questions about microwave amplifier measurements

HI:

I designed an amplifier at 5.5 GHZ.
I tested the amplifier from 2 GHz to 6 Ghz. I would like to ask some questions here:
My S11 at 5.5 was -2.6 dB (based on mismatched input) and S22 was
-11.4 dB. when I was tuning the output matching network in the lab, I was able to get S22 around -16 dB but I found S21 at the same time around 1.5 dB. Since my design was based on available power gain (GA), should not be S21 independent of S22? (or GA independent of gamaL)

should the magnitude of S11 and S22 be below 0 db from 2 GHz to 6 GHz? as I wrote above at 5.5 GHz it is good but at some other frequencies i found the magnitude of S11 and S22 above 0 dB, does this mean my design is bad at other frequencies and good at the design frequency?

thanks for your help
ayhz2002

oscillations

If your S11 and S22 are positive dB readings it means there is a negative resistance at the two ports which will cause oscillations at low source and load impedances.

Hello ayhz2002 !
What amplifier design do You need: narrowband or broadband ?
If You need the narrowband design - You must make the good in/out matching only on 5.5 Ghz, but in narrowband the 'good' matching usually is S11,S22<-30 dB. In this case You must aim at minimum gain on out the working frequency band. Usually it can be made by using the narrowband filters as matching circuits.
In broadband case You must match the amp. device in the broad frequency band (it's not a simple problem !). In this case the 'good' matching usually is S11,S22<-20dB. If You haven't the hard limitations on input noise temperature, You can use the internally-matching IC, it's simplifies the problem.

The available power gain can depends on S22, because if You changes output matching circuit - You change it's input impedance. As I understand, the available gain is S21 with ideally matching ports, and GA of active device (transistor, IC etc.) independent of matching, but GA of whole device (transistor+matching circuits) depends on matching circuits construction.

Best regards,
Kit-the-great

ayhz2002,

If you get S11 or S22 larger than 0dB at specific frequencies or with a specific matching network then your amplifier is conditionally stable.
Because then your amplifier is oscillating under specific conditions (temperature also).
It might oscillate inband as well as outside the band. Generally you want to avoid any oscillations that might affect your complete system.

The amplifier would be unconditionally stable if S11 and S22 <0 dB for the band no matter how you match the in- or output.

Certainly S21 will vary depending on the input and output match. You will get the available power gain when your amplifier sees the perfect load on the input and output. You will still need a matching circuit to match that load to 50OHMs for example. You can measure the available gain with a load-pull setup for example that uses tuners for the input and output.

harkonnen

It might be best if you could get the S parameters from the manufacturer, or the actually measured ones for that exact one if you want to build only one device, and simulate the circuit. Try to get rid of the instabilities with the correctly chosen matching networks. Quite often the actual S parameters differ from manufacturing lot to lot, so it's best to measure them if possible. They may also change a bit after soldering.

But usually the sad/hilarious fact is that when you want to design a good power amplifier, you'll end up with an oscillator, and vice versa...

That's why you may need to tune the matching networks with additional, soldered stubs, using a network analyzer and power meter.

Hobbie