A defective nice XPA125B (100W out) was brought in by a friend OM ... with 'no power out' as reported trouble.
Quickly found out that one of the two RF final transistors (in push-pull configuration) was dead : in TX mode very low voltage on gate (should be about 4 V) and no idle current induced ... As there are no other active devices in the amplifier chain, for sure this was the the cause of problems. On internet, could not find schematic diagrams, but a kind of basic 'maintenance manual' (download here), not of big help.
Original devices are RD70HHF1 (see datasheet here) MOS FET type transistor specifically designed for HF High power amplifiers applications, good for 70W out max, thus a pair in push-pull in theory good for 140W out maximum - that gives 40W 'headroom'. As my components suppler didn't have this in stock, he proposed me the RD100HHF1 as replacement - the 100W version, apparently used as well in some YAESU gear (FT-450, FT-991A, FT-DX101D, ...). It has very similar characteristics (see datasheet here), and is in fact a 'drop in' replacement, at about same cost as the 70W version (about 40,-EUR/pce).
In this case it is off course required to replace the 2 transistors to have the push-pull stage in balance.
The upgraded configuration would be good for in theory 200W out (providing 100W headroom vs 40W) and the big advantage of the 100W version is the fact that it can accept up to 12,5W drive instead of only 5W for the 70W version. As we know, overdriving these devices is the prime cause of failure ...
Exchange of both transistors is an easy job, during operation one leg of a feedback resistor is to be unsoldered to allow access to device and the temperature sensing diode somewhat bent out of the way to allow extraction of MOS FET.
Adjustment of IDq (quiescent current) is facilitated by playing with the 2 jumpers providing gate voltage in TX mode. Turn both multi turn trimmers completely anticlockwise to start. Measure (in TX) the idle current of the PA, it should be about 0,3A. Now by swapping the 2 jumpers one by one, adjust the current with the relevant trimmer to have an increase for each of 1 A (= the IDq recommended by manufacturer Mitsubishi) - so consumption will increase till 1.3A for each adjustment. The gate voltage should be around 4V at that time. In the end, put the 2 jumpers back in place.
Then the final test : gradually apply RF at input, and check power out on dummy load - with 2 or 3W carrier input on 20m band, 100W is easily obtained on output.
Good luck with repairs !