`Scoping SCR trigger signals? Single phase TIG welder repair

[Chris Wilson]

Going on from this posting:

https://www.eevblog.com/forum/beginners/tig-welder-possible-triac-problem/msg439650/#msg439650


A local transformer company have done a full rewind of both bobbins of the choke (reactor), it looks great, but is slightly bigger, like me after a night out Had to shift the mains transformer up a bit, and move a terminal strip, but it's in the chassis. had a bit of a panic thought, it was tight.


But I now have an issue with what may be the control circuit on DC. Can anyone help with the below please?

I am repairing my failed TIG welder and a shorted choke coil (called a reactor in some regions) has been rewound. See photo attached of the original. One of the two bobbins had an inter winding short, effectively shorting the output of the mains transformer when on DC.  Out of circuit testing of the two high amperage SCR's shows them good, but I am having issues where I believe on DC setting the output is not being rectified correctly. It may be a control circuit issue, or a phasing issue of the new choke. I would like to look at the trigger signals to the gates of the SCR's. The circuit is hopefully attached, if not it can be seen at http://www.gatesgarth.com/schematic.jpg There is also a function description from the user manual as a jpg at http://www.gatesgarth.com/function-description.jpg I have both a conventional mains powered dual channel scope, and a more usable (given ease of portability to the TIG, as the TIG ain't going nowhere far, it's a 300 pound plus brute), a 200MHz USB dual channel scope that I can run on my laptop, fully isolated from the mains> I do believe the transformer coupling in the single phase TIG welder should preclude mains ground loop issues, however?

I need to know where is best to take the ground lead of the probe to, chassis ground, or the cathode of the SCR, and what sort of trigger pattern one might expect to see? Thanks.

[Paul Price]

If it works on AC and not in the DC welding, and the SCR's are OK by test, then your issue is in the circuitry which creates the SCR trig. pulses, not the SCR's. And this circuit is something that I cannot find in the schematic.  I could not trace R1-23 R1-22 to a trigger generating circuit.

[Chris Wilson]

Hi Paul, thanks for the reply, There is another big control board in a panel mounted box, pretty full of IC's. All are marked and seem common devices. Unfortunately I have been unable to get a schematic from the maker for this board, so testing may be a bit tricky. it's all through hole though,  and 90% of the controls seem to function fine, such as gas pre and post flow, current ramp up and down, stitch  and spot weld timing, blah blah. As the front control knob is on this box and board I assume it also has the trigger generating circuitry. I will scope the gates,  and see if there's ANY sort of waveform there, but as it does control power output I can only guess the SCR's are seeing some sort of control signal though? Thanks again.


EDIT:

OK, I have scoped the two gates, ground lead of probe to cathode, probe clip to gate lead, for each of the two SCR's. Only one seems to be getting a trigger signal. It's at 50Hz, and is the same trigger, on the same SCR in both AC and DC mode. The gates go to the box for which In have no schematic, to a computer parallel type and size plug and socket on the box. There is continuity up to the plug. I intend having a good look at the board and to try tracing the signal path to the gate pins. Hopefully I can "reverse engineer" it enough to work out what may have failed.


Trigger signal screen shots are at

http://www.gatesgarth.com/trigger1.jpg

and


http://www.gatesgarth.com/trigger2.jpg

The timebase is changed in trigger2 to show two events, and I put the vertical cursors in to show frequency at 50Hz.

The probe was set X10. The scope is not exactly the best in the world, but hopefully it shows enough for someone to say if they do indeed look like the gate signal? Cheers. I'll also try and attach them here.

[G7PSK]

I usually just get these people out to repair my plant if one goes down rather than mess around myself.

 http://everark-weldtronics.co.uk/

Always found them very helpful and will often give advise over the phone on how to fix something myself if just a simple job. Plant down time is not worth my time fixing it as a rule better doing what I am best at and letting some one who repairs welding plant all day every do the job, its faster and cheaper in the long run, also I cannot give myself a warranty on the job.

[Fank1]

Chris;
It looks like trigger 2 is the good one.
When the scr gates on it clamps the gate to about .7 V ans is very quick.
The trigger 1 shot looks like an open gate on the SCR, there is no clamping action and it doesn't look like the gate is taking any current.
If you want to test it, remove it from the circuit and series it  and an auto tail light bulb with a 12 V battery.
Connect the anode to the positive terminal.
Then take a 1K or so resistor from the positive and just touch it to the gate lead.
The light should come on and stay on.

