Matching for output power spoils the S11 and S22

[rf_enthusiast]

Hi,
I want to design an OFDM transmitter for which I am designing a last stage power amplifier using the CREE CGH55015F device. The reference design from the maufacturer shows the S11 and S22 to be above -3dB which I feel is quite bad. They say it has been matched for max output power. Is there anything to be taken care of when matching for max output power by compromising the reflection coefficients?

Any help on this is most appreciated.

 

[BigBoss]

As you know, power matching may be quite far away from conjugate matching in power amplifiers.In your case, I might understand that S22 would be bad however I couldn't understand why S11 is so bad ??
Normally S11 is dependent to S22 mostly by S12 that should be generally low.
S11 should have not been impacted too much by this mismatching at the output.

 

[PeterMartin]

Yes its possible that both s11 and s22 could be quite bad, many transistors are quite 'transparent, although as Bigboss says the problem should be worse for s22.

You have several options:
1) Accept the less than ideal match or power transfer. Or , some in between compromise.
2) Add one or more isolators to the circuit.
3) Build a balanced amplifier

Incidentally its the same story for low noise transistors where the input of the transistor usually has to be in a mismatched condition for minimum NF

 

[rf_enthusiast]

Hi,
Thanks for the comments.
Big Boss:
Infact as you pointed out, S11 is a lot better compared to S22.

My next question is that is there any problem living with this value of S22? Please help me understand this. We are a bit reluctant to use isolators and balanced circuit due to cost considerations.

 

[FvM]

Please notice, that all S-parameters in the reference design are said to be small signal. So they are of limited relevance to
determine the actual amplifier matching at full loaded operation. S12 is however sufficient small to see, that S11 won't be
strongly affected by the output matching.

A more interesting question is, if the impedance matching has been actually optimized for maximum output power
(output network) respectively power gain. This can be mainly expected for the output network, but the input network may
need further optiomization to improve S11.

 

[rf_enthusiast]

Hi gentlemen,
Can anyone tell me if there is any issue in going ahead with a RL of -3 dB provided output power is optimized? What will be the consequences of such an approach?

 

[Abhishekabs]

Hello,
Have you designed & simulated your power amplifier? or you are asking doubts related to application note provided by supplier?
What is your frequency of operation & operating power ?
The losses will also depend on what type of board you are using.

There can be only one issue with moving forward with 3 dB RL whcih can be solved.
You PA may become unstable later due to various reasons, so you need to make sure that you are using buffer amplifer or to stabilize your PA.
So you are on safer side.

I hope I have answered your question.

 

[rf_enthusiast]

Hi,
Thank you very much for your suggestions.
Good question. Please find the answers to your queries below:
We have simulated the amplifier and the S11 (-8 dB) and S22 (-12 dB) are better in simulations. The reference design in datasheet shows S22 of -3 dB.
The output power is 30 dBm, Gain is 10 dB.
The frequency band is 5.7 GHz to 5.9 GHz.
The PCB used is 20 mil dielectric, 1.4 mil Cu RO4003C.

 

[kspalla]

put an isolator at the output section if possible