how to measure impedance

what do you mean exactly ?
the impedance of a medium facing an electromagnetic wave ?
i guess if this is what you mean then it doesnt differ according to the signal strength
but it depends on the medium .. and the type of polarixation of the wave

I know what you mean; it's called large signal s-parameters with a big term.

You can measure them with a network analyzer, as long as the non-linearity is not too big and port input power is not too high.

There are network analyzers that excite non-linear networks with a pure sinewave and measure both basetone and harmonics. These are made for non-linear measurements. They do this with sampling gate techniques.

Go to http://www.maurymw.com/ and look for large signal network analyzers.

Techniques are explained in detail at the website of Jan Verspecht
http://www.janverspecht.com/

Did this help?

I want to measure the input or output impedance of PA, under TX on. Is it able to do that?

Yes it can also measure the S11 or input impedance.

What frequency is the PA?

And why would you need large-signal in-and output impedance? Load-pull measurements are a lot more common and cheaper to optimize PA's!

S11 can be tested in saturation. For S22 (input signal saturates he device)you have to saturate in input with a different frequency signal you use for s22.

Hello all.
I think I understood the problem.
Katykaty: I think you just wat to measure input and output impedances of an amplifier in normal operatin conditions, i.e., large signal.
Let's say, in normal test you are exciting this amplifier with a signal source that already has some quite high power. The output of the amplifier is connected to a load.
1-Between the preamplifier and the amplifier ( the one you are testeing) you put a coupler (for example 30db). This coupler should have the 2 coupled outputs available (A=direct and B=reverse).
2-Put another coupler between the output of the amplifier and the load. A similar coupler as I described in item 1 (C=direct and D=reverse) .
3- Now:
S11= B/A
S22=D/C

4- And:
S21=C/A
S12= A/C

To do this measurement you just bias the amplifier for normal operation, class A, AB, C, etc....
Apply the right input power.
Check that the output power is around the expected value and then follow the steps I described before.

You can also use a Network analyser to do the measurements for you. Just switch off the internal generator and put it into A/B mode.
You can calibrate it in low level, so that you don't burn the calibration kit! Assuming that the only non-linear device that you have is the amplifier that you are testing, the calibration in low power will be also valid for high power.

Godd luck....
Greetings.
S.

Thank you for all of you.

radiohead,
how load pull method can help on this issue? could you explain more detail
Thx

use the load pull test measurement instrument

Some indications how you can use load-pull for this application
http://www.maurymw.com/support/PDFs/5c029.pdf

Load pull is a measurement where you automatically and mechanically connect different loads from all over the smith chart to the output. If we measure the efficiency each time a different load is connected, we can look for the optimum load for efficiency. Compare is to best match circles and noise circles on the smith chart, it's quite similar.

radiohead,
while load pull at PA output, any effect on the input matching?

Hello Katykaty.
As I undertood from your first question, you already have the amplifier and you would like to characterize it, i. e., measure the s-parameters under normal working conditions.

In this case I still believe you should follow the procedure I described in my previous post.

The load-pull is a technique to be used when you are beginning a design. This means you have the transistor and you would like to design the matching circuits for the input and for the output.
When you design an output matching network you may have 2 different ways to follow: conjugate matching and power matching. According to the book from S. Cripps (RF Power Amplifiers for Wireless Comunications), the creator of the load pull method, you can use it to get an optimum output power match, what would give you maybe 2dB more output power comparing to a conjugate match.

I think, in your case if you really want to make a new design, you can simply design a standard conjugate matching network using the standard methods, and then just optimizing the circuit under the testing conditions I described in my previous post you should be able to get the maximum out of the transistor you chose......

Hope it helps.
Greetings.
S.