I had the request to repair an ACOM 1200S LDMOS based HF power amplifier, capable of 1k2 (full scale), nominal output 1k.  As I have repaired in the past without  troubles my SPE 1k3 FA PA, I initially expected that this job would be in line with previous experience, taking maybe one hour time... but this was a  'big misjudgement' !

This PA was no longer capable of outputting more than 700W, even with 50W input, as per owner's report. - where in the past only 18 to 20W was enough to drive it 'end scale'.

In the ACOM the BLF188XR LDMOS measured on one gate 0,3V , on the other gate 0,99 V, with zero idle current  (as per PA service menu, which indicated for both gates 1,07V). Remark that both gates will always indicate same value, as there is only one line to monitoring circuit ...

ACOM 1200s service menu, with idle current enabled

One gate measured to earth (PA not powered) R=82 Ohm, the other one (where 0,3V was present) R = 20 Ohm.. So, according to my experience & past repairs, I suspected one of the gates was defective and 'short circuited' to ground, normal gate voltage on BLF188XR should be around 1.9 V to produce 0,9 A idle current (Idq). No other defects (like burned components) to see. Typical cause for failure of this type (gate 'burnt' ) is overdrive of input ... 

PA module top view

The PA module within ACOM1200S exists in several versions,  the version I had to cope with was the most complex : it consists of 2 decks of PCB's, mounted one above the other. The top deck being the control, biasing & protection circuits, plus RF output transformers (PCB in one piece- version PA1 F2 V 4 may 2018), affixed by a multitude of screws and two connectors to  the deck underneath (a couple of PCB's screwed to the copper heatspreader)  responsible for LDMOS input circuits, additional protections circuits  etc (2 board connected to LDMOS version PA1 F1.1 V 4 may 2018 and   PA1 F1.2 V 4 may 2018).

PA module top PCB

PA module bottom PCB's, with LDMOS in the middle

Last but not least, the BLF188XR was soldered to the copper heat spreader .....  ARGGGH... too bad :o(

I contacted the service department of ACOM who was very helpful and reactive, hereby I proposed to replace the BLF188XR (capable nominally of 1k2 CW @ 50V)  by a much more rugged MRFX1k80, capable of 1k9 @ 65V, this suggestion was accepted by ACOM as being feasible. As far as I have experienced, on HF bands (including 6m) these LDMOS are quite equivalent, except on their gate voltage required to operate in class B.  The BLF required 1.9V, where the MRFX series are more around 2.6 to 2.9 V. 

Instead of sending the complete PA, or at least the PA module, to ACOM for repairs it was decided to perform the job myself ... despite knowing this would not be an easy task like the SPE ! I had built so far several LDMOS PA's where I soldered the device to the base plate, but never unsoldered and resoldered one ...

First, the complete PA module/pallet had to be removed from PA, then stripped of completely of all PCB's till the defective LDMOs was the only part remaining on board.

For heating the copper heat spreader, I used a thermostatically regulated electric stove. To set the correct temperature, I used an infrared 'pistol' digital thermometer, but attention : this will NOT work with the highly reflective copper surface.... You need to measure the temperature of the stove round surface (black), then put the copper heat spreader on it when appropriate, and wait ...

Heating the copper heatspreader to remove defective LDMOS

Once the defective LDMOS has been removed, it is imperative to clean very well the area where the new LDMOS will be seated, hereby removing any trace of the old solder - as we are not sure about it's melting point. - so the surface must me absolutely flat. Wipe off the old solder when liquid, let it cool down, then carefully sand the LDMOS contact surface with very fine waterproof sanding paper (minimum P1000 type). For the removal and cleaning operation,  a higher temperature than I expected was required, it must have been around 300 �C or so ... the result was that the copper surface became dull by oxidation - but this is only of 'cosmetic' importance.

Cleaning the copper plate

Now, before installing the new LDMOS, first measure resistance between both gates & drains to the source : all should be 'infinite'. After the solder operation, measure them again : they all still should be 'infinite' ...   else for sure the LDMOS is fried :o(

IMPORTANT : then check if the two PCB's will fit with the new LDMOS ! If the gates / drain terminals are too long, it is now time to shorten them by cutting 1 or 2 mm off with scissors ... because once the LDMOS will be soldered in place, the only alternative will be a DREMEL tool ... like I had to find out.

Apply some solder flux to copper plate and LDMOS contact surfaces, and gently wipe off any excess, but try to keep the surface 'wet'. Apply special low temperature solder paste (melting @ 200�C) on the entire LDMOS contact surface, approx.  0,3 mm thick , evenly spread (I used a cutter knife for this).

