I am working on the matching network of an RF power amplifier (3.5GHz).
The result coming out of S-parameter simulation (sweep from 3-4GHz) is
-- @3.4GHz, S21=11
-- @3.5Ghz, S21=11.5
-- @3.6Ghz, S21=11
So the gain vs. freq seems to be flat.

Then I did the Harmonic Balance simulation with very small input power, so I'd expected the result close to what I got from S-param.
However, there's a big difference
-- @3.4GHz, Gain=15
-- @3.5GHz, Gain=13
-- @3.6GHz, Gain=9
Now the gain vs/ freq looks like too curvy!

I am confused with the discrepancy, I'd high appreciate if anyone could help me explain it.
Thanks.

Steve

Hi Andrew:

Thanks for your notes.
First of all, I'm using Agilent ADS.
My S-parameter simulation is just like the standard one, put two 50Ohms ports, and simulate the S parameters.
As for the HB simulation, I put a 50Ohm resistor in series with the source and have another 50Ohm resistor as the load.
Everything else for the power amplifier is all the same.
Yeah, the gain I'm talking about in HB simulation is the power gain, but it shouldn't make any difference, cuz I'm using all 50-Ohms systems.
I did the power sweep of input, from -10dBm to +10dBm, and I plotted the gain vs. power.
The input power level is as low as -10dBm, and it seems to be the asymptotic value in the lower range since the gain plot is flat with input lower than -5dBm.
But still, for the gain with a low input, it doesn't match the gain given by the S-param simulation.
I understand that these two simulations are intrisincly different, but when my input power's low, they should be telling the same story, shouldn't they?
Confused here!
Hopefully you could give me a hint or something.
Thanks a lot for your time.

Steve

Hhi Vuk:

Glad to see expert like you.
Yes, the gain I was talking about is the ratio of the power delivered to load to the power available from the source.
The problem is that in some of my simulation, the two gains (one by SP and one by HB) agree with each other.
But when I change the parameters, the difference started to show up.
So sometimes it's correct and sometimes it's not. Isn't that weird?
And also, from the S-paramter, the gain over frequency is flat (from 3.5~3.7GHz, say, the difference is only 0.5dB).
But in the large signal HB, at very low level, the gain differs by as much as 5~6dB.
That's totally different from what I got from small signal simulation.
As you said and I understood, when at small signal level, HB should be the same as SP.
Can you think of any possibilities that can cause this?
Thanks a lot.
By the way, I happened to see your PhD dissertation online, that's awesome!!!
Actually I'm very interested in simulation technology besides using it. I'm writing my own time-domain nonlinear simulator (not shooting Newton though).
Maybe we could talk more about it.

Steve