how to decide the optimum load impedence of power amplifier

[whiteshirtee]

cripps input matching

Hi, all.

I'd like to design a power amplifier using ADS. In order to output the maximum power, I have tried to find out the optimum load impedence of the amplifier transistor. Although ADS offers the Loadpull simulation, it doesn't work well and the simulation takes a very long time. Can anyone have a good idea to decide the optimu impedence of the amplifier? Thanks a millian.

 

[madengr]

+maximum power transfer +output load +transistor

You need to take a look in Steve Cripps' book "RF Power Amplifiers fro Wireless Communications" to fully understand what's going on. In short the optimum load presents a load line (class A) that swings between gate-drain breakdown and maximum current at pinch off.

Ropt=(Vdgb-|Vp|-Vk)/IDSS

Vdgb= gate to drain breakdown voltage
Vp = pinch off voltage
Vk = knee voltage

You'll also have to resonate out the output capacitance.

Instead of load pull analysis just tweak the tuner values until you get maximum power.

 

[whiteshirtee]

Re: how to decide the optimum load impedence of power amplif

Thank you, madengr!
It's realy a easy way to find the appropiate load impedence in your way. But there is another problem. While we are changing the load, S12 changes also. That is to say, there is a matching problem at the input. We can get the maximum output power at this certain input matching network, but it isn't the maximum output power for all possibility. I want to find the maximum output power with the best input matching. I'm really appreaciate for the answer!

 

[dingjingfeng]

Re: how to decide the optimum load impedence of power amplif

how to get the correct value of output capacitance,
and resonate it out.
we should resonate it out with pi match network or shunt inductor?

You'll also have to resonate out the output capacitance.

 

[brumi]

Re: how to decide the optimum load impedence of power amplif

The best power transfer is when the impedances are matched.
If you don´t know the PA impedance you could try to obtein it with try and error.
Test the power with diferents terminator resistors . a simple three rule can help you.
Before that you must transform the PA impedances to 50 ohm (line impedance).
Thats simple but with a few instruments you can do that .
Remember the resistors must be no inductive type.

 

[manojshukla]

Re: how to decide the optimum load impedence of power amplif

Just measure the output resistance with the help of multimeter at the output.
It is easy and effective, if you have made the circuit.

 

[madengr]

The input and output tuning is an iterative process. Get some tuners and try it in the lab on a real part. If your load impedance is varying that much you may want a circulator on the output. The input match that yields maximum power may not have the best return loss. The FET is a voltage controlled current source so you want the optimum gate voltage waveform, which may not yield good return loss.

The highest power generated from a transistor is not when the impedances are matched, rather when is is presented with the optimum output load and the gate is driven with the optimum waveform. Maximum power transfer occurs between two "black boxes" of conjugate impdedance.

The value of output capacitance is typically taken from the small signal (s-parameter) measurments when dveloping a model. Of course if you are using a tuner (and your reference plane is at the FET drain) the output reactance is the conjugate of the tuner reactance.

The output capacitance is usually lumped into the impedance matching network. Yes, a PI low pass section is often used. Shunt L would require a DC block so it's not too practical. Again, see Cripps' book as he shows several examples. All this assumes you are relatively close to the FET dice, so if you are at package reference plane than a different network may be needed.

From what I have read (If I remember correctly) the output resistance (on GaAs) changes drastically at a few MHz so a multimeter won't yield the proper results. Check a book by Mike Golio as it talks about model extraction.

 

[whiteshirtee]

Re: how to decide the optimum load impedence of power amplif

We can loadpull the transistor if we get it. That's helpful. However, what I'm doing is to design a MMIC. So I have to find the optimum load impedence and the appropiate input matching network by ADS. I can find the optimum load impedance by tuning the load resistance and canceling out the output reactance at certain input matching condition. That's good.but when I adjust the input matching, the optimum load impedence changes also, then the optimum load impedance has to be found for another time. and so on. It's a iteration. I think it is not wise to do this. So what is needed is how to find the optimum load impedance and oppropiate input matching network by computer or at ADS in a easy and wise way.

 

[dingjingfeng]

Re: how to decide the optimum load impedence of power amplif

when I ajust the input power, the load pull result will change, no to say to change the bias situation.
so when we design a linear PA and the input power changes greatly, how to find the best load.
Is there a method to get the different optimum load under different bias and input?
so we can select the best load.

 

[madengr]

Re: how to decide the optimum load impedence of power amplif

That's good.but when I adjust the input matching, the optimum load impedence changes also, then the optimum load impedance has to be found for another time. and so on. It's a iteration. I think it is not wise to do this. So what is needed is how to find the optimum load impedance and oppropiate input matching network by computer or at ADS in a easy and wise way.

That's the way it's done, at least for me, with real hardware and in simulation, mostly hardware. I assume you have been source pulling as well. You alternate back and forth between source and load pull, picking the peak output power, and eventually it will converge (or at least about a mean value). There is always going to be some change but the change becomes irrelevant since your matching network tolerance won't be high enough. Now with the load pull contours you can overlay a monte-carlo simulation of the matching networks and get insight into how much variance you will have in the output power.

when I ajust the input power, the load pull result will change, no to say to change the bias situation.
so when we design a linear PA and the input power changes greatly, how to find the best load.
Is there a method to get the different optimum load under different bias and input?
so we can select the best load.

The main spec for a PA is usually ouput power and gain, so if you drop the input power and try to find a another load then you through the spec out the window. Maybe you want the load that gives the best efficency, regardless of power. This would have to be a class B or C amp so then you would plot efficiency countors for various power levels, overlay them, and pick a point that is a tradeoff between all levels.