WWS200, Parweld XTS162 welder repairs

[Arvorsteve]

Hello and welcome to the Forum !

Since it is the same model of welder and has similar faults, I think it would be OK to just carry on with this thread, so please go ahead here.
Here's a similar inverter repair thread, by the same OP - @The_Motorcyclist - which ended in success:

Parweld XTT182DV no power

I have an XTT182DV inverter set, which is currently not showing any signs of life. A preliminary investigation is showing that I am getting AC voltage to the first pair of rectifiers and when the DC side of the rectifiers is disconnected from the main PCB there is 240V DC. When the DC side of...

www.mig-welding.co.uk

Hi Eddie and thanks

Please see photos of my IGBT gate drive signals and the front of the model I have.

I have replaced the IGBT's and the welder ran for about 10 seconds off load ( after replacing the trashed start up thermistor) until one of the IGBT's failed again. I am wondering if the gate signals are the issue. There are basically 4 x FGH40N40 IGBT's, configured as two pairs, with their collectors and emitters paralleled up for current handling.

Both gate drive signals are in phase and appear to be at about 57Khz and 22V, although the square waves are not too good and there are a few overshoot spikes on the edges, that might have caused damage?

I would be interested in seeing a PDF of that circuit that The Mororcyclist posted, although I don't think it's for the model I have.

I like the idea of running the IGBT's up on a lower voltage current limited supply as someone else suggested. I think I might try that when the replacement IGBT's arrive, but I am reluctant to do that until I get some comments back about the gate drive signals.

Thanks guys.

 

[eddie49]

The oscilloscope traces are interesting, this is the first time we have seen any IGBT gate pulses from an inverter welder.

As mentioned in Post #16 of the "successful" thread:

Parweld XTT182DV no power

I have an XTT182DV inverter set, which is currently not showing any signs of life. A preliminary investigation is showing that I am getting AC voltage to the first pair of rectifiers and when the DC side of the rectifiers is disconnected from the main PCB there is 240V DC. When the DC side of...

www.mig-welding.co.uk

if the four IGBTs are connected in two parallel pairs then they form the two "effective" transistors of a half-bridge inverter. In that design, I believe that the transistor pairs must switch on and off alternately to send current through the primary of the step-down transformer in opposite directions. As such, they must not be enabled into conduction at the same time - which the in-phase gate pulses seen in the 'scope traces would do?
If you set the scope to Chop rather than Alt is the phasing still the same?

 

[Arvorsteve]

I had exactly the same understanding Eddie of the need to have antiphase gate drive signals, so I was also surprised to see both signals in phase. Changing the scope settings from ALT to CHOP makes no difference. I even tried my Hantek scope to be sure!
I isolated the output from what appears to be a transformer before it feeds into what looks like a totempole stage on the IGBT gates and got a better looking square wave, but again, both in phase.I will send more photos over weekend.
I could really use a circuit diagram at this stage. It might be the case that one pair of IGBT are switching the high side and the other pair the low side simultaneously so producing a series path through the transformer winding, although if that were the case there wouldn't be a reversal of the magnetic field, as in conventional bridge set ups.
I'm going to have to spend some time tracing the pcb tracks and attempt to reverse engineer the circuit. What fun!

 

[eddie49]

This discussion about a homemade inverter welder initially gets the gate signals from a UC3844 PWM controller connected via a pulse transformer with two secondaries - which must be wired in opposite directions:

100A inverter welder ( IGBTs keep blowing)

Hello, I am trying to design a 100A smps for welding , the design is two switch ( IGBT) forward converter using current transformer for feedback in primary current, the switching tranformer is a commercialy available toroid transfomer recovred from already working welder , the issue is when i...

www.edaboard.com

Later that is replaced with an IR2110 2-channel gate driver, which has complimentary high and low outputs.

The suggestion of using a current-limited low voltage external DC supply instead of the usual rectified AC mains, to avoid damaging new power switching components, came from Page 47 of the Selco repair manual shown in Post #33 here:

Can anyone recommend a repair place to fix inverter

I think the problem is your asking quite a specialised technical question - not many people would be able to confidently give you an answer on fault-funding to this level - I would think it should trigger the hf first but that’s only a slightly educated guess :dontknow:

www.mig-welding.co.uk

This bench supply for the IGBTs would be 30 to 40 volts DC, limited to 2 Amps, but you would have to arrange for the gate driver circuit to still have adequate voltage.

 

[eddie49]

Apart from the in-phase/out-of-phase issue, if you adjust the current setting does the mark-space ratio of the trigger pulses change?

