After years of faithful service, I noticed my SPE 1.3k EXPERT PA produced from time to time some smell of a 'burning resistor', later some 'crackling' noise, typical of RF flashover situations. Also, in the low and mid power settings, it refused from time to time to amplify (despite going into TX mode), unless I switched to high power mode, and reverted to mid or low power ... very strange !
Anyway, when I opened the PA top cover and the inner cover plate, I immediately ascertained what was going wrong: the LDMOS device was dying ... despite the fact that I did not mistreat the unit (always driving it in RTTY only till 750W about in mid power setting, never reaching the 75°C temperature limit ...).
It seems that one of the LDMOS drain leads was literally 'burning up' :o( In fact inside these LDMOS devices there are many tiny leads from the outer connections to the substrate, and if they start to burn up one by one, you will have a degradation in a chain reaction. See pictures below of the MRFE6VP61K25H device pulled from the amplifier... you can see through the drain terminal !
By chance, I had a MRFX1K80H (1k8 version, see specs sheet here) on hand, intended as spare for my homebrew 2m PA - so I decided to give it a try with this version. Quite an easy job, performed in less than 1 hour, as the whole PCB can remain in place ! Apply a minimum of heat paste on the LDMOS, well smeared out along it's surface. Be sure the copper heat spreader is clean (use alcohol and a cotton bud) and without any dust particles before putting the new LDMOS in place. Do not over tighten the standoffs used to bolt it in place.
Take care of adjusting the trimpot on the PCB (close to LDMOS) completely CCW before applying any power. This trimmer is the adjustment of idle current through LDMOS (IDq). Then, while in TX mode, with NO DRIVE at all, watch the current consumption (on the display of PA, it should be 0,1 Amp) and by VERY SLOWLY turning the trimpot clockwise, set it to 1.6 A. Watch out, because beyond this value, it increases quickly (with as final result sure a 'BIG BANG').
The MRFX1K80H is designed for output power of 1k8 continuous @ a supply voltage of 65V, so in the SPE amp (max 1,3 kW @ 50v ) it feels very 'comfortable'. In my howebrew PA for 2m I had to replace as well the 1k25 version (which went SK) by the 1k8 type, and found that this version was somewhat 'lazier' (less gain), because circuit components need some fine-tuning when operating at this high frequency. But on HF bands, like in the SPE, there is no difference, 1k8 version is considered as a 'plug&play' replacement, and I experienced that you will more easily reach 1k3 or 1k4 peak. In fact, now I usually work in 'mid power' setting, and it delivers 1 to 1,2 kW with very limited heat dissipation, where with former 1k25 LDMOS I had to run in high power mode. To minimize stress on the LDMOS, always try to run on LOW or MID power and apply enough drive to operate PA in saturation (=ALC limiting output), this will ensure best efficiency (= minimal heat dissipation) .
The following graph - provided by NXP - clearly shows that the MRFX1K80H is more efficient running on 50v @ 1k2 output power on 27 MHz than the MRFE6VP61k25 version, without retuning of any components:
Two tone test with bare IC-7600, ALC just coming up, Po 15W PEP
Two tone test with SPE 1.3k PA, LOW power setting, 500W PEP out
Two tone test with SPE 1.3k PA, LOW power setting, 800W PEP out
Two tone test with SPE 1.3k PA, HIGHpower setting, 1000W PEP out
This is the third MRFE6VP61K25H device that I am sending into semiconductor's heaven ... I have the impression that, since these very first high power devices went into production, a lot of research and improvement have been made by manufacturers to make them more reliable, resulting in the 1k5 and 1k8 versions.
Good luck with the repairs !