Foundry PDKs specify that RF Spice model of transistor is valid up to certain frequency range, for example upto 20GHz for 180nm node. Does this imply that designer can design RF circuits upto that frequency limit with that technology PDK? Is there any criteria for frequency upto which RF circuits can be designed in specific technology node PDK.
I think model validity range and useful frequency range for design are two different things.
The model will usually be valid over a wide frequency range. Foundries often validate the model up to 40 GHz. This does not mean that you cannot use it beyond 40 GHz, you can, but it is only unverified. Most models do not abruptly collapse at say 41 GHz, so it will work.
If you design at 40 GHz, then even if model is valid, the design might not work very well if you are working at transit frequency fT of the transistor. Also, for nonlinear simulations, you will need to understand how the harmonics play out, and you will need 3-5 of them included. Model will require to correctly present harmonic impedances as well. In terms of criteria, you need to operate well below fT of the transistor (fT/3 to fT/2?) for you to get good amplification from the transistor.
Is it possible for design companies to adopt any other model type other than provided by foundry? How does that work, where they get measurements data for such models.
It is possible. Mid to large companies have DC and RF wafer probe capability to measure and model transistors. A design of experiments is required to cover all required device sizes and large sets of automated measurements are made. That data is used to create any other model needed.
Sometimes foundries are willing to provide their data.
The overview provided a good insight into various SPICE model types. However, does a design engineer have any flexibility in selecting the model type for simulations? Typically, they're constrained to using whatever is part of foundry PDK?
Yes that’s true. They are constrained by what is available to them from the foundry most times. Some organizations have teams that either enhance foundry PDKs to suit their specific needs, or help adopt completely alternative model types if foundry PDK doesn’t meet their needs.
Foundry PDKs specify that RF Spice model of transistor is valid up to certain frequency range, for example upto 20GHz for 180nm node. Does this imply that designer can design RF circuits upto that frequency limit with that technology PDK? Is there any criteria for frequency upto which RF circuits can be designed in specific technology node PDK.
I think model validity range and useful frequency range for design are two different things.
The model will usually be valid over a wide frequency range. Foundries often validate the model up to 40 GHz. This does not mean that you cannot use it beyond 40 GHz, you can, but it is only unverified. Most models do not abruptly collapse at say 41 GHz, so it will work.
If you design at 40 GHz, then even if model is valid, the design might not work very well if you are working at transit frequency fT of the transistor. Also, for nonlinear simulations, you will need to understand how the harmonics play out, and you will need 3-5 of them included. Model will require to correctly present harmonic impedances as well. In terms of criteria, you need to operate well below fT of the transistor (fT/3 to fT/2?) for you to get good amplification from the transistor.
Hope that helps.
Thanks very much for this explanation
Ok, got it. Thanks for explanation. And congrats for your articles. They are really useful for ic design community especially analog and RF domains.
Glad you like it!
Is it possible for design companies to adopt any other model type other than provided by foundry? How does that work, where they get measurements data for such models.
It is possible. Mid to large companies have DC and RF wafer probe capability to measure and model transistors. A design of experiments is required to cover all required device sizes and large sets of automated measurements are made. That data is used to create any other model needed.
Sometimes foundries are willing to provide their data.
The overview provided a good insight into various SPICE model types. However, does a design engineer have any flexibility in selecting the model type for simulations? Typically, they're constrained to using whatever is part of foundry PDK?
Yes that’s true. They are constrained by what is available to them from the foundry most times. Some organizations have teams that either enhance foundry PDKs to suit their specific needs, or help adopt completely alternative model types if foundry PDK doesn’t meet their needs.