S11 and S22 parameters

[Nahim]

For a powerful Gan transistor (100W) I calculate the matching
I received the
S11 and S22  -12dB
S21  -0,5dB
Freq SPAN 500-2500MHz
How bad or good are the results? What consequences can appear?

[TheUnnamedNewbie]

The idea that S11 and S22 must always be great (and under 'great' is usually understood <-10 dB) is a myth. It all depends on the situation and networks around the device you are discussing the S11/S22 of.
In fact, demanding good S11/S22 when it is not in fact needed can lead to sub-optimal performance.

Low-speed analog design usually has S11 = 0 dB or something, and I've never seen anyone complain about that.

That said, in situations where it does matter, having bad S11 or S22 can result in standing waves, oscillations, suckouts, damaged components, poor performance, bad SNR, etc...

[Nahim]

In fact, the parameter S21 is more important? Is it responsible for the loss of the signal?

[TheUnnamedNewbie]

I would say in amplifiers (and even some passive networks), S21 is often more important, but again, it depends on the situation.

A classic example is a power amplifier, where it is not uncommon to sacrifice S22 and even S21 in order to improve the maximum output power and compression points.

And then you haven't even started going down the rabithole of power/voltage/current gain, which are not the same if input and output impedance are not equal. For example, your transistor has an S21 that is less than 0 dB, ie, it seems to be worse than a wet noodle. But I suspect that might be because you are plotting it with 2 50 ohm ports, instead of actually matching the thing for gain or power.