You sure do a lot of electronics for a race car guy !!
Jerry

[Chris Wilson]

Hi again Jerry! I didn't explain my screen shots too well, sorry. They are both with the same probe on the same gate (of T1). Only the time base of the scope is changed. Same events, different time base settings.  Trigger 1 is one trigger event with a fast time base. Trigger two is the same gate but a slower time base showing more than one triggering event, but in less detail. If I scope the gate of T2 I see no trigger signal at all. Does that make better sense? Thanks for the reply, I must owe you beer by now

The TIG is intrinsically tied into race cars and engines. The PSU you helped greatly with was ham radio, and "fun". I attempt both fixes to learn and save money. Mainly to learn, as time is really money, as G7PSK infers.


G7PSK: Thanks for the reply, this thing is late eighties, I doubt any pro repairer would want to get down to the board level without a schematic, but for sure I will ring them tomorrow, thanks for the link!

[tautech]

Often SCR's are triggered via a pulse transformer, usually 1:1, to provide isolation.
If you trace the trigger gate wiring to the other board (missing schematic) and test for waveforms there. If it is part of the control panel, you may have to solder leadout test wires to points of interest so it can be powered up safely.
Also check for continuity from that board.

[Chris Wilson]

tautech, you are spot on, the gates trigger via two toroidal transformers.

Good news this morning, a schematic has been e-mailed to me.

It's too big to attach, so I have put it at http://www.gatesgarth.com/71616531.pdf


Before this arrived I had pulled some semiconductors and tested them out of circuit.


T11 to T15 inclusive test good out of circuit, as do D73 to D77
inclusive. Still wondering if IC8 the LM234 is the problem. I
thought it  looked like it may have been hot. See photo of
board please.  Hopefully I can do some scoping now!

Any suggestions where to probe welcome. Thanks.

[tautech]

Scope the primary of the pulse transformers. Do continuity checks of primary, secondary and connections to SCR's.
I will do some study of the schematic.

[tautech]

LM324. Pins 3 & 5 are AC ref from 24 VAC for SCR firing timing. Check there is a AC reference on each.
You have one channel working so check signals and take notes and then compare the dead channel and work backwards towards signal source. The 324 does look different to others but it is a readily available IC and cheap. Consider fitting a new one in a socket.
I'm wondering if the control failure caused the choke meltdown?
Any other faulty component you may find, you should consider upgrading it. (higher voltage/current/wattage)

[Chris Wilson]

OK, great tips, thanks very much. I will have to use extension leads to the scope / meter as the box has a PC parallel port style connector with a short lead, so access to the inside of the box is impossible when connected to the machine. I was looking to see what would power up if I ran 24V AC to the unit on the bench, but it's so complicated (to me, at least ) I am not able to tell if it would do anything useful in that state. I have ordered the IC's I do not have access to via a pal, so this week i should be able to swap out any of them.   For sure IC8 looks "different" to the other LM234's. Maybe I should consider an infra red imaging thing? My pal who repairs Bosch automotive diagnostic gear swears by his, he does one model that has a plethora of chips in it, and any one of a bunch of about 12 is prone to failure. With his gun he can see a hot one and do the job in a tenth of the time it used to take.


I have a question about the circuit: It looks "balanced" in so far as there are two identical circuits around two halves of IC8, T11 and T!2, and T14 and T15. Save that is for T13 and its associated bits. What is the function of T13 please?  Cheers! 

[tautech]

Not brilliant on transistor theory, but I strongly suspect it is to invert or phase shift the AC for the SCR triggering. If you scope T 11 & 14 collectors(1 for each channel) they should be out of phase. (Refer to theory of operation waveforms.)
Check Diodes 66 to 71 also. Resistors also, often they measure less than their value in-circuit. But you are looking for open circuit.
Also check pins 8 & 14 IC8, that their output is even in amplitude.

[Chris Wilson]

I probably won't be able to get into this until tonight or tomorrow, that damned nuisance called work is intervening

I thought it was to phase shift, perhaps under control of the AC balance control. This gives different waveform shapes for either better oxide cleaning or better penetration when welding aluminium alloys.

