Continue to Site

Welcome to

Welcome to our site! is an international Electronics Discussion Forum focused on EDA software, circuits, schematics, books, theory, papers, asic, pld, 8051, DSP, Network, RF, Analog Design, PCB, Service Manuals... and a whole lot more! To participate you need to register. Registration is free. Click here to register now.

How to get the correct circuit of power amplifier design?

Not open for further replies.


Junior Member level 1
Jan 29, 2002
Reaction score
Trophy points
Activity points
mrf21125 ads

When I design PA,I find that using the typical circuit in the datasheet sometimes lead to bad result.Furthermore,the recommended microstrip circuit always show it's poor performance in EDA simulation.How
do this happen?Would you mind tell me how I can get the correct circuit of PA design.
By the way,the PA transistor I used were from
Ericsson and Motorola.

Please be more precise with design specifications and include bandwidth, output-power level and transistor used. Have you tried to simulate with MicrowaveOffice?

It is difficult to give good advise without having knowledge about any details regarding your design, but it sounds as if it is about a HF PA. Generally, correct decoupling of the power supply is a must in a practical circuit. Also, remember that for GHz-range PA:s capacitors do not always behave perfect. Furthermore, the datasheet specs are often typical values and the deviation between samples could in reality be noticeable. One must also pay attention to small details not usually thought of; vias for example will work as small inductances at very high frequencies. The type of circuitboard laminate also has an effect. Often Teflon boards are used. Design of microstrip circuitry takes into account epsilon_r, which is different for various circuit board materials.

I use MRF21125(MOTOROLA)LDMOS power transistor and the circuit operation frequency is 2100M-2170M.The software I used to simulate are Microwave Office (version 4.02) and ads1.5.The datasheet provide three Input and Output Impedancefor 2100M,2110Mand 2170M,When I use these parameter to simulate the circuit recommended in the datasheet I found the result is not ideal.Now I think that in PA design using the Input and Output Impedance for s-parameter simulation(s11,s21,s12,s22)is not enough,what parameter should I use and what should I simulate in order to get best performance in PA's pcb board?Thank you!

First, you have to choose the class of your PA such as A, AB, B, C, F, G, H, HG... which implies different choices... say we choose class A that means that you have to bias your transistor in the middle of the V-I characterics... for your frequency range 2GHz I would not use distributed constants but concentrated (capacitors, inductors...)
In PA you generally match the input and tune the s22 to have highest PAE (this is a quantity to look at...) send me the configuration you wanted to used i''l give a look at it... bye

PA is always work in nolinear area, which S parameter for linear simulation is no use in deed.
So U must use a nolinear model to simulate Ur circuit. Usually the spice model will work well, have U one ?
In MWO ,it can be placed just though nonlinear lib in element window, while in ADS, U must input every parameter by Ur self.

Good luck.

But I still think that it is no use to spend a lot of time to design and simulate, for the S parameter the manufactue supplied is just for small signal,
and the spice model they supplied is not accutate enough either.

Just follow the application note, use the mictosrip with the same impedance ane electronic length , and the cappacitors with value proportional to 1/frequency.
That is OK.

Remember however, that if you use nonlinear simulation you'd better do one-tone simulation. I suggest you also to look at Pin-Pout curve, which shows the compression point, look at harmonics level that should be kept not so high and at PAE = Pout/(Pin+PDC)

Why is nobody mentioning Large-Signal Analysis. One cannot have an accurate output pwr read only relying on small-signal and S-par are small-signal.

Only to those who don't know, designing a PA with large-signal analysis consist of performing a Load-pull analysis to determine the Zopt. Then your matching depends on this. There are steps to follow but if you go to DesignGuideLine in A/D/S and following at the same time some books (Walker is 1), then you may get it right.
If more details required, please ask.

What about the bias condition?

Collector / Drain current? Vcc/Vdd ?

In general, the higher the drain/collector voltage, the more power you can develop across a load ((E^2)/R). Also, the higher the current, the higher power ((I^2)*R). Obviously the device can't handle infinite voltage/current so you must choose one that fits. You are usually stuck with a VCC since maybe you are doing a cell phone so you have only three volts. Then you must have higher current. If your amplifier is 100% efficient, then VDD*IDD would be your power output. But that is not the case so you will need more DC volts or current since the remainder of the power is lost as heat in the inefficient amplifier transistor.
And I agree, S-parameter and linear simulations are not much use here. You need to use harmonic balance and source/load pull simulations to do the job right.

Not open for further replies.

Part and Inventory Search

Welcome to