Linearity for handset Power amplifier

[haitaoz]

gtc3 wcdma

When I measure the WCDMA PA,I found the ACPR is not symmetric on the lower and higher side.The difference is about 6dB.Does anybody know how to make them even and better?Thanks a lot in advance!!!

 

[RFCMOS]

Hello,

what about the signal at PA input?

 

[haitaoz]

The input signal is standard WCDMA GTC3.I did the measurement without PA,the ACP looks good.

 

[RFCMOS]

hi,

Do you see this disymetry on the ACPR for all power levels?
Do you have several bias mode on your PA?
regards

 

[haitaoz]

I have two modes in this PA: high power mode and low power mode.
This disymetry on the ACPR only happened in 10dB range(from 6 to 16dBm) on low power mode.

Thanks!

 

[RFCMOS]

Hello,

Some non-linearity there...
what if you put a WCDMA signal generator at PA input just to to ensure that there is no interface issue betweenypur transceiver and PA (mismatch, harmonics, intermodulation ...)?
Make the measurement just at PA input and then after your diplexer.

 

[haitaoz]

This is the way I did the measurement and notice the problem.

Thanks again!

 

[RFCMOS]

Hello,

Did you perform the measurement on application board or PA demo-board (PA only)?

 

[haitaoz]

application board

 

[RFCMOS]

Hello,

If you have a demo-board, try to do the same measurement on the demo-board.
Otherwise, you have to investigate around the PA control lines, supplies and and look carefully to your layout to see if you are not coupling something (any sensitive lines close to your PA or bad ground).
Make the measurement just at PA output and for different freq. Try to do the measurement also just with a pure sinewave.

 

[springf2000]

this is the memory effect of your power amplifier, y should test the following steps:
1)improve the power supply video bandwidth;
2)improve the heat sink conditions
3)improve the matched network

good luck

 

[totowo]

this is the memory effect of your power amplifier, y should test the following steps:
1)improve the power supply video bandwidth;
2)improve the heat sink conditions
3)improve the matched network

good luck

Springf2000 is right. The phenomenon is caused by the memory effect which is descripted by many papers.

The narrow-band matching network differs the impedance at upperband from lowerband, differs the impedance at w=2f from w=0....so that the nonlinearity product is quite different....3~6dB or even 10dB difference is often observed.
So you could improve your matched network.

Heat sink usually affect the lower modulated freuencency because the heat time is about kHz level...

Use an power supplier as good as possible and carefully decoupling.

Best regards.