Welder Tig AC/DC blown house fuses, shorted IGBT

[florkk]

Hello, i have strange problem, i have my welder about 3 years, its ac/dc tig , few days ago, when i weld in 160A ac range, tig blown house fuses, fuse was 10A, after third shut down time, when i go to turn on automatic fuse, welder try to starts, light all indicators, and after 1 sec blown fuse again. i was know that something broken inside. i open welder, and after few minutes of searching i sound two shorted igbt, in some kind of power supply board. in total the are 4 igbt, connected to 4 individual wires, going to strange looking transformer. The igbts are IKW50N60T. i bought 2 new for replace, replaced shorted one, and welder work again, i also replace home fuses from 10A to 16A with delay. i use welder for about 1h, and on my mistake when i was finishing weld, i touch tungsten to base metal( where i have connected ground lead) and it tungsten stick for 1s. house fuse blown again, and again when i turn it on, and turn on welder, it again blown fuses, so again it seems like shorted igbt, i open again welder, and again the same 2 igbt that i replace got short. right now i order again new set of IKW50N60T, but why its again blow, any ideas? its any way to protect igbs with some diode or something? igbt gate is controlled by transformer, and few resistor and diodes.

[florkk]

PS. IGBT after damage have all 3 legs shorted to each other, so G E C are shorted

[TheMG]

Sounds like there may be a deeper problem that is causing the IGBTs to fail.

Welder should be able to handle short-circuiting the output without any damage, it should current-limit. So I would suspect maybe there's some sort of issue in the control circuits responsible for limiting the output current, causing the IGBTs to operate beyond their maximum ratings and failing.

[florkk]

again i replace igbt and after 30min ac welding its short again, i check temperature of this two transisors and they were cold. they are k50t60 600v 80A , gate is controlled by one small transformer two diodes and few resistors. maybe someone know what can damage this igbt, i think its not heat, and its not over voltage and current. without scope i can measure anything? idk why but im still thinking about gate voltage, or something connected with gate

[MadTux]

Sure you got original IGBTs?
Chinese bastards especially like to fake power semiconductors (experienced that myself), since they get best margins there.

Apart from that, perhaps find a cheap high voltage differential probe (7A22 or 7A13 are probably the cheapest, if you get a Tek7000 mainframe) and see, wether the gate drive signal is good/the same on all 4 IGBTs. And perhaps find some real heatsinks for them, looks like the usual chinesium crap welder, with heatsinks made from whatever scrap there was available.

[transistor12]

Chinese bastards !

[tokenpete]

Here is a video to help you diagnose. You probably have a blown zener on the other side of the board. You need to use a 18V and 8.2V zenner in opposite polarity to protect the gate on the igbt.

[fenugrec]

+1 on checking gate protection diodes - possibly 2 zeners back to back. Find their voltage ratings and replace, they are cheap, not worth testing. If your IGBTs shorted G-E-C then some high voltages may have fed through the gate and damaged those diodes.

[FIXITNOW2003]

what make this welder @? may have circuit

[tokenpete]

There are many brands globaly using (used now obsolete) the same design sort of with minor changes, they come from NANTONG FREE ELECTROTECHNOLOGY

In the UK R-Tech welding had these, looking at the OP's images I think his is an Everlast TIG judging by the front panel feet

I have attached

1. The schematic for the power section
2. The schematic for the Main Control Board
3. The parts listing as referenced in the schematics
4. A generic manual for this series of machine

I have been fixing these machines for 20 years, they are like my babies 

[tmazz]

@Tokenpete Your a legend mate! I do think it's a copy of an Everlast welder as I have seen a picture somewhere of the output section of the control board which was exactly the same!

[tokenpete]

  You are most welcome

[tmazz]

@tokenpete attached is the secondary inverter board schematic for said welder.

[tokenpete]

Great work. I have plenty of these boards as new pieces. Mostly the caps die on them after not being in stoarge for a while.

[tmazz]

Cheap caps are always a problem.

I would like the AC frequency range to be a bit higher (up to 200hz) so I might lower the value of resistor R2, on secondary board, and then change the pot to a higher value so that I still get the very low frequency range. The balance is a bit ordinary as I don't think it is in spec with the dial. My Tungsten tends to ball up a bit too easy when I push the current up and dial set to minimum, i.e., 30% EP.  I'll scope it out and see what is going on. Being able to adjust a bit lower than 30% EP would be nice anyway.

[tokenpete]

Normally with torch in negative you can crank 200 amps with 30% balance and a 2.4 tungsten and it won't ball up.

[tmazz]

A simulation with ltspice of the initial frequency and balance circuit shows a how cheap and nasty the design is. Not even a square wave but sort of a sawtooth where the wave is simply pushed up or down around ground (O volts) to adjust your EP and EN.

[tokenpete]

You are simulating a circuit with it's load disconnected. Add the IGBT and accompanying gate circuitry and you will see a square wave.

[tmazz]

Ah, my bad your right!

[Njk]

I have been fixing these machines for 20 years, they are like my babies 

In that case you must be familiar with the Jasic machines, because it's a big manufacturer and the products are distributed worldwide under many different brands. But sometimes, for the same product, installed options are depending on the region. Actually, the option set is up to the distributor.

Consider Jasic W212 model (known in NZ as XA-TIG200PRZ, see the image). Similar machines of 160A and 180A also exists. It's a DC-only welder that has digital and analog RC interfaces by the design. For some reason, only the models for NA, Australian, and NZ markets are actually equipped with the analog interface. While in all European models (even that of Wilkinson in the UK, AFAIK), all the related components are not populated on the PCB. So the users wishing to get that option have to install the missing components by themselves (and perhaps to adjust the firmware). With that purpose, I'd prepared the wiring diagram of the interface. Although it's perfectly possible to deduce the missing values from the schematics, it would be nice to know the original ones for reference. This is where the modern telecom can be helpful, isn't it?

P.S. Sorry for drifting off-topic but some time ago I'd started a dedicated thread and no one responded