measuring power amplifier noise
Assuming that your power amplifier is linear an not class C, you should have no problem as long as the output power is higher than the input noise of your spectrum analyzer.
Some problems come in when you have a class B amplifier and the noise source cannot drive it out of the slightly nonlinear quiescent state.
you should use a attenuator as the load ensure the security of your instrusment
This brings up a very important concept. The sequence that you turn on and connect the boxes makes a big difference. You do not want turn on transients damaging the following stages. A limiter at the analyzer input is always a good precaution.
Pay attention, with a PSA you will have a bigger measurement uncertainty than with a NFA.
The noise source power must not drive the PA in nonlinear behaviour, the NF meas is a linear measurement.
In addition an attenuator in front the PSA will raise the NF of the instrumentation and so the meas uncertainty.
Some choiches must be done taking in consideration the expected PA NF.
What is the reason to measure the NF of a PA?
The NF, as correctly said, is a linear measurement (that sometimes it is done with a large signal, for example in basestation receivers, that is useful to evaluate the noise behaviour of the circuit in operating conditions) and for a PA is useful to know what is the absolute noise level at its output, so why NF?
Thanks
Mazz
Usually Noise Figure of a power amplifier is not so important.
More important is the Noise Power:
PNoise@Output (dBm) = PNoise@Input + GainSystem + NFSystem
cheers
it is possible to find a convergence on opinions wrote by springf2000, flatulent, and marcomdd.
Me too i've experienced a destructive spike from PA. Instrumentation seafty coming firts.
But the use of the internal attenuator (if mechanical) is enough. Connect PA, set Attenuator to 70dB, then turn-on PA, then step by step, set RF attenuation to 0 dB. During this operation take alook for the full band power injected into PSA ! Now you are in a steady, secure state. The same procedure will be done at power-off.
Uncertainty: of course you cannot get the same uncertainty of the NFA. But for limitated Y factors (i.e. 2 or 3 dB) the detectors inside PSA are more linear than detectors inside other SA's.
So for my opinion, if you don't require very accurate measurement, the PSA my be used.
But... in any case, due to High NF of PSA, to get a relaiable measure, the Gain of the PA shoul'd be high enough. A pratical way is look at the noise level; with noise source off, connecting PA the noise shoul'd rise some dB's (5...10 at least).
Two ways to express the same thing: the noise power generated inside the DUT.
Thank you for all replies. I need to make a adaptor to convert an existing RF output to another f0. That why i am so worry about noise. I could not access IF from the existing device. That why I need to know about the new PA noise. What I worry , as flatulent
mentioned, if my amplifier is not class A, I will have a problem. I usualy test the power amplifier with the agilent PNA and an attenuator, it work ok I can get P1, Gain, but not IP3. I dont know that the PNA can do the noise figure measurement or not. If I connnect the attenuator, how can I caribrate it out?
A noise source -> an athennuator->PSA for caribration.
A noise source -> PA->an athennuator->PSA for measurement.
Is that correct?
boy, it seem you have more than a PA, it seem a system with a conversion a non linear PA and more...
In this case the NF measurement may be possible, with non conventional techniques.
I suggest to try the following.
A rough expression of NF is:
NF = (Sout-Nout)-(Sin-Nin) [dB]
It is valid for S>N (at least 30 dB).
The ratio Sout-Nout may be read with PSA properly tuned at the right frequency.
The Nin must be known, and this is the greatest difficulty. Generally you cannot measure it using PSA...but if you have a very noisy CW generator (a bad generator), maybe used for this scope.
To find a noisy CW generator try the PNA used as CW generator and connecting to PSA, take a look its noise; if the noise rise, OK! you can use it.
In any case the network analizers cannot be used, itself, to measure noise, also if they are calibrated... If you have some doubt about calibration and how it works... may be the subject for another post.
Yes, It is.
But, pay attention : if the gain of the PA is low (less than 30 dB) you will have a very high measurement uncertainty.
I suggest you to read Agilent AN 57-1, 57-2 & 57-3.
bye for now
Pay attention please ! I remember that using a SA to make power measurements (especially noise power) is not a very accurate way.
In particular thare are, at least, 3 sources of error (systematic errors ) related to:
1 - Rayleigh distribution
2 - Log amplifier
3 - 3 dB/noise power bandwidth
These effects need a correction of (more or less) 2 dB.
Read AN 150 by Agilent for more detailes.
Bye for now