Electronic welding machine in short circuit

[DisasterNow]

Hi, I have an electronic welder that no longer works. It turns on regularly, the green light works and so does the fan, but I have no voltage on the output. It should have about 80 volts at no load. Visually I don't notice any burnt components, but when I power the board via variac and a light bulb in series, there are moments in which I feel the fan speed drop and the light bulb lights up slightly, indicating that there is a short circuit on the board. I checked the temperature of the components, but I can't find anything with high temperatures. Do you have any advice for finding the short circuit? The card is very similar to this one, thanks.

[DisasterNow]

I add the electrical board diagram (not 100% identical, but almost the same)

[DisasterNow]

Maybe you can see it better in the PDF (it's from an inverter of the same brand, but I'd say that the PCB is almost identical)

[coppercone2]

its something on a heat sink if you cant find it

[DisasterNow]

The bulb dissipates the short circuit, so no protection fails and perhaps not even the faulty component. There are many heat sinks, is there a way to understand if it really depends on a heat sink? The board also has thermal control, but I believe only on one of the two heatsinks with 4 IGBTs. Thank you

[coppercone2]

Ok I am saying if you think there is a short circuit, but you can't find any hot parts, chances are its a part on a heat sink, or with a heavy ground plane.

For high voltage power boards you really want a thermal camera that can point you in the right direction. Usually bad welding machines have blown parts.

My broke welder had clear signs of burn damage. Did you check perhaps to inspect the parts with a mirror to look for a cracked transistor? They often blow a corner off or something like that and look mostly OK

This one looks really simple. You could replace every single part fairly cheap, its always an option


My hunch is something bad with parallel MOSFET. I would just buy a dozen on digikey and give it a shot replacing them all lol, for 20$ + thermal grease. Chances are it might be a cheap welder with counterfeit parts (suspicious this is from amazon) and you might improve the thing. Plus, its kinda safer then probing around those. And you can probe most of the diodes too.


With parallel mosfet it might be complicated to try to determine if one of em is acting up, their like $2.95 so is it worth it ?


A good machine would have factory made IGBT bricks for this purpose. Parallel FET is alot cheaper but yeah who knows what can happen there!



I can ensure you, that probing those FET near the heat sink while its running is perilous. If you wanna try to do something there, get a fluke insulated grabber hook, turn off the machine, make sure its safe, connect the hook, measure, and repeat for every single measurement. Probing there  while its on is the risky part for this machine. Those TO packages near BIG heat sinks are recipe for short circuit trouble

[DisasterNow]

Ok I am saying if you think there is a short circuit, but you can't find any hot parts, chances are its a part on a heat sink, or with a heavy ground plane.

For high voltage power boards you really want a thermal camera that can point you in the right direction. Usually bad welding machines have blown parts.

My broke welder had clear signs of burn damage. Did you check perhaps to inspect the parts with a mirror to look for a cracked transistor? They often blow a corner off or something like that and look mostly OK

This one looks really simple. You could replace every single part fairly cheap, its always an option


My hunch is something bad with parallel MOSFET. I would just buy a dozen on digikey and give it a shot replacing them all lol, for 20$ + thermal grease. Chances are it might be a cheap welder with counterfeit parts (suspicious this is from amazon) and you might improve the thing. Plus, its kinda safer then probing around those. And you can probe most of the diodes too.


With parallel mosfet it might be complicated to try to determine if one of em is acting up, their like $2.95 so is it worth it ?


A good machine would have factory made IGBT bricks for this purpose. Parallel FET is alot cheaper but yeah who knows what can happen there!



