Mike M0ZDZ reported that his beloved FT1000 had failed during the ARRL 10M contest last year. He thought it had good power output on other bands, but not on 10 metres. I wondered if that was a failed low pass filter or relay in the LPF board. However when I opened it up there was no damage evident to the LPF, and there was only 2 watts of output power on any band. That indicates PA or power supply failure; the PA supply remained at (I think) 31V so that wasn't the problem.
You can still download the service manual for this radio. The PA is quite different from the FT1000MP, so you do need the correct one. I didn't write down where I downloaded it from, but look for "FT1000D_serv.pdf".
The PA itself isn't easily accessible, You remove the metal "cage" over the PA and power supply then remove both heatsinks as a pair. Unfortunately there are a lot of connectorised wires to both, so you need to draw a diagram showing what went where. The PA includes a pair of driver transistors (MRF486) followed by a pair of final transistors (MRF422). The fault could be any of these 4. I desoldered the PA transistors and bent up the legs, which were east; I could then use a multimeter to check that I got a diode junction from B-E and B-C on each transistor. Those were OK. Doing the same for the PA was a lot harder because of the copper tabs being much larger, but having done that I found that one transistor just didn't have en emitter connection - the bond wire must have failed.
Mike was keen to attempt repair, so we purchased a new pair of MRF422 from Mouser electronics. removal and replacement was quite simple, using new thermal component underneath and to bond the temp sensing diodes. The alignment process is tricky, because you need to monitor the voltage across a couple of resistors (R7014 for the driver stage, R7027 for the finals). There aren't any test points provided for this. The driver alingment was simple, but the finals always had excessive bias current (with a measured voltage of 30kmV minimum, whereas the target id 6mV across R7027). The bias is set by a series combination of a fixed resistor R7025 and variable one VR7002. The manual says R7025 should be 150R, but it was actually 100R; I had to add a further 680R series resistor to get the bias low enough. I've attributed that to the new Mouser transistors (from a second source supplier) having higher gain than the originals.
There are good audio quality reports after doing this. Strictly I should get a 2 tone oscillator and re-check linearity using a spectrum analyser but so far we've settled for checking for signals above and below the operating frequency.