But that seems to be under the control of  IC3 on page 1, and it seems to work. The odd thing is current control (seemingly set by AD1 into pin 12 of IC8) seems to work fine in AC mode, and none linearly in DC mode, yet I see no pulse width or pulse frequency change of the SCR trigger pattern. Maybe it's because I am not actually striking an arc when I test. To strike an arc means having the HF on, and it has already cost me a cheap multimeter, as I hadn't realised it was automatically switched in on DC as well as AC. on DC it runs until an arc is established, then shuts down. I was looking at the DC voltage at the machines ground and electrode sockets, and the meter did NOT like HF through it!   HF on a TIG welder is is RFI on steroids.

Thanks again!

[Chris Wilson]

I have very carefully soldered leads to pins 1, 3, 5, and 7 of IC8, and brought them out through an existing case hole. They are each about 18 inches long. I can then scope the inputs and outputs reference to  ground, with the unit in the machine and connected up. I am hoping the lead length won't matter at the frequencies involved?  I will do this tomorrow when I am fresh. Hopefully it will show some useful info. Cheers.

[tautech]

Chris, pins 2 & 6 should also be checked, they both get a SCR stop or run signal from P5 schematic via a 2 diodes. (AC or DC?)
It is +13.5 or -15, so you can check it with a meter.
If pins 2 & 6 measure ~13V for stop and ~ minus 14V for run they are ok.

If pins 5 & 7 show a square wave phase shifted they are ok. Could possibly be a triangular waveform.

Pins 1 & 7, identify which one feeds the SCR gate good waveform you posted the other day

[Chris Wilson]

I used my normal bench scope, so don't have screen shots, but I feel I do have some pertinent info.

All data relative to PCB ground.

Pin 7 which is to the firing SCR has a perfect square wave trace of about 30V when the gun trigger is pressed. (6 divisions at 5V per division). No trace when not pressed. Same on AC and DC.


Pin 1 which is to the none firing SCR has no output, AC, DC, trigger not pulled, or pulled.


Pin 5 has  a shark's fin trace, all the time the machine is on, AC and DC, trigger not pulled, or trigger pulled. It's there from the get go. About 10V P to P


Pin 3 has the same trace exactly as at Pin 5 under the same conditions.


Now I think this is the interesting bit.

 
Pin 6 has 13.02V with machine on, AC or DC setting. Pull the trigger and it drops to 0.64V. On the scope it's showing as a flat trace that just jumps down when pulling the trigger. No waveform as such. Voltages also checked with a multimeter.


Pin 2  shows a similar 13.02V with the machine on in both AC and DC settings. But when I pull the trigger it drops just a fraction (but repeatably) to 12.98V

Do I need to now look at pins 9 and 10 (working SCR "side" of IC8) and pins 13 and 12 (none working SCR "side"?


The none working side is seemingly fed by AD1 which appears again on page 1 at the bottom, part of the AC balance stuff. I also see at the very bottom of some pages reference to "Udgaade Komp", which I guess means updated or upgraded components?  There are cited in this region.

Your thoughts appreciated! Thanks.

[tautech]

Pin 6 has 13.02V with machine on, AC or DC setting. Pull the trigger and it drops to 0.64V. On the scope it's showing as a flat trace that just jumps down when pulling the trigger. No waveform as such. Voltages also checked with a multimeter.


Pin 2  shows a similar 13.02V with the machine on in both AC and DC settings. But when I pull the trigger it drops just a fraction (but repeatably) to 12.98V

Do I need to now look at pins 9 and 10 (working SCR "side" of IC8) and pins 13 and 12 (none working SCR "side"?

Chris, we may have overlooked the obvious.
P5 top right, ED1, ED2 have only 2 states: +13.5 or -15V
Check the minus 15 V rail. The 324 op amp is supplied with + 15 and -15 and can swing nearly to each rail. Both rails need to be near spec for it to do so. Pins 2 & 6 results are revealing IMO.

Pins 10 and 12 is what I would check but after + & - rails were checked

[Chris Wilson]

Good and bad news.

The good (I think) is there's no 15V negative rail voltage. Just under 1 volt positive is showing on pin 11 of those LM324's that use the negative rail. At IC26 the 15V negative regulator, I see about 32V on the centre pin (input) relative to ground. Nothing seems to be on the other pins, which seems odd, given pin 11 on IC8, for example, shows a tad of positive voltage on it.