I can ensure you, that probing those FET near the heat sink while its running is perilous. If you wanna try to do something there, get a fluke insulated grabber hook, turn off the machine, make sure its safe, connect the hook, measure, and repeat for every single measurement. Probing there  while its on is the risky part for this machine. Those TO packages near BIG heat sinks are recipe for short circuit trouble

Thanks, it's very kind of you. I'm at work, tonight I'll reflect on your advice 

[DisasterNow]

My experience troubleshooting an inverter electronic welding machine board is that of a beginner.
There are few transistors, they don't seem to be shorted, but they should be desoldered for an accurate check. Then there are some power diodes (they seem ok) and 8 IGBT (soldered onto the circuit, it is difficult to verify if they are efficient).
Of course I could replace the power components, but first I wanted to understand where the short circuit was located; It's not easy because for many minutes it doesn't absorb any extra current, and when it happens the bulb disperses it without causing further damage.
In reality the welding machine is not a super cheap model, even if it is many years old, but it has not worked much. The model is EPS energi@130 (manufacturing company is eps-inverter.com).
When it was given to me as a gift, the electrode only produced a few sparks, in fact the output is zero volts (it should be 80 DC at no load).
I carried out the voltage check on the component side only with the probe cable that detects the temperature, while I carry out the measurements on the track side.
For now I have verified that on the large filter capacitors I regularly have 310 DC; the relay works; the fan works. On the gate of 4 mosfets I have a voltage, while at the moment I don't have any voltage on the gate of the other 4 mosfets, but I have to check better. The IGTBs are IRF P 460s.
I add some photos: secondary diode

[DisasterNow]

transformer area

[DisasterNow]

8 IGBT

[DisasterNow]

hybrid logic board (covered by a transparent protection)

[DisasterNow]

Track side (I scratched the transparent protection, otherwise it wouldn't conduct on the multimeter tip)
Note the fixing of the 3 large heat sinks, with rivets...

[odemiriz]

If the welder powers on regularly (meaning green led on and no error led), primary switches - mosfets or IGBTs - are fine. You should first check output diodes with a multimeter. If there is a short there, you need to change the faulty ones. If not most likely you have problem with control board. Check the aux power for the control section.

[DisasterNow]

If the welder powers on regularly (meaning green led on and no error led), primary switches - mosfets or IGBTs - are fine. You should first check output diodes with a multimeter. If there is a short there, you need to change the faulty ones. If not most likely you have problem with control board. Check the aux power for the control section.

Yes, the welder turns on regularly, the fan turns and the green light is on; the yellow light does not come on (not even when the bulb indicates a slight short circuit). The output diodes (BYV52) verified in the diode test (on the board), would appear to be ok. I haven't yet done accurate voltage and signal tests in the control part, on the gate of 4 mosfets I have a voltage, while on the other 4 gates I have a zero voltage, but from the diagram it seems to me that it refers to the ground so maybe ok. The resistors and diodes between T4 and the gate seem ok. I don't understand what it is that doesn't allow the current to reach the output.
When the light bulb signals me the short circuit (only occasionally while the welder is running idle), I then lower the variac below 200AC and the short circuit disappears. I almost want to let him continue to see if he smokes anything, but I don't want to do any more damage. I still have to check better, but I had the feeling that by leaving the current regulator towards the minimum, the short circuit is more frequent, while if I raise the potentiometer towards the maximum (130 A), it doesn't signal any short circuit, but I have to do further tests . However, the output always remains at zero DC.

Thank you

[coppercone2]

well if the current is not reaching output that means the bridge is stuck or fried or control is fried. is the control card making a wave form into the gate of the transistors?

[DisasterNow]

well if the current is not reaching output that means the bridge is stuck or fried or control is fried. is the control card making a wave form into the gate of the transistors?

I think it's quite difficult for me to understand this fault. I have attached part of the diagram where I took some measurements. In the moments WITHOUT a short circuit, at the input of TR4 I have the right voltage (13.7v); at the output I have about 18/20v, which arrive at the gates of MF1-2-3-4. On the other output of TR4 I have the ground reference, which continued up to the gate of MF5-6-7-8.
During short moments, the output voltage from TR4 to MF1-2-3-4 alternates from 20 to 120 v!, and the input of TR4 drops from approximately 13.7 to 10 v. I also saw that the yellow LED lights up in moments of short circuit.
In the diagram I see that they distinguish between two types of mass (one has the symbol F). However, I see that there is continuity between these two masses on the PCB.
Thank you

[DisasterNow]