 

[Arvorsteve]

Hi Eddie and thanks for the links. I didn't get a lot of time over the weekend to look at the welder, but I did manage to start tracing out the gate driver circuitry and it looks very similar to the circuit of the DIY welder in the link you just sent. Not a conventional totem pole setup with one NPN and one PNP transistor as I was expecting, but there appears to be a single PNP transistor and four conventional diodes, plus a couple of zeners, although not wired exactly as per the DIY model if I traced it correctly. There is the isolating transformer though, so the circuit is very similar.
I expect this transformer and my earth reference point for my scope is the cause of my in phase signals.
Whilst I have removed the earth lead on my scope in order that I can have a floating probe, I was referencing the scope traces to the 0V DC rail of the internal circuitry! Given the isolating effect of the transformer, it's probably the reason for what appears to be in phase signals. Later in the week I will try and find some more time to reference the two channels to the separate transformer windings, as opposed to 0V and let you know if I then get the out of phase signals. It might also be the case that I observe a better square wave as well after doing this.
I haven't found a UC3844 PWM yet on the board, but several of the IC's are under a varnish coating with no markings underneath. The circuit is very similar to the one in the link you sent, so thanks for that. At least I now have something to work with.
I will also try altering the current setting and observe the mark space ratios but it looks like the feedback for the PWM comes from another sensing transformer in series with the main output transformer, so the feedback would be load derived as opposed to front panel setting wouldn't it?
Thanks again.

 

[Jan-DIY]

Hey.
Generally these SMPS are designed as an H-Bridge config.
What gates did you measure?
Although the driver doesn’t boost the gates very fast.
This is from a Lorch long time ago.

Yes. Some sort of ringing is on driver side. But gates are much better.
Your signal is weak on transition from off to on. This can cause trouble because transistor are in non SOA.

Cheers
Jan

 

[Arvorsteve]

Hi guys.

Firstly, I don't think this is a conventional 'H'bridge drive Jan. See attached sketch of what I believe the cct to be. I have also drawn my understanding of an 'H' bridge at the top LHS. I think there's generally 4 drive signals providing dead time to prevent cross conduction in an 'H' bridge. This seems different to that, with just the two signals coming from the two secondaries of the gate drive transformer.

Although, having said that the half bridge cct's iv'e seen tend to have a centre tapped transformer to the supply rail. This transformer does not have a centre tap, but there are two sets of IGBT's in parallel.

So, having traced the circuit I could not find a common point that both of the drive circuits between the gate drive transformer and the IGBT's shared. For that reason I am still not convinced that my gate drive signals are 180 degrees out of phase, because I only get in phase signals when measuring direct onto the IGBT gates to 0V. I might be scoping the wrong points.

Having isolated the 350v by cutting the track I might scope Q3 and Q4 gates to the 0V rail and the Q1 and Q2 gates to the 350v rail, then see if I get out of phase signals.

As you can see the rounded edge is improved when the transformer output is isolated, although having individually isolated each winding in turn I can't establish what part of the circuit, if any is causing the rounded edges when I reconnect the transformer.

My oscilloscope traces attached are actually a bit of a cheat to my thinking, because the only way I could get to see two signals out of phase was to cut the track at the points marked X with a circle around them and then put my scope across the four pins of the two secondaries ( TP1 to 4) on the transformer.

It's a cheat because if you notice the earths are on points 1 and 4 ( refer to circuit) and I am scoping points 2 and 3, which I would expect to be the case as I am looking a different ends of the transformer secondaries. It's interesting to see that the circuit of the DIY welder that you posted Eddie also showed the two gate drive signals for the MOSFETS coming from the top end of each winding on the transformer ( 1 and 3 on my circuit same as mine), which would also produce in phase signals.

I think that circuit works by rectifying the signal through the two series diodes. The PNP transistor just switches the MOSFET off quickly when the signal goes negative. I'm not sure how mine is supposed to work

I will report back after further scope probing !

 

[Arvorsteve]

Incidentally guys. How do I set up my account so I receive an alert if someone posts something, because I'm not getting anything. Thanks

 

[Arvorsteve]

Update on this thread is that the welder is fixed. There was another fast switching diode under the white caulk that had gone short circuit and was causing damage to the IGBTS. I also upgraded the 40 amp IGBTS to 60 amp ones. When reassembling It's important to ensure you use insulators on the devices and heat transfer compound before bolting to heatsink. I also bolted the IGBTS to the installed heatsink before soldering, which gets every device flat and levelTwo of the collectors are common, but two are not. The absence of a clean square wave gate signal was a red herring in the end and probably due to the absence of the gate capacitance On insertion of the new devices I obtained good square wave output before rectification. I also tested the welder with the new IGBTS on the hestsink but with leads to the boatd, reason being the printed circuit was stsrting to object to having the devices changed twice! A few photo's attached. Thanks for the help given