Designer’s Guide Community

jb_savage

New Member

Offline

Posts: 4

Can Spectre/SpectreRF be used at 60GHz?
Jun 11^th, 2011, 2:20am

Can someone comment on using Spectre/SpectreRF at 60GHz for transceiver design (LNA, VCO, PA, etc...) in mmwave? Is it out of the range for Spectre? Have high frequency models (TSMC N65) and Momentum created s-params for inductors. Agilent GoldenGate seems to work OK as does ADS. Using Spectre APS seems to get different results. Could this be because Spectre does work at 60GHz or maybe the APS model interpreting which I understand is simplified is too aggressive. I really don't like a simulator that is messing with the model parameters and accuracy. Or, maybe I am expecting too much from Spectre at 60GHz.

Back to top

IP Logged

RFICDUDE

Community Fellow

Offline

Posts: 323

Re: Can Spectre/SpectreRF be used at 60GHz?
Reply #1 - Jun 11^th, 2011, 7:19am

What analysis engine(s) are you using within Spectre (HB or shooting method)?

If you are using shooting method then you really need to verify the nport model that Spectre creates for your s-parameter blocks (perhaps its just as important for HB). The nport model can extrapolate to bad low frequency impedances if the data is not defined down to DC. This could produce different results if the extrapolated impedances are changing your DC loads.

Back to top

IP Logged

jb_savage

New Member

Offline

Posts: 4

Re: Can Spectre/SpectreRF be used at 60GHz?
Reply #2 - Jun 11^th, 2011, 7:48pm

RFICDUDE,

OK thanks! -- I think it was a shooting issue..and dropped back to hb for better results with more harmonics. Also, dropped APS for the time being but want to use multi-core. I can use multi-core with Goldengate for about the same performance. I think there are enough points close to DC for sparam models. Using MMSIM7.2

Do you have a preference between SpectreRF and Goldengate? It seems SpectreRF has more analyses and options but Goldengate easier to use.

Back to top

IP Logged

RFICDUDE

Community Fellow

Offline

Posts: 323

Re: Can Spectre/SpectreRF be used at 60GHz?
Reply #3 - Jun 12^th, 2011, 3:37am

You might as well compare HB with and without APS. If there is no difference then APS is ok (most likely).

It is still a good idea to separately evaluate the wideband (down to DC) response of the nport model to the original s-parameter data.

I use whatever tool my employer will pay for that helps me verify the performance of the design. In the past, some have paid for both and others have paid for only one. But I have used Spectre more often simply because it is native and it does work reliably for most problems.

Back to top

IP Logged