We are working with a mixer that produces large out of band spurs. Our concern is that these spurs will push the amplifiers on the input of the noise receiver into compression. We have done this kind of testing before using a E4440 spectrum analyzer with the noise marker and a series of external block gain amplifiers and this was a problem. Is there a pre-selector on the input of the noise receiver before the amplifiers>
So, if your large signals are more than about 1/2 to 2x the freq of the measurement, they will be filtered out.
The 029 hardware has some gain stages, so you can lower the gain to avoid overdriving but only a limited amount.
Using a Yig Filter is possible but often the reproducability of the YIG S21 loss is not so good. For noise figure it's not much of a problem as the SC21 error due to the YIG error will match the error of the noise power (that is the error due to the YIG happens to both output power for SC21 and the noise power), so noise figure would be OK, but the SC21 of the noise measurement would have the YIG error.
However if you used a bypass switch around the YIG, you could use a second channel to measure just the SC21 of the mixers and not worry about the YIG error.
There might also be a way of using triggering to do this. If the NF application uses two sweeps, one for gain and one for noise power, single triggering could be used to sweep the gain with the YIG out of the path and then sweep the noise power with it in the path. I'm not sure if this would work during the calibration though. Perhaps the YIG could be switched in the path when the noise source standard is being measured for calibration and out of the path for all other calibration standards.
> I take it that each noise figure channel will have its own gain measurement associated with it be default. That this gain measurement is taken when the noise figure channel is triggered. When triggering a channel that uses the noise figure application there are actually two different measurements taken, one for gain and one for noise power. Not to belabor the point but this is a complicated machine and I feel understanding details to this level is best.
yes - your understanding is correct. none of the PNA applications by default depend on data taken by other channels. And yes, the more you know, the more you can get out of your PNA-X.