Success! The charger is repaired and has charged a battery. Just one battery charged so far and time will tell how reliable the repair is, but I am fairly confident.
As I suspected in my last post, the final repair was just the replacement of D3/D9 and Q2. I used the MBR20200CT dual schottky diode to replace both D3 and D9. See the attached diagram but this device is a diode pair in a common cathode arrangement in a TO-220 package. Each diode has a current rating of 20A so much higher than the original 5A rating for each of the MBR5200 diodes. Just one half of the pair in the MBR20200CT should handle the job but the holes are in the board so why not make use of both. Yes I know it's not good practice to use diodes in parallel but there is no harm, there may be some current sharing because the two diodes are probably on the one silicon die and should have very similar characteristics, and the component is physically more stable using all 3 legs.
I also added a small piece of aluminium as a heatsink as it seems these diodes generate more heat that anything else on the board. The average current through these diodes is much less than the 5A rating of the original diodes but I suspect that the heat generated by the switching process is what tends to kill these diodes so anything we can do to keep them cool is probably a good idea.
This charger failed during the summer months in Australia. It was located in a garage, in the shade, but it may have been quite warm in there when it failed. I wonder whether this is a factor in the failure of these chargers?
A couple more suggestions before I finish:
- Despite being careful, I somehow damaged the track between R12 and the gate of Q2 when I had unsoldered Q2. This track runs underneath Q2. Q2 was so cracked that it came off in 3 pieces and yet the track was broken. I am not usually heavy handed so be gentle when removing the dead MOSFET. I used a single stand of wire from some multi-strand hookup wire to bridge the gap.
- If you use a heatsink like this be careful of the ferrite bead FB1. Speaking from experience, it can short out against the heatsink which is connected to the tab of the diode package. Luckily no damage done!
- Not shown in these photos but I glued a small brace from the heatsink down to the board to make the whole thing more stable (and to make sure it never makes contact with FB1 again!!)
So in the end, once I had the replacement parts, this was quite a straight forward repair.