Hi Arwah, Well, the precise answer depends on the circuit, but I'm assuming this is something like a mixer. If the input and output frequencies are different, there must be some non-linear behaviour in the circuit, and so s-parameter analysis won't do the job. Analyses such as ac, noise, sp are small-signal linear analyses which work by linearising the circuit equations around a DC operating point. As such, they can only generate the first harmonic of the response to the input signal (which is assumed to be small enough to not create any non-linear response), so clearly that can't handle frequency translation. There are several ways you could compute the conversion gain of a mixer. For example, you would apply the LO frequency (or clock - the terminology depends on the system) as a large signal (e.g. sine wave or square wave) and then run either a PSS (Periodic Steady State - using the shooting method) or HB (Harmonic Balance) analysis depending on how non-linear the response is. Then you can do a small signal analysis to go along with this - for example, a pac or psp (if you use the pss analysis) or an hbac or hbsp (if you did the hb analysis). This then applies a small signal but linearises the circuit equations over the period you've captured in the PSS/HB analysis. As such this gives you a time-varying linearisation, which can then capture the frequency translation caused by the large signal periodic input to the circuit that you've applied. For more details, I suggest you look in the Spectre RF documentation in /doc/spectreRFTheory/spectreRFTheory.pdf - or look it up if you run /tools/bin/cdnshelp. There's also a workshop at /tools/spectre/examples/SpectreRF_workshop/rfworkshop.tar.Z . If you unpack this somewhere (using "tar xvfz rfworkshop.tar.Z") then there's a database and a doc directory with various workshops to talk you through how to do various analyses of such circuits. If it's a non-RF application, you may find it useful to take a look at various documents on The Designer's Guide , particularly the one on simulating switch capacitor filters. Hope that helps. I'm guessing you're using Spectre as the simulator - you say "in cadence" but there's no tool called "cadence" - that's the name of the company that provides the tools. Regards, Andrew.![]()
↧