I'm checking all the components one by one, for now I've reached after the large capacitors and everything seems ok (in moments of short circuit they drop from 310 to 200 DC, but I think it's normal, I also have the lamp in series).
When I get to the power components (mosfets, diodes, transistors, voltage regulator), I will disassemble them from the board including heatsinks, for better control.
For each component that I desolder, I will test the board under voltage to see if the occasional short circuit will disappear, in this way I should limit where the problem lies.
Some doubts:
-I didn't understand why they fixed the large heat sinks with copper rivets, in this way, for example, the leveled voltage of 310DC that reaches the drain of 4 mosfets, we also have it on the entire heat sink (I don't understand this choice).
-I read that desoldering/soldering the mosfets could ruin them, are perhaps precautions needed in this regard? Is the classic CPU thermal paste used between the heat sinks and the MOSFET body?
-the only components that I hope do not have problems are the two chips on the hybrid board (Lm324 and TDA1060), I will have to find a way to verify their integrity carefully before changing the other components of the board...

[coppercone2]

yeah it might be risky to remove all those MOSFET if you plan on reusing them.

The keys are
1) esd protection
2) thermal protection for the chip. You can use a heat sink clip from gootwick, which is a aluminum solid alligator clip that does not stick to solder and prevents heat from going into the part.. their very useful for repairs IMO

I don't know how to test MOSFET good for the conditions they are in in the welder though. I think there still might be problems even if a simple test says its OK, like switching speed or something causing something. I think you need a fairly elaborate test fixture to really 'test' a power MOSFET for complex tasks other then switching on a heater you know?


I think you should try to map the signals out though, before you desolder heavy parts. It should be getting a oscillator waveform of some kind to trigger the transistors. You need to find it on a scope IMO


The gate driver circuit and the gate driver control circuit can be evaluated with a oscilloscope. The gate driver control circuit is some kind of oscillator that makes a low power signal square or other wave form. The gate driver is a pulse amplifier circuit that allows the little control signal to drive the big transistors fast enough. More then likely its a PWM type signal with A and Abar phases to control the bridge.




Just to give you an irrelvant annecdote:
my welding machine had a ton of problems like exploding diodes (miller dynasty 200). Some how Eventually  I found a leaky ceramic capacitor like 7 inches away. The problem stopped when I replaced it. This leaky capacitor was causing all sorts of trouble, but their circuit is rather complicated compared to this one. And yes, I did replace ALL the power parts in it, and it was still doing the same thing. So be careful, I am pretty enthusiastic about replacing power parts, but its NOT a cure all. It is possible some little piece of crap part is making it go hay wire and that all the big parts are OK.

[coppercone2]

Do you have a control signal on M_13 and M_14 on your diagram coming out of the board? That is the start of it. Then the transformer in this case is like a BALUN?, which makes the signal differential (A and A_ ) to drive the H bridge.

If one of those 4.7 Ohm resistors is off, it could mean a bad FET. Those resistors are power resistors. I think D1 is a clamp, and T2 and T3 is your gate drive amplifier. The part there is a power amplifier D45H8

DZ1 is limiting the voltage to 18V, the FET probobly does not like more then 20V on the gate.



The thing I am confused about is the D45H8... because every H-bridge I worked on, it has a power source there, it looks like that chip is just floating. The MOSFET are fairly beefy and they take alot of current to turn on and off. I am not getting this part since its connected through a transformer only . My miller had isolated 20V rails to power the driver, like an auxillary IGBT gate drive power source. The gate drive then get a signal from a MCU that it amplified to control the transistors.

I think you need someone with a little more knowledge of this circuit then me.

[DisasterNow]

I think you need someone with a little more knowledge of this circuit then me.

You also do too much Coppercone, thank you for your availability; in the next few days (I'm away for work now) I will continue the research, also taking advantage of your precious advice. If anyone else wants to help me, I will be grateful. Unfortunately I have never studied this kind of circuit (inverter), I'm doing it now step by step, I started from the side with mains voltage and will continue until the output.
I would add that it is not easy not having the precise diagram, the one in the pdf is very similar (same brand but not identical, even if in the first part of mains voltage, it is practically identical except for R39 which is not there in mine).
If the problem is not on the vertically soldered logic board (I attached the photo previously), I think I will be able to analyze the components one by one even by removing them from the board. As long as the problem is not on the LM324 IC which would not be easy to replace, both due to the wax they put on the PCB and the very close contacts.
In the meantime, thanks, I'll let you know the progress (or even failures) 

