virtuoso , can we sweep the real inductor value by script or by sweep itself or are we forced to vary the parameters of the inductor ?

It appears the component for which you show the query GUI is an inductor from some PDK. As such, the parameter values that have "greyed out" backgrounds are likely computed and set by callback routines of your PDK based on the values that are editable (i.e., parameter values with white background). Therefore, as long as your PDK allows for varying the inductor value and adjusts its other parameters accordingly (and has error checking to prevent a non-realizable component), setting the desired inductor value with a design variable appears possible to me.

As an example, in most PDK MOS device models, the number of device fingers of an MOS device can be set to a design variable and used to sweep circuit performance over a range of number of device fingers.

Hopefully, I understood your question and was able to provide some insight Detailer!

Shawn

It appears that the inductor has been characterised by the foundry using the EMX Continuum "OIF" mechanism which allows a inductor to be specified via electrical parameters and then a mini-optimiser used to find a good fit based on precharacterized data performed by the foundry using EMX simulations. It does not directly choose the physical parameters from a given "desired inductance" - you have to press the "Find Optimal Inductor" once you've set up your goals.

This is not going to be suited to either having design variables in the inductance fields nor using the ADE parameterization capability (available in Explorer, Assembler and ADE XL/GXL) which is useful in the case where the component does not directly support variables in CDF parameter fields. That's because there will be nothing to explicitly trigger the "Find Optimal Inductor" during the sweep.

I'm not sure there's any easy way to directly solve that right now - perhaps you should contact customer support to see if we can find a way of supporting this through such Continuum/OIF inductors.

Regards,

Andrew

It appears that the inductor has been characterised by the foundry using the EMX Continuum "OIF" mechanism which allows a inductor to be specified via electrical parameters and then a mini-optimiser used to find a good fit based on precharacterized data performed by the foundry using EMX simulations. It does not directly choose the physical parameters from a given "desired inductance" - you have to press the "Find Optimal Inductor" once you've set up your goals.

This is not going to be suited to either having design variables in the inductance fields nor using the ADE parameterization capability (available in Explorer, Assembler and ADE XL/GXL) which is useful in the case where the component does not directly support variables in CDF parameter fields. That's because there will be nothing to explicitly trigger the "Find Optimal Inductor" during the sweep.

I'm not sure there's any easy way to directly solve that right now - perhaps you should contact customer support to see if we can find a way of supporting this through such Continuum/OIF inductors.

Regards,

Andrew

Hi,

First of all thanks for such attentive and fast answers. So it seems I can't sweep real inductors. I will either contact the pdk provider to see if there is anything I can do. Either way, I was wondering if there was a free real inductor model that by chance has a high correlation with a 130 nm pdk inductor?

For the time being, since I could vary the real inductor parameters, I was able to get an R/L plot where R represents the parasitic resistance and L the obtained resistance. With this, ignoring the capacitance between the metal layers, I could do what I wanted initially

Once more, thanks for all the support given

Alex

Detailer said:

So it seems I can't sweep real inductors

Alex,

Being pedantic, it's not that you can't sweep "real" inductors, it's just that you can't sweep the inductance for this specific PDK inductor. Sweeping physical parameters should be feasible, as you've found. I'm not sure how useful a "free real inductor model" would be given that this is tied to a physical implementation for the specific 130nm technology you're using. Correlation to "a 130 nm pdk inductor" is not likely to be of any use unless it's to the same foundry and metal stack you're using, with the same inductor winding layers. If you're just interested in finding an inductance that matches you could use either ind from analogLib (maybe a bit too ideal) or indq from analogLib which at least allows you to specify the Q and gives you a more realistic non-ideal inductor - then when you've found the likely value you want, replace with the PDK device.

Regards,

Andrew