Inverter Welder Strangeness

[Crystaleyes]

Re: Inverter Welder Strangeness

What went up in smoke?

No smoke. Just a crack and pop. Then the unit was dead. When I went back ten minutes later it fired up again.

???????

[Crystaleyes]

Re: Inverter Welder Strangeness

Ok.

So I've been in the garage and tried some welding. The unit does exactly what it was doing when I bought it.

It sparks... But doesn't 'arc'.


Anyway, something is clearly amiss, and it seems that something is drawing the current away from the output cables.

Perhaps a clue can be found in the fact that the neon inside the on/off switch lights up even when the unit is not plugged into the wall. It must be getting the power from main caps, as it stays on for some considerable time.

This indicates to me that there is a closed circuit on the power supply section which shouldn't be there. Looking at the original sketch I made, the only path is through the bridge rectifier and the PTC, although surely the capacitor shouldn't be passing DC? I think I'm missing something...

Could this loop be why the dimbulb was glowing so much, and why the unit can't arc?

Looking for solutions online, one possibility is that the main 30A relay fails under load. This was tested for funcionality but not for integrity. It is certainly a possibility.

[CapLeaker]

Re: Inverter Welder Strangeness

Not necessarily. Could be also a sensing circuit problem. Do you see any large resistors looking like 2W axial resistor on steroids? May have weird colour bands / resistance values? If the sensing isn’t proper, the PWM can’t fire the IGBT’s full bore either. There is something we are missing. The crack / pop noise was weird too… I’d investigate that as well.
I would say the unit should work “underpowered” with the dim bulb tester inline. However straight out of the outlet, the caps should have full juice available for the inverter.

[Crystaleyes]

Re: Inverter Welder Strangeness

A fault in the sensing circuit is also definitely a possibility.

So, there are actually four 2 watt resistors on the board. 2 x 22 ohm, 1 x 150k ohm and one which has lost it's markings but reads 82 ohms.

The '82'Ω resistor with no markings is connected to the input pin of an LM7805, so I don't think that this one is a problem. The others measure fine - 22.something ohms and 149.something k ohms.

On a different tip, one thing that I've noticed is that the two heatsinks which hold the IGBT's are at completely different potentials. I noticed this after getting a belt from one of them last night! The IGBT's are not insulated.

In DC, (to mains ground) one heatsink measures 15v whist the other is 55v

In AC (to mains ground) the first one is at 80v 6KHz with the second measuring 112v 60Hz

In AC with no ground (only the red lead) reads 9v 44KHz and 12v 60Hz


Personally, I would have expected the two IGBT's to be running at identical levels.


The loud CRACK occurred again, when switching the unit on today. It's not too dissimilar from a capacitor shorting...

[CapLeaker]

Re: Inverter Welder Strangeness

To 9999999.9% resistors either goes higher in value or open. There was one case in my lifetime it went short. Other suspects also would be the blue caps and those red brownish caps that look like a hand bag (pic 4). Go after the cracking sound. Maybe a cold solder joint? I wonder if it has something to do with all the weirdness going on. Take all the boards out and have a good look top and back side. While your at it do some straight shot, high resolution pictures of the boards.

[Crystaleyes]

Re: Inverter Welder Strangeness

To 9999999.9% resistors either goes higher in value or open. There was one case in my lifetime it went short. Other suspects also would be the blue caps and those red brownish caps that look like a hand bag (pic 4). Go after the cracking sound. Maybe a cold solder joint? I wonder if it has something to do with all the weirdness going on. Take all the boards out and have a good look top and back side. While your at it do some straight shot, high resolution pictures of the boards.

Thanks for the Sunday homework...

But yeah, I'll do all that, and check those caps.
I was looking at the 'handbags' today. Pretty sure I've tested all the blue ones and the brown ones are perhaps the only caps which haven't yet been pulled.
So many ideas have crossed my mind, including bad solder joint. Was trying to figure out if there was any possible route for one of the caps to short on startup, but I'll dismantle the unit tomorrow after church and get back to you.

[CapLeaker]

Re: Inverter Welder Strangeness

You don’t have to take these caps out. If there is something wrong, you should see it or be able to maeasure with the DMM. Not sure what to think about that snapping noise. Maybe there is a cold solder joint on the new caps you put in? Somehow the snapping noise came after the new capacitors, right?

[Crystaleyes]

Re: Inverter Welder Strangeness

i shall lift one leg on two of them, as they are in two pairs.

And yes the cracking did start after I installed the new caps, but also after the IGBT's and bridge rectifier were reinstalled.
All the holes were cleaned with IPA and fluxed along with the pins before soldering, and I did retouch one of the joints after you mentioned that yesterday, but I can have a good look at them for sure.
The noise has only happened when starting the unit from 'cold'. Hasn't happened once the machine has recently been running. In fact, I'll go check now...

No. At least no crack!

What did just happen, was that the relay had to click three or four times before turning the unit on.
Therefore I am definitely gonna swap this relay out now. As mentioned, I found a tech here in Brazil explaining how these welders sometimes fail due the the contacts in the relay becoming damaged and unable to handle the correct load current - even though the relay clearly activates and even switches the unit on.
And as I doubt I have this exact relay in my parts stash, that shall have to wait until tomorrow when the shops open.

