How can I improve IMD3 ?

Something is squaring up the signal, back off power density
/ bigger devices (at cost of more C throughout)?

dick_freebird said:

Something is squaring up the signal, back off power density
/ bigger devices (at cost of more C throughout)?

Click to expand...

If i lower the power, would not I lost the power and drain efficiency ? and I can not reach the required levels.

This is a 25W-Class Device and you overdrive the transistor so transistor gets into hard saturation/hard nonlinear mode and exhibits strong nonlinearity.
Also there are some design and layout errors. You cannot design such Power Amplifier without Load Pull simulation/measured data.
Re-visit your design steps by taking Load Pull concept into consideration.

Brkyapali said:

I have been designing Power amplifier and I got a serious problem with the IMD3 point. I have got all the power and efficiency I wanted but IMD3 is very high and I don't know what to do.

Click to expand...

What do you think your IMD3 is? What do you expect it to be?

Brkyapali said:

IMD3 value is almost same on the schematic results.

Click to expand...

What do you mean by schematic results?

Brkyapali said:

In the fifth photo it says 41.7 dBm for fundamental

Click to expand...

That is just a measure of output power for one of the carriers, not IMD3...

Brkyapali said:

but normally I get 45.6 dbm

Click to expand...

"normally" meaning what?

It's unclear whether you are measuring IMD3 correctly in the first place.

Brkyapali said:

In the fifth photo it says 41.7 dBm for fundamental but normally I get 45.6 dbm

Click to expand...

That is with two tone excitation, correct?

The level for two tone measurement is usually 6dB down from the single tone measurement, i.e. 45.6dBm - 6 dB = 39.6 dBm
I think you drive the amplifier too much into saturation with two tones of 41.7dBm

Check first P1dB Compression Point to achieve the capability of your Amplifier. In order to do this, sweep Pin from very low levels to appropriate level.
If your P1dB Compression Point is not around almost 10 dB lower than expected OIP3 value, there is absolutely an design error.

RF and Microwave Design are done step-by-step.

If your target is to achieve a good OIP3 ONLY, you have to find the "sweet point" of this transistor by derivating Id-Gm curve.
Gm value will exhibit a sweet-point that reflects minimum 3rd order harmonic.
So third order derivation of this Id vs. Gm curve will give you an insight.

Usually I measure IP3 setting the two tone level even lower then 6dB from the level of the single tone test.
In this way I am sure I don't saturate the PA.