Hi all,

I have seen that many technology-kit also provide dedicated RF transistors. Apart from an optimized layout p-cell I don't see any difference to the regular transistors, but they gives somewhat different simulation results. Can somebody tell me what typically is the difference between dedicated RF transistors and the regular devices?

Regards

hi berti,
          i think in RF they will concern about less gate resistance so they would having more contacts on gate and good silicidation on poly...

Thanks,
Rajasekhar.

Berti wrote on Jun 28^th, 2009, 10:45pm:



Hi Berti,

I cannot think of much that would change the DC operating points except for the inclusion of source/drain contact resistance and perhaps the source/drain diffusion resistance being different due to a different p-cell layout. But these should not make a big difference. Maybe you have more terms for leakage and impact ionization in the RF model. Perhaps you can try probing the various terminal currents including the bulk current and gate current.

There is also some difference in the use of NQS (nonquasistatic) parameters in some RF models but I don't believe that this will cause a change in DC behavior. In fact, many RF models simply include a series resistance with the gate to capture some NQS effects.

Finally, it could simply be due to the fact that your RF device model was not updated as recently as the standard device model or was not centered correctly for the DC case.

Regards,

Vivek

Hi Berti,

Yes the RF device layer is typically just a CAD layer, although it can be used by the extraction tools (i.e. not to double count fixed pcell metalisation during RCX) or metal fill process (for metal fill exlusion purposes) or physical verification.

I agree the DC parameters should be very similar for both the RF and normal device, I've never seen an RF device which is not just a fixed layout. Having said that, its always good to look at the model card/equivalent circuit for the device. There are two common ways the RF devices are made:
1) normal FET model with added parasitics (like Rg, Rd, Rs, Rb, etc.)
2) complete extraction, normally BSIM. In this case you'll see that the RF devices have a different BSIM level to the standard devices. This could lead to discrepancies in DC behaviour, but one would hope these discrepancies would be small.

Cheers

aw