AnsweredAssumed Answered

PNA-X Noise Figure Calibration

Question asked by Madasila on Sep 24, 2009
Latest reply on Sep 25, 2009 by Dr_joel

I am trying to make accurate NF measurements on a LNA. The LNA has 32dB of gain, 0dBm output and a simulated NF of ~1dB between 4 to 6GHz. I'm having difficulies in obtaining a measurement in which I can have confidence since measured results and simulations comparisions show too high discrepancies (the first measured NF was close to 1dB but had 0.3dB ripples (slow ripple across the band) and other subsequent measurements were closer to 1.4dB). Which one is true???

The help files on the Noise Figure Calibration gives some hints on how to get accurate measurements, but in general, not much info is provided regarding some of the settings during calibration.

To start with, the LNA is a MMIC and as such, cannot be connected easily as close as possible to the noise receiver. I have to go through an RF Probe and a couple of foot long cable to connect to port 2.

I've seen somewhere on the web that the correct way to calibrate with the noise source is to have it connected directly to port 2 (this is the first step in the calibration). How would connecting the source to the cable instead degrade the measurement (does it really have a clearly adverse effect around 5GHz?).

Also, it is stated in the help files that power seen by the noise receiver should as close to +10dBm as possible. Of course, in my case, since my LNA outputs 0dBm, and I want to measure it at backoff, it's only outputing -10dBm... I understand that I can calibrate with higher power and then reduce it to my LNA's level, but how much will that affect the measurement? I've made calibrations with 0dBm and brought down the power to -40dBm after that (I can do this with the step attenuator set at -10dB and not changing it during and after calibration), I am just not sure if that is ok or not.

Another point: What Noise bandwidth do I want to be using? I am not constrained by time, I want accurate results. So, should I calibrate at 800KHz??? I've tried several bandwidths and in one instance, I had used 4MHz during calibration and switched between all possible bandwidths during measurements, for some reason, the 2MHz BW gave the best results... Which measured NF is the true one???

My IF bandwidth is very low (1KHz I think) and the Noise Averaging is set to 8.

Here are the steps I take during calibration:

Port 1 Output power = 0dBm (Power cal done prior to this including a 0.25dB loss for my RF probe)
Connect Noise Source directly to Port 2 using a female to female adapter.
Replace female to female adapter with male to female of same length and connect Ecal.
Measure 1 port cal
Disconnect Ecal, connect RF cable to same adapter and do a 2 port "on wafer" TRL cal.
Port 1 Output power decreased to -40dBm (without touching the step attenuator setting).

Am I missing something/doing something wrong??

Any advice?

Thank you