Any plans for a smith chart?
I had two other questions but just found the answers on own so going easy on you today.
Thanks… Kevin
1 Like
Smith Charts have not been on my mind. I’ve only used them once – to cut lengths of open and short stubs to noise match a signal to a narrow band receiver for some nuclear instrumentation back in the 1970s.
That said, I do think RF people would benefit from using SPICE more than they do – I’ve had the impression that RF people shun SPICE for cultural reasons and not on valid fundamentals. Myself, I’ve used SPICE for a SIS Josephson Junction mixer to mix 500GHz radiation from space to a 2GHz IF. Piece of cake.
–Mike
1 Like
There is some truth in your observation about spice and RF. I think that is changing though. In old days the rf transistor vendors provided s-parameters and not much else. No non-linear models to run on spice unless you dummied up one yourself. The other thing is a lot of the early RF courses were centered around s-param design methodology. So, noise figure and gain circles and all that plotted on smith charts. The RF simulators accommodated that kind of design whereas spice didnt. However, with the right test sheets and some .meas statements you could get spice to do most of it. Spice is still generally short of microstrip and stripline models though. RF or microwave design/simulation on PCB takes some extra effort if only using spice. The RF harmonic balance simulators have all this. Tradeoff is cost though as licenses are expensive. And when license expires you often lose access to schematics. In any case, we do a lot or rf design on spice these days but I have workarounds so dont worry about smith chart. BTW… why arent the quorvo guys asking for this given quorvo= rfmd + triquint? I thought they did RF design there. 
Spice in RF design “if” models, models, models.
And teeny-tiny steps making long sim’s.
And then you still have to verify EM for physical coupling and “stuff”.
Smith Chart quickly loses its luster with wider bandwidths and modulated signals and Monte Carlo(process variation verifications).
We certainly do RF work around here! I’m in conversation with our modeling experts about traditional RF simulation tools versus SPICE. From what I can gather, SPICE was used in the ‘early days’ but then the specialized RF tools came along, and SPICE wasn’t up to the task at the time.
That crack was just me teasing a bit Jeff to motive the eventual inclusion of a Smith Chart. BTW, in most applications only a linear AC simulation is required to say extract and plot s-parms on smith chart and this is very very fast. Useful for matching network and filter designs and a lot of other RF techniques. We actually do a lot of RF simulation on spice (or spectre) especially for RFSOC designs where we have to cosimulate with a large amount of digital. The RF simulators often have a rough time under this condition, especially spectreRF, they just wont converge. Behavioral system modeling is another area where we often use spice in place of matlab. I did just run an RF oqpsk transceiver system on qspice a couple of days ago… seemed to work just fine once I fixed some syntax incompatibilities in our behavioral libraries. The graphic shows the system simulated. The qspice plot shows bits received matching bits transmitted and the expected oqpsk I/Q plot.
wow- that is very interesting! I’ll be talking to the head of our RF modeling group next week. This thread will be of interest to him.
1 Like