Attempt to repair a welder

[Laci]

A couple of months ago I got a welder from a friend of mine to repair.



It is the awelco arc250 type switch mode welder.
My friend said that he used bad electrodes,so at a moment it exploded with a cracking sound and smoke was coming out of it.



After opening the case,the first problem was obvious:the output filtering inductor blown up.



The welder is made out of two identical power supplies in parallel.



When I turned it on for the first time,the room breaker immediately popped and I've seen some sparks at the driver circuit.





After taking it completely apart,some new problems show up.The inverter section was completely melted:the switching IGBTs,their gate resistors,the filtering ceramic capacitors and even two of the output rectifiers.






(The bottom side of the driver circuit,over the dead IGBTs.)

Thankfully this was one power supply so I could test things accordingly to the working one and turned out that the driver circuit also had problems.The working driver circuit was producing a square wave at the connectors to the IGBTs,but the other one not and the voltages were lower at the damaged board.Also the power resistor of the bootstrapping IC was getting hot so the first thing I did was to remove the L6386 bootstrap IC.



The next thing I want to do is to order a new IC and test the driver circuit.Hopefully the output stage of the BJTs at the bootstrap IC is not dead.Then I should repair the inverter.I already removed the burned gate resistors and got into the first trouble.The inverter board has a thick aluminium plate on the bottom,which is connected to a large heatsink.This makes almost impossible to solder anything on the board,because even the plate conducts the heat away.The legs of the IGBTs came out by heating and pulling them strongly,but their tab(collector)is soldered under them so I can't heat them up enough to be able to remove them.What should I do?By the way here are some pictures of the other inverter,which was already repaired in a service and there were replaced some of the output rectifiers which are the same package as the IGBTs,so it's definitely possible to replace them!

[Mads Barnkob]

Good breakdown of the job ahead and you found quite a load of defects. That can make it hard to find the root cause of the failure.

From my experience you are properly looking at dirt being the culprit or improper insulation on the choke. Flash overs from dirty PCBs is not news and when you have this unit repaired, you should replace the fan with a bigger one and with a filter on it to avoid a future incident from the same problems.

That repaired PCB with a huge bodged solder bridge to replace a blown trace is a little worrying, is there a copper wire put inside that blob? It could be that is it not dimensioned correctly and overheats/blow up, the same trace does not look faulty on your failed PCB, so you should properly reinforce that trace as well.

Wash all the PCBs in luke warm water with soap, best way to clean them up, let them dry out for some days or put them in a oven at 50 degrees Celsius for some hours.

Repair the driver sections first and get them up and running before testing with the power section on.

Good luck!

[the_anomaly]

This welder has limited duty cycle (use time), which could have caused the IGBTs to burn.  Looking at the PCB, I don't see much for heat removal on the D2Pak.  You could try to add small heatsinks to the un-populated D2Pak spots if possible to help with heat removal.  Once its all working of course.

[Laci]

After a painful explosion and fail on this project,I thought of trying the repair of this welder again,because it only takes up space in my lab otherwise.

Since many components blew up last time,I replaced those,the UC3845 current mode IC,the L6386 bootstrap driver,the BJT output transistors and some infinite ohm resistors on the current sensing part.

Right now it turns on and nothing blews up,but compared to the other module(the welder is made out of two same inverters in parallel)I get a lower,1.6V at the compensation pin 1 of the UC3845,while the other has 2.7V.It's weird because this pin is the output of the error amlifier,which has an internally referenced input and another is the voltage feedback pin 2,which is 2.5V on both modules!If you take a look at the schematic you can see that pin 1 only connects to higher or equal potentials(V_FB=2.5V, V_REF=5V) and a connector.Probably this wrong voltage results in a low pulsewidth driving signal,so the inverter's energy transfer is limited.

Does anybody have an idea what causes this?Maybe replace the UC3845?

[petespaco]

It's good that you still want to repair that welder, but, if you only miss one component, you could be back to "square one" in a few milliseconds.  I wonder if that particular welder has had a good track record or not.  When I looked at your first picture, I saw the name "Awelco" as the mfr?  At first I thought it was "Welco" with a preceeding Triangle.  But "Awelco" it is.  apparently made in Italy?  On one site, it seemed to imply made in India???
Anyway, when a company names itself by hitchhiking on a well known name like "Welco", I begin to wonder-------.
  I couldn't find any reviews after about 15 minutes of trying.  Maybe someone else can do better.

[Laci]

Thank you for the answer and searching, petespaco!
The manufacture is Awelco,made in Italy.I only found some informations and troubleshoot on a russian website.The welder is nice and compact for the first look,but it has many disadvantages for the person who wants to repair it.Not only the SMD parts make the repair hard,but the shunt based current sensing kills many components in case of a failure.

While probing on the working module I accidentally shorted the UC3845's pin 7 to pin 6.This made the chip to act the same as on the not working module: low duty cycle driving signal.I think these ICs can partly die.
I'll order some replacements and try again...