No announcement yet.

Dynasty 200DX Output Inverter IGBT and PC6 (IGBT Drive and Snubber) Repair

  • Filter
  • Time
  • Show
Clear All
new posts

  • Dynasty 200DX Output Inverter IGBT and PC6 (IGBT Drive and Snubber) Repair

    I finally found a dynasty I could afford! It didn't work, of course, but that's to be expected of any equipment I can afford.

    Last time I posted a repair process, a couple people commented it was helpful, and I couldn't find a single forum thread about repairing this part of the machine at all, so I figure I'll do a quick writeup for the next person who searches.

    Dynasty 200DX, serial LC623385. Used but not beat up. A crapload of hours and cycles according to the counters, probably used for production.

    The seller's description was that the unit had a help 10 code. I couldn't reproduce this. I did, however, notice that it put out 00.0 volts, despite powering up properly. I did find the LED board loose (screws missing) when I opened the unit, so it's possible it was bumping into the control board and causing an intermittent fault, but I don't think that's in any way related to what I did find wrong, and much more likely to have been a previous shop's lazy re-assembly after a failed diagnosis attempt.

    So, now that I had some idea what the problem was (not a blown skiip, unlike just about every other one I've seen for sale), time to start diagnosing.

    First step, find a service manual. Which is always harder than it should be! I finally found one, which had some of the diagnostic checklists, but no schematics. I wish it had schematics. Schematics would make everything easier. If anyone has one with schematics, I'd love a copy, for next time it needs fixing...

    I ran through the pre-power checklist. I'd already powered it up, of course - the seller's description of it giving a non-terminal help code meant it was probably safe to do so, and I didn't notice any obvious damage (no black marks next to the skiip module, for example) with the cover off. Since it powered up and did everything except actually put out power, I expected the boost circuitry and such to be intact, and it all measured good. I also measured the bus voltages and they were spot on. You should measure them too, even if you haven't powered it on - make sure they're near zero before poking anything else! I figured I'd run through the whole checklist anyway, my first time working on one and all.

    Everything on the pre-power checklist measured good until I got to the bottom board, PC6, IGBT driver and snubber. R22, supposed to measure 10K ohms, read a dead short. Progress!

    Ran through the rest of the checklist, didn't find any other issues. Back to R22.

    R22 is obviously in parallel with D9. Far more likely D9 is shorted than R22 is. Sure enough, lift one leg of D9, and it's the problem. Time to see what these parts do. Start tracing traces, draw a little schematic of this corner of the board. I used a postit.

    Click image for larger version

Name:	gatedriver01.jpg
Views:	295
Size:	229.3 KB
ID:	608077

    Ok, that sure looks like a gate driver. Look up the part number of the opto, and sure enough it's a high-current gate driver.

    I also notice that R28, which seems to be intended as a fuse for the isolated 20V power supply feeding the low-side gate drivers, is rather toasted. So we have a number of bad parts here.

    Now, experience with power semiconductors has taught me that any time you find a damaged gate driver, current gain stage, etc, it's because the biggest most expensive part connected to it fried first. But the big IGBTs passed the tests earlier in the checklist, and the checklist says to only replace PC6 at this point, not the IGBTs! This seems really, really fishy to me. By far the most likely thing to take out the gate driver is a blown gate.

    So, I built a trivial IGBT tester. A 9V battery, two 470 ohm resistors, an LED, and several little jumper wires. From the positive of the battery, go to one of the resistors, then from that resistor to the positive of the LED. Connect the negative of the LED to the collector of the IGBT you wish to test. There's two separate IGBTs in each package, so you have C1 E1 G1, and C2 E2 G2. C2 and E1 share a terminal. Let's say we're testing #1, so we'll connect the negative of the LED to the C1 terminal. Next, connect a jumper from the emitter of the same IGBT (in this case, E1) to the negative of the battery. Last, use a jumper to go from the associated gate terminal (in this case G1), which will be one of the small pins at the bottom, not a big bolt terminal, to the second resistor. Hrmm, that's a messy description, hold on a second while I grab another postit... Touch the other end of the resistor to either the positive or negative of the battery. When you touch it to the positive of the battery, the LED should turn on, and should stay on until you touch the resistor to the negative of the battery, which should turn it off. It will likely also go off after a few seconds if you don't touch the resistor to either side of the battery, but it should stay on at least a moment. If it does this, the IGBT is likely good. If it doesn't, the IGBT is bad. Or you got your wires wrong or one popped off when you weren't looking... Also, you'll need to make sure to lift up the bus bar between the two IGBTs, so you can accurately test one at a time. Now test the other half of the module - remember there's two separate IGBTs in each module.

    Click image for larger version

Name:	igbttester01.jpg
Views:	297
Size:	246.0 KB
ID:	608081

    So, tested both modules, and sure enough, one of the four IGBTs doesn't do squat. And it's the one with the blown gate driver components. Big surprise there, right?

