This all looked fine, all of the 5 FET's where able to set at the desired idle_current. Before building the RF-deck, one important thing needs to be done: Exchange the last hybrid_coupler in the final-stage for a more capable version. There are a lot of options in the market as well as HAM-solutions like W6PQL can deliver. For the time being, not sure whether the PA could work at 2320MHz, I choose the simplest solution and soldered amn Anaren Hybrid_coupler instead of the smaller coupler. The Anaren coupler I selected is pin-compatable to the existing coupler, however can withstand 200W CW.
The soldering of the hybrid_coupler is not easy...Just like soldering FET's to a copper heat-spreader, one needs to bring heat into the baseplate of this Powerwave amplifier. I used an electric stove and carefully exchanged the coupler for a new one Xinger XC2500A-03S sourced from Mouser:

Time to build the PCBA into a shielded enclosure and start testing the perfromance!
I was lucky to be offered by Wim, ON5KTO a fitting aluminum housing which already had been used for exactly the same Powerwave amplifier. Looking at the remains I found inside this housing, the former owner was unlucky...several burned parts as well as parts taken from the remainng PCB where found. I simply exchanged it by my PCBA and started testing.
First thing to perform was a frequency-sweep. By reading the French articles carefully I already found out the first isolator in the PCBA must be bypassed, for that reason the input semi-rigid is mounted to the coupling_capacitor at the gate of the driver_stage.
The frequency_sweep was initilally not very promising, however a trimmer_capacitor at the output of the driver_FET was all it needed to be adjusted. After this the frequency_response was flat to at least 2320MHz!

The nice thing about there Powerwave amplifier PCBA's is the seperate supply of Vcc for all the five stages. This brought me to the idea to have a 0.01 Ohm in series to each of the supplies, and have a 5-position rotary-switch to select to measure across the 0.01 Ohm resistor. The common from the rotary-switch is connected to the banana-plug chassisparts so we have a 10mV/A reading during tests that will follow.
The multi-pin connector in below piture is the 28V supply input. Half of all the pins are positive, the other half is negative (GND).

Now it is time to finish the mechanics of the PA like sufficient cooling and a dedicated 28V power_supply that can generate enough current. According the the 'French', we need to count for at least 25A or so.
To have sufficient cooling, I mounted in total four 60x60mm fans, two are feeding air into the heatsink, two on the opposite side suck the warm air out. The heatsink is covered by an aluminum tunnel to guide the air through.
The power_supply is slightly moddified by incorporating the wellnknown chinese volt/ampere meter to indicate up to 50V and 100A.

Meanwhile also the SG-Labs 13cm transverter had arrived and tested to deliver over 2W output power. With some care the transverter was connected to the PA and the power at the output was slowly increased. All of a sudden, arcing at the output of the final hybrid_coupler occurred! It was NOT the hybrid_coupler itself, however it turned out arcing from the 50 Ohm trace to GND. The same happened to the French team if you go over the document in which they posted all the measurements: French-team measurements

We repaired the issue by replacing the final hybrid for a fresh one, but removed the SMD pad from the bottom of the hybrid! Also we removed all the copper from the top-side of the PCB and soldered the output semi-rigis directly to the output of the hybrid_coupler. The picture below is a close_up, in practice the RF-path to GND is still within acceptable tolerances...
Measurements after this repair indicated the power is there, I stopped pushing at 150W RFout...but still the power increased linear to the applied input power, so more to come later!

To conclude, above is a picture of the completed PA for 13cm. The tunnel over the heatsink is visible, as well as some of the fans.

Next thing to do is to create an external Brenchline_Hybrid (W6PQL or similar) so there is much less risk to have flash-over in the PCB. One should always realise these kind of PCBA's are never intended to be pushed to the full potential of the FET's.

As an example is below the solution which is maximum safe, RF-wise, however there is also a intermediate solution possible in which 2 of the 3 output Hybrids will be replaced the Anaren Xinger XC2500A-03S, and only the final hybrid is external (W6PQL).