Hello

Can we get Noise Figure measurement uncertainty for PNA-X for all methods of measurement that this Network Analyzer can perform.

There is a plot on KeySight web side that compare noise figure uncertainty for: Noise Figure Analyzer, difference Spectrum Analyzers and PNA-X vector cold source Noise Figure measurement. We need something similar just for PNA-X it self.

The PNA-X have now so many variation in Noise Figure measurement ( vector, scalar, noise source , power meter, noise receiver, RX receiver ...) that it will be good to tabulate the advantages and disadvantages in terms of uncertainty, measurement speed and calibration time for each method.

For example; If you have amp that have 6dB NF you probably do not need to run vector, just scalar is more then sufficient and probably faster.

I find out that if out if you use Noise Source instead of the new method with Power Meter, the calibration time is much shorter. It is better or not I have no Idea.

Best Regards.

Can we get Noise Figure measurement uncertainty for PNA-X for all methods of measurement that this Network Analyzer can perform.

There is a plot on KeySight web side that compare noise figure uncertainty for: Noise Figure Analyzer, difference Spectrum Analyzers and PNA-X vector cold source Noise Figure measurement. We need something similar just for PNA-X it self.

The PNA-X have now so many variation in Noise Figure measurement ( vector, scalar, noise source , power meter, noise receiver, RX receiver ...) that it will be good to tabulate the advantages and disadvantages in terms of uncertainty, measurement speed and calibration time for each method.

For example; If you have amp that have 6dB NF you probably do not need to run vector, just scalar is more then sufficient and probably faster.

I find out that if out if you use Noise Source instead of the new method with Power Meter, the calibration time is much shorter. It is better or not I have no Idea.

Best Regards.

> Hello

>

> Can we get Noise Figure measurement uncertainty for PNA-X for all methods of measurement that this Network Analyzer can perform.

> There is a plot on KeySight web side that compare noise figure uncertainty for: Noise Figure Analyzer, difference Spectrum Analyzers and PNA-X vector cold source Noise Figure measurement. We need something similar just for PNA-X it self.

Great Idea, but it would be just an example. There is a complete PNA-X noise figure uncertainty calculator that is available to run on the PNA. Of course the uncertainty depends on the properties of the DUT; I know, alot of users don't like that we say that, but it is just the fact. And, I will point out that the SA and NFA uncertainties do not account for any source mismatch effects; they are ignored (even though they tend to be the largest errors) because there is no way to quantify their effects without a full noise-parameter characterization (to find out how the NF of the DUT changes with a change in match). It's not that they are exactly untrue; they are clear that the match effect is not included, but it is significant.

> The PNA-X have now so many variation in Noise Figure measurement ( vector, scalar, noise source , power meter, noise receiver, RX receiver ...) that it will be good to tabulate the advantages and disadvantages in terms of uncertainty, measurement speed and calibration time for each method.

That's what the NF uncertainty calculator is for (not sure it handles scalar because, as I say above, without Noise Parameters there is no way to quantify the error).

Here's the link (I had trouble finding it too, I'll give jvall a stern talking to): http://www.keysight.com/main/software.jspx?ckey=2226545&lc=eng&cc=US&nid=-11143.0.00&id=2226545

(I eventually found it by using the search in keysight with the keyword "PNA noise uncertainty"...he'll tell me "google it!" I'm sure)

> For example; If you have amp that have 6dB NF you probably do not need to run vector, just scalar is more then sufficient and probably faster.

> I find out that if out if you use Noise Source instead of the new method with Power Meter, the calibration time is much shorter. It is better or not I have no Idea.

>

The errors are almsot identical below 20 GHz, as the noise source is slightly higher uncertainty but the power meter has added uncertainty for the power transfer step. Roughly they are both about 0.15 dB uncertainty after cal, not including source match effects.

In PNA, the error is almost unrelated to the noise figure, and almost entirely related to the calibration accuracy of the noise source or the power meter. So a 6 dB NF device will have 0.15 dB uncertainty and a 1 dB NF device will have the same. This is of course only true when the total noise power is at or above the noise power of the PNA-X receiver. As the noise power gets lower than the PNA-X (about 7-14 dB above kTB from min to max freq), the error from jitter increases but this can be reduced to an arbitrarily small value through noise averaging. At 0 dB excess noise, the drift of the reciever and test cables will be the limiting factor as you are subtracting the noise of the PNA from itself and digging down 7-14 dB. I usually say that you don't need to worry about this if you are at or no more than 3 dB below the PNA-X noise floor.

Thus, if you have a 20 dB gain amplifier with 0.3 dB NF, the uncertainty is the same as a 14.3 dB gain amplifier with 6 dB noise figure. Figures, eh?