[CapLeaker]

Re: Inverter Welder Strangeness

Yeah... almost it's starving somewhere. It's like voltage is there but can't handle the current needed. Interesting repair so far. lol

[Crystaleyes]

Re: Inverter Welder Strangeness

Obviously I couldn't resist trying something else whilst the Sunday ticked on by...

So I bridged the two relay contacts with a fat wire and tried some welding.


Unfortunately it still just sparks!

Mind you, at least we now know that the relay isn't the cause!


My mind keeps going back to why the IGBT's don't measure equally?

Gonna have a closer look at that circuit and locate up to which point the signals are the same?



And yes... Interesting


EDIT:

The datasheets below are for the original and replacement IGBT's (just in case I missed something?). The IKW50A is the one currently in use, as was unable to find the original. That said, I have just found them for sale online...

[CapLeaker]

Re: Inverter Welder Strangeness

Are you pointing at the missing body diode?

[Crystaleyes]

Re: Inverter Welder Strangeness

Are you pointing at the missing body diode?

You've lost me.

Are talking about the IGBT's?

[CapLeaker]

Re: Inverter Welder Strangeness

You've lost me.

Are talking about the IGBT's?

yes.
The body diode is there so the current can only flow in one direction. But this shouldn't cause this behaviour.

[Crystaleyes]

Re: Inverter Welder Strangeness

yes.
The body diode is there so the current can only flow in one direction. But this shouldn't cause this behaviour.

Which was why I chose the current IGBT's over other non-diode options.

I've ordered a couple of the original ones which will arrive in a couple of weeks, so let's see if they make some difference?
Personally, I doubt it, as the welder was performing in the exact same 'spark, no arc' mode when I bought it.

Checked the handbag caps and they are all fine.

Found the source of the burning.. And I think the source of the crack.

On one of the traces which connects the mains AC wire to one of the large caps, there has appeared a small patch where the copper trace has become visible through the green tint. It is here where the smoke comes from at start-up, although it only lasts for a second or so...
I soldered the bare trace to see if that helped but no. It still smokes.
You can see the wisp of smoke in pictures 3, + 4, below..

[CapLeaker]

Re: Inverter Welder Strangeness

Just don’t forget to put a coating on top of the repair.

[Crystaleyes]

Re: Inverter Welder Strangeness

Yes. It'll all be nicely cleaned up, but I just need to work out what is impeding the system from working properly. I have desoldered and resoldered so many points and and components so many times that I have become somewhat slack at keeping things tidy, but once this is fixed I'll clean up then lacquer the whole PCB.


To recap...

An overview would be, that this is an AC-DC-AC-DC welder.

For sure the mains in, is 127v AC.

For sure, (as discussed) that is magically rectified into DC, which after supplying the IGBT's, then uses an LM7805 to power the display and PWM circuit which in turn fires the IGBT's, through two sot-8 mosfets and a small transformer.

For sure, the IGBT outputs then pass through two TO-220 (schottky?) diodes and a fat-arsed inductor before leaving the unit as 64vdc across the main welding leads.

In amongst all of this is some serious current restriction.


Anyways, I'm having a break...
The alternative IGBT parts will only arrive in a couple of weeks and I have various other things to be getting on with.

Thanks sincerely for all the suggestions, and if anyone has any other idea as to what the problem may be then please, don't be shy.

[Crystaleyes]

Re: Inverter Welder Strangeness

The replacement 'original' IGBT's arrived this week and were only put in today.

Trying to weld again in the garage still didn't work properly - only sparks - however I am pretty sure that the problem has revealed itself.

That trace which smoked on startup, which I had 'repaired' has had a catastrophic failure, as can be seen in the photos...

I didn't notice it whilst in the garge as all the sparks were making enough smoke on their own. What I did notice though was that the green 'power' light on the front panel was pulsing after the machine had been switched off.
It was only when I went to take some readings on the bench that I noticed the thermo-nuclear meltdown which had occured on the trace which connects the AC mains in to one of the 2200uF filter caps.

As mentioned, I bought this in order to learn how these welders work, I think it is safe to say that I have learned a valuable lesson here in NOT to ignore, or take lightly, smoking traces.


The truly mad thing, is that the unit still powers up.... !!??

[CapLeaker]

Re: Inverter Welder Strangeness

Well, I guess you found the problem. It did kind of act like it had power, but not enough current to do anything. Now you have to grind all the black carbon off, seal it or fix the hole and fix the trace.

[Crystaleyes]

Re: Inverter Welder Strangeness

I had been seriously thinking that the fault was due to a weak link somewhere. I'd been checking all the removable connections through the mains switch and also from the board to the welding cables outputs.

Before I went to the garage today, I ran the welder for a good 15 minutes or so, taking readings and the pwm section through the IGBT's appeared to be on the button.
There was a stable 46KHz frequency with a 50% duty cycle, and without any load the machine appeared fine.

Anyway, now gonna clean it all up and have another good look, coz at the moment the burnt trace is still sparking and glowing when the machine is running, even though there is no longer any clear electrical connection.
How this welder stills turns on when one of the two AC mains inputs has a severed trace is gonna be interesting to see.

The solder station is on.

I'll be back...

[CapLeaker]

Re: Inverter Welder Strangeness

The black carbon is conductive and gets worse if high humidity is in the game. You have to get all that black stuff removed using a rotary tool or cut it out and fix the hole and the trace. It’s a wide trace, so it needs to support a lot of current.