how to design a output network in power amplifier

[smartme]

these days, i design a PA, well, a problem about the OMN(output matching network) puzzles me,
from the loadpull simulation, the optimum loadline is 3Ohm, while the output impedance of power cells is 30Ohm,
so the dB(S22) between the power cells and the loadline is -1.7dB, if i design the OMN using the ideal components,
the dB(S22) between the antenna and the output network also is -1.7dB, while the design goal tell me the dB(S22) should be less than -10dB, oh, how to design this network?
while the resistor could change the dB(S22), while the PAE is so bad...

 

[BigBoss]

There is always a copmromise between "Complex Conjugate Matching" and "Optimum Load Impedance".You don't have to use exaclty Optimum Load Impedance point that you have found on the Smith Chart because there are Delivered Power Contours around this optimum impedance point.So, you should renounce a little bit power to get significant PAE improvement and realtively easier matching.
I suggest you draw Optimum Load Circles and PAE circles for some expense of Delivered Power and then find a compromise point that satisfies your sepcs.

 

[smartme]

so, when a PA see the load (antenna), maybe the load is changed according to the temp, cell phone structure, and ... how we design a PA output matching network in these mismatch condition, if we decrease the output reflection coefficient , does the PA could work under worse mismatch condition?

 

[aniltunwal]

Hi

PA is always matched to the Optimum Load which we can get from the load pull simulation of the device if you have model available for the simulator.
dont worry about the mismatch between antenna and the PA output. Use isolator between antenna and PA output.

Anil

 

[tony_lth]

Use isolator between antenna and PA output.

I don't think so. 1st is the insertion loss of isolator. 2nd is the big package of isolator, especially in very tough volume available.

- - - Updated - - -

Use isolator between antenna and PA output.

I don't think so. 1st is the insertion loss of isolator. 2nd is the big package of isolator, especially in very tough volume available.

 

[E Kafeman]

so, when a PA see the load (antenna), maybe the load is changed according to the temp, cell phone structure, and

PA optimum load can be changing due to frequency, band, and power level too (TRP). Antenna should also be impedance optimized against RX for best S/N (TIS), a impedance curve that seldom is similar to TX impedance. Commonly is the worst mismatch due to that antenna impedance changes when phone is in free space relative handheld. That problem can be reduced by finding an alternative antenna geometry and matching such that optimum load is somewhere between. TX and RX curves must often be weighted for best compromise. Some phone chips have internal switchable capacitive (cmos) load for RX which can improve situation a bit.
Frequency depending conjugate impedance match is a must. Matching is often a iteration process adjusting matching network and antenna shape. A few years ago was a customer specification, for me as antenna designer, something related to 50 Ohm and verified by passive measurements. It was so much simpler then.
A cellphone PA that is mismatched is seldom a functional problem, I test them very hard, with and without antennas or even accidentally DC shortcuts and they seem to survive the most. 10 years ago there was chip-sets that could die due to to much reflection but I guess internal protection in general is much better today.
Have however seen minor mismatched PA, that increase current consumption a lot, and some PA have impedance requirements also for harmonic frequencies.
A common request after the antenna and impedance matching design is done, is if I can do a minor adjustment to make the antenna a bit less effective in the range 3-7 GHz without changing anything else, as the phone not did pass EMI tests. "Is two days enough" is their second question.
That kind of job can make me a bit tired. If you design a PA, please design it free of harmonics.