[DisasterNow]

Meanwhile, I only analyzed the part of the Tr4: 13.7v arrives at the central pin of the Tr4 (from the regulator IC1, Lm317), the other two pins (which arrive from M13 and M14 of the hybrid board, have the same potential (if I look the signals on these 3 pins, referred to ground, are 13.7 DC). The low output of Tr4 (the one towards the MF5-8 mosfets), has zero volts a sinusoidal signal, but at this moment the light bulb indicated a slight continuous short circuit, so the signal was very dirty (it increases by many volts, the regulator Ic1 does its job, coming out at 13.7). that on the 3 input pins of Tr4 I have continuity, as on M13 and M14 (after all they all have 13.7v).
In the meantime I'll take one step at a time and as you advised me I won't desolder the mosfets. If you could find where the short is coming from, but apparently it doesn't heat anything abnormally. I also found two schematics on the 15 pin hybrid board, not exactly the same layout, but close enough, I'll attach them.

Thank you

[DisasterNow]

[DisasterNow]

Just to give you an irrelvant annecdote:
my welding machine had a ton of problems like exploding diodes (miller dynasty 200). Some how Eventually  I found a leaky ceramic capacitor like 7 inches away. The problem stopped when I replaced it. This leaky capacitor was causing all sorts of trouble, but their circuit is rather complicated compared to this one. And yes, I did replace ALL the power parts in it, and it was still doing the same thing. So be careful, I am pretty enthusiastic about replacing power parts, but its NOT a cure all. It is possible some little piece of crap part is making it go hay wire and that all the big parts are OK.

As recommended, I will first map all the components, trying to understand without desoldering them which ones are ok and which ones could have problems. If the problem lies in a leaking cap, I would find it, as I desolder those and check them on the component tester. My fear is that there is some transformer with problems, there are 4 of them. For now I have only found the R 10 (output area) of suspicion, I have not desoldered it but it gives a low resistance, instead of 3, 9K, but I don't think that's the source of the problem. Thank you

[coppercone2]

A ceramic hybrid card is a pretty good heat sink. Its possible there is some power being dissipated there that is difficult to trace.


For heat sinks, if you glue thermocouples on them, and run it for a while, while the difference is small, your meter can even resolve 0.5 degrees normally. It might seem abnormal. But if you just press a thermocouple into it, it might not be stable enough of a measurement to find a small difference. Clipping it on with a alligator clip that has rubber on it might work without glue for a heat sink


and I never tried this one before, but I did get a interesting idea : what if you insulate the heat sinks with some cloth strips or something, to contain heat, or plug the fins. Then it might start showing up. If the heatsink is effectively isolated, it just a thermal reservoir. Eventually it will heat up from even a minor short. But its mechanically complicated. I imagined wrapping it up like a mummy with wool strips.. just have the fire extinguisher handy and dont leave it unattended LOL. I think it would work to increase the 'resolution' but its going to be hard to apply for some designs.

[coppercone2]

and another place you can have a nasty semi short is under power connector plastic. Its a heavy power/ground plane, if you have corrosion or whatever there, it could be getting hot UNDER the plastic, not hot enough to melt it, and going right into the ground plane.

I am sure you noticed how hard it is to desolder those power planes, aka they can sink ALOT of watts before things get interesting.

I.e. I saw this with a sorensen 30A supply. The main input connector terminal block (3 screw) was soldered to the PCB. They did not put enough solder into one of the holes, so over time it cracked and kinda wiggled and sparked. In my case it went open circuit after some bad sparking, but there was soot present. If more soot accumulated there, under the plastic connector, I could see it POSSIBLY forming a parasitic resistor of sorts, which would be well heat sinked by the huge power pins and the power planes and hidden by the connector body.

Sometimes I really miss the 1960's style design, where all the power pins had no insulators, so if there is anything going on, you can see right down to the board. When you start having plastic connector housings, you can hide things under the housing, and its damn near impossible to find unless you remove the connector ! Corrosion goes there too, if there is batteries around. The battery juice wicks up the wire and ends up depositing in/under/around the connector. Mad hard to clean unless you disassemble and scrub wash everything.