    The pre-power checklist is, IMHO, wrong. If any of the resistance measurements are off, you should assume the associated IGBT module is blown unless you can prove otherwise. Replacing just PC6 will quite possibly result in immediate failure of the new board, and at $570, that would be an expensive mistake.

    So, now onto the opto. With everything connected to it fried, it's probably blown too. I used the diode check scale on my meter and found it didn't measure the same as the other three, reading completely open on some pins that the others had a diode drop on, but you can probably just assume it's bad too. D9 is a 1N4748, a 22V Zener, not a regular diode. Put in a mouser order for the little parts, and found an IGBT on ebay. Time to hurry up and wait. I was really tempted to change out all the green front panel LEDs for pink ones while waiting, but fortunately my parts got here first.

    I also measured every easily-measurable semiconductor on the rest of the board, and they all measured OK, and I couldn't find any other damage. I did notice one of the 100uf 25v electrolytics, not one of the ones connected to the part of the board with the other failures, was domed slightly on top, indicating over-voltage, over-ripple, over-temperature, or over-age. I replaced it, and later verified the power to it was the correct voltage and free of excessive ripple, so I'm not sure what caused it to start failing. It could have just been under-specced and started slowly bulging from how many hours the machine has on it, or, since it was above one of the IGBTs, it could have been overheated. I also double-checked the isolated 20V supplies on the power board, and everything was good.

    R28 was crispy and unreadable, so I assumed it matched the also-likely-a-fuse resistor on the other side of the plug, and got the same value, 2.74 ohms.

    Yay, parts! The IGBT replacement is pretty simple. Be sure to clean off all the old thermal goop from the tunnel heatsink and apply a thin, even layer of goop to the new module before installing it. And nothing but basic through-hole rework for the parts on PC6. Tested the new IGBT for the heck of it, and both halves tested good. Time to put it all back together. I decided goodntight was the proper torque for everything, not having a torque screwdriver. Also checked the output connectors (they were tight), the skiip module clamp (also already tight), and anything else that looked like it needed to be confirmed tight. The zip ties on the snubber inductor were broken, allowing it to flop around, so I replaced them with new ones. I also added another two globs of hot glue to the push-on connectors for the small pins on the IGBTs, as they didn't seem secure, and both were at an angle when I took it apart. Apparently newer units replaced the globs of hot glue with a plastic clip, which seems like a far superior design.

    Plugged it in, hit the power switch, and... 6.7V. ****it, there's still something wrong! GRR! [email protected]#$!! Ohhhhhhh, yeah, these have low OCV mode, don't they? ok, no need to panic, that's probably normal. Unplugged it, put the rest of the unit back together, grabbed torch, ground, filler, argon, and some random metal, took it outside for a test run.

    Yay! It makes sparks! It actually makes too big of sparks, as it turns out the foot pedal I got with something a while back and never tested is bad, and it both only runs at full current (whatever the machine is set to) and only turns off if you slam your heel on it as hard as you can repeatedly. But the welder works. The other kind of welder (i.e. me) doesn't work quite as well, having never touched a TIG torch before, but hey, I made some things that don't look like entirely total crap after a minute of trying, with no current control. I've since fixed the foot pedal, but haven't had a chance to try it out yet, since my workspace is the driveway, and it's been wet in the mornings, and dark when I get home from work.

    Click image for larger version

Name:	tigwelds01.jpg
Views:	225
Size:	303.3 KB
ID:	608079

    The gas valve is rather leaky, will need to clean/repair/replace it, but that's a pretty minor project. I has a tig welder now! *does happy wolfy dance* Now to practice, practice, practice...

    Hopefully this helps anyone else who come across a unit with no weld output and PC6 or output inverter IGBT problems!

    Attached Files

  • #2
    And if anyone else has a RFC-14 foot control that doesn't work, it's probably the same problem mine had...

    Click image for larger version

Name:	footcontrolbrush01.jpg
Views:	260
Size:	253.1 KB
ID:	608084

    Just the brush stuck in the brush holder. Removed the brush holder (there's a slightly-hidden allen screw), unscrewed the back cap, knocked the brush out the back with a flat punch, cleaned up the holder and the brush with sandpaper, put back together. I also loosened the screws holding the contactor microswitch in place and moved it slightly further out, to fix it not shutting off reliably. I haven't actually welded with it yet, but will hopefully be able to find time this weekend.

    I think I really don't like this foot control. I have a bad back, and the extra six inches of height is really unpleasant. It also pivots in the wrong spot for my feet. And I plan on doing mostly portable welding, where a foot control might be awkward in general, especially a ten-pound one.

    I'm torn between just buying a SSC foot control now, which looks much smaller and pivots at the end instead of the middle, or springing for a tig button. I need to learn how to properly hold a tig torch before a button control would work well, or not? It also costs a lot of money. Of course, the SSC foot control isn't cheap either. Can we do wanted posts anywhere on these forums? Anyone have a SSC they want to sell cheap? Or a tig button? Or maybe one of those CK belt controls? Although I like the idea of the pressure-sensitive button a lot better...
    Attached Files