Mastech HY1803D power supply modifications

[MrSlack]

I bought a cheap Mastech HY1803D power supply to run a hot wire cutter and mini drills off. Quick Google around suggests that there are a number of weak points in this supply as expected and the schematic agrees so I'm going to replace a few bits to undo some of the cheap as I have the parts lying around. The failure points appear to be the rectifier diodes melting, the power transistor shorting out and exploding electrolytics. They're a relatively clean design otherwise, just very underrated parts. It does however seem designed to be repaired which is nice.

1. Wun Hung Lo 2n3055 as the regulator and no thermal grease. Replacing this with an ON Semiconductor 2n3055 with some proper thermal grease and mica pad. The 2n3055 doesn't do a whole lot thanks to the relay switching turns on the transformer to keep the drop voltage low. I suspect the provided one isn't up to the job as there are a lot of fake ones floating around.
2. Rectifier diodes are rated at the same current as the supply which is a little low (3A). They're also sticking right up so they don't desolder themselves. Pah. Going to rip these out and stick a 5A bridge I have floating around in bolted to the case.
3. Main smoothing cap is a purple no brand hand grenade. Replacing this with a Panasonic one. Actually will replace all of them I think with stock items I have.
4. Board traces for the output lines aren't particularly heavy so will solder some wire up and down these to keep the board cool.
5. Have tested that it's properly floating unlike some cheap and broken ones.

I'll then run it at full whack 18v 3A on a resistive load for 6 hours and see if it explodes.

Anyone else think of anything worth looking out for or replacing/modifying on these?

I'm really not expecting miracles for what I paid for it but it'd be nice if it lasted more than a year

[MrSlack]

Mods done. Can handle a solid 3A resistive load at full whack now for an hour without failure. The extracted 2n3055 base leg fell off after a touch with a solderinng iron which is a bit worrying. Shonky transistor!

Also cleaned up some naff soldering and fixed some lead stress crimes. The current sense resistors were installed by a blind drunk.

[macboy]

Good call replacing the shady '3055. No thermal compound you say. Yikes. 
The rectifiers should be OK since alternating pairs of them conduct for each cycle, so each carries only half the load. Even with derating for capacitor power supply type load, they should be OK if they are genuine parts and truly capable of 3 A average, which is dubious to begin with. OTOH I bought a pack of 100 1N5408 from China through ebay, and a torture test of 4 A continuous did not cause a failure over 24H.

[MrSlack]

It had a mica pad but that was it. And that was pretty much falling out.

Wasn't too happy with the diodes. They have thick leads and whoever did the lead forming crimped one of the leads half flat. Probably rated ok but shoddy assembly job. There was a 7812 regulator on a small heatsink as well - disassembled that and there was no thermal compound either.

It's doing a nice job for what I bought it for though. Completely standard parts so repairing it if it ever goes pop is a short job.

[JBeale]

Thanks for posting this!  My HY1803D has worked OK for over a year, and I liked the fact that it's quiet due to no fan. Today it suddenly decided to give me +35V DC output regardless of the setting of the voltage knob. That's highly suspect anyway, because this model is only supposed to go up to +18V max so where did +35V come from? Straight from the rectifier cap I assume, so maybe a problem with that 2N3055 power transistor on the rear heatsink.

EDIT: Yeah, I unsoldered it and find that the 2N3055 base-collector junction looks like a 47 ohm resistor and measures the same with either polarity. Not at all a diode-junction anymore, so at least that is one bad part.

I guess this is the schematic: https://partiallystapled.com/~gxti/trash/HY1803D.pdf

EDIT2: Next, I learned about fake 2N3055 parts. My replacement part immediately failed at 3A current. The die inside is much smaller than original part. Arghh.

EDIT3: Finally fixed it, using a real 2N3055 from Digikey (instead of the fake cheaper one from Amazon). That was the only part that was bad. It is back to normal operation now, 0..18 Vdc and 0..3.0 A output.