Have the electric stove heated up to 220 .. 230* C, and when temperature is OK, lay the copper spreader on it. I used two long M3 screws as 'guides' to keep the LDMOS positioned, else it will drift off it's position on the liquid solder ...  and you risk that solder goes into the 2 screw holes. After 10 minutes or so, you can put the LDMOS in place, hereby using tweezers - the solder paste will almost immediately melt and become liquid, with the LDMOS 'drifting' on it...   at this stage, move the LDMOS back and forth laterally a couple of times, between the positioning screws, to remove any air bubbles possibly trapped under it.

New LDMOS soldered in place, between the positioning screws

Now promptly remove the copper plate and  put it on some cooling surface (like a flat stone). Immediately, position the LDMOS well in the middle between the 2 screws, and press it's body firmly against the copper plate (pushing on the ceramic casing with some tweezers), till the solder becomes solid again ( 1 or 2 minutes more of patience ...).  This whole procedure will avoid as much as possible heat stress exposure to the LDMOS.

Let everything cool down, and measure the gates & drains resistances to the copper plate ... if all OK, the LDMOS should have survived the operation  ;o)

Mount again the copper heat spreader on the aluminum heat sink, but as it's surface is not very plane, I advise to apply some heat conducting paste on the 'hot spot' under the LDMOS area - spread it well and evenly over a surface of about 5 x 5 cm.  An old credit card is the ideal tool for this ;o)

Then reassemble the PCB's on the pallet, taking care that the two interconnection  terminal contacts are well positioned for mating. After re-assembly,  measure the resistance of gates to GND, they should both read same value of 450 ... 500 Ohm about.

For the MRFX, as already mentioned, it is required to raise the gate voltage to induce some idle current.  Some components need to be replaced for this on the protection PCB (as per ACOM instructions).

ATTENTION : the trimmer to adjust the gate voltage/idle current increases current by rotating  COUNTER CLOCKWISE. So, at the start of procedure, set if FULLY CLOCKWISE - the gate voltage should be about 1 V. 

It is now time to apply 50V to the pallet and check that LDMOS is holding stand ...  As a precaution,  first temporarily insert a 'safety' resistor of about 10 Ohm / 5W or 10W in the + 50V line to pallet....  In worst case, at power up, you will have some smoke, but no 'big bang', and LDMOS cannot be destroyed should the gate voltage be excessive for some reason.

Using the amp service menu and  enabling the gate voltage, check the idle current, while  adjusting very slowly the trimmer ... at some point, it will increase... good ! ...  Then the amp will go into error, as the 50V  supply will drop because of the 'safety' resistor in line.

At that point, reduce a little bit the trimmer setting, remove the 'safety' resistor, and re- adjust again  till reaching 0,9 A (with PA cold).

Then it is time for first RF checks ... starting with low input power, perhaps 1 or 5 W,...  Full power (1.200W)  should be obtained with 15 ... 20W input power about.

PA back alive & kicking !

If all OK, let the PA run and warm up till 80�C, then adjust the compensation circuit for Idle current (PA 'hot status') of 0,8 or 0,9 A.

Finally,  adjust all protections as per ACOM's instructions.

In total, expect this job to last a day, taking into account the time for cooling down the pallet, testing the amp @ 80�C etc...

As a matter of facts, this repair can only be carried out by an experienced technician, with all the special material on hand, and with the support of the ACOM service department  - which was very efficient and interactive in communication.

Final thoughts and conclusions ...

This well designed ACOM PA is fitted with many active protection features and internal devices avoiding damage by excessive voltage, current  etc.   But some transceivers produce a short burst of high power at key down, before going into QRP in order to drive this PA ...  the time of their ALC to react.... a not so healthy situation for your PA ! LDMOS devices can withstand some mistreatments, but not indefinitely ...they are less foregiving than tube PA's !  Unlike the SPE PA, this ACOM unfortuntely doesn't provide a RF power feedback to transceiver by the ALC , in order to avoid any overdriving or reducing power when excessive SWR is present.

Regarding power dissipation, another major possible cause of LDMOS failures, it must be taken into account that LDMOS are mostly efficient (or otherwise said : dissipate  less heat ) when being driven into saturation. When driven into saturation, because of non linearity, the IMD products will increase in case of complex modulations, but this is not the case with 'single tone' modulations, like digital signals. So, for example, the LDMOS will be operating much more in comfort  outputting 750W RTTY while being saturated under 35V supply than with 50V supply, running at least 15% more efficiently. In my opinion it is a pity that the ACOM doesn't offer, unlike the SPE, different output power settings (LOW / MID / HIGH) by adjusting the supply voltage. 

Good luck with the repairs !