Bad news is I may have either slipped with a probe or allowed something to short, it's a nightmare trying to access the box whilst it's connected to the machine. Now the gas valve is actuating all the time the machine is on. It only used to (correctly) come on when the trigger was pressed. If I move the control box the gas valve actuated LED will go out, but the valve is still open. Also the current display reads something like 285 amps all the time, and the current control knob has no effect on the display at all.

I am thinking either there's a short on the 15V negative rail somewhere (maybe a chance to try my Tone Ohm clone short detector I got at a rally last year?) or the regulator is dead. Maybe both?


I am kicking myself for not soldering an extension lead on the negative pin, but instead trying to probe it live with the box half in and out of the machine. Mea culpa.


I am wondering if feeding it 24V AC through the Variac with a very low amperage fast blow fuse in line would allow more civilised bench testing of the negative rail issue?


Thanks for the idea, I was assuming as so much of the box seemed to work the voltages would all be good, but i think a lot of the devices run off a single rail?

Thanks to my stupidity I feel I have gone a few steps forward, but also possibly a few strides backwards


I think I need to address the negative rail first.

[Chris Wilson]

I removed the -15V regulator and bench tested it OK, I then refitted it and for some reason the -15V to IC8 is back! I will get a new regulator in case it was the heat of removal that did something.

I have also fixed one of my self inflicted issues, in that when i shorted the PCB trying to measure things half in and half out of the machine I found the gas valve permanently open. I found T4 shorted B,C and E. A new one has sorted that.

My issue now is that it gives full current output both on the digital meter with the torch trigger off, and also attempts to weld at full power even when set to minimum on the current adjust.

On page 1 of the schematic I see IC2 should go 0 to 10V as the current adjustment knob is swept max to mininum. Here is what i am seeing on IC2 pin 8, relative to PCB ground

Torch trigger OFF, machine ON:

Current set minimum: -2.9V
Current set maximum: 0.1V



With the torch trigger pulled I see:

Current set minimum -9.2V
Current set maximum -27.5V


The way i read it maximum output should be at 0V and minimum output at 10V

I powered the box with 24V AC on the bench and had a poke around.

Again on page 1 D4, D5, D6 and D7 voltages are correct. IC2 and all other op amps see correct +15V and -15V on pins 4 and 11.

On the bench (and I am not sure it doesn't need two different feeds of 24V AC, I just fed pins 36 and 34 on M1) I see the following on the pins of IC2, which given what i see on pin 8 of it, when installed in the machine suggests I am not feeding it all it expects when it's on the bench as a bare box.

IC2, on the bench supply of 24V AC to pins 36 and 34 of M1:

Pin 1: 0v
Pin 2: 0V
Pin 3: 0V
Pin 4: +15V
Pin 5: 0V
Pin 6: 0V
Pin 7: 0V
Pin 8: 0V
Pin 9: 0V
Pin 10: 0V
Pin 11: -15V
Pin 12: 0V
Pin 13: 0V
Pin 14: -0.63V

I am a bit stuck now as to what I may have done to lose the current control 0 to 10V output on pin 8 of IC2.

Any help much appreciated. Thanks.

[tautech]

Chris
I notice on the schematic there are numerous circled test points.
Get back to the guy whom supplied the schematic and see if he will send you the info for those test points.
They should provide voltages and possibly waveforms for troubleshooting. There may be specific test setup conditions for the test points and you will need these also.

Rob

[Chris Wilson]

Hi Rob, thanks for that. I noticed the test points myself, but he doesn't think he has that info any longer. He's double checking. He may have some waveform data from a more basic control box (Type 3) that could conceivably help. I  have detailed waveforms for the inputs and outputs of IC8 now, done on the bench. That was the original issue, one SCR not appearing to fire.  By the time they get to the primaries of the two 1 to 1 transformers they seem a bit shaky. with some pulses seeming to be being dropped. I'll collate the info on just what points I probed into a sensible format and post them later. Cheers Rob!

[G7PSK]

Here is the another company that may have some information, at one time I purchased all my equipment and consumables from them. All they sold back then was Migatronic. Then they went to pushing Fronius and I have not dealt with them much since then, but they used to be very helpful and would come and fix down plant the same day (usually by a board swap).

http://www.miganglia.co.uk/

Does the unit you got have scratch and or lift tig as well as HF start, if so you should switch over to that or pull the plug on the HF generator before making any test measurements, the HF is an absolute bugger for breaking test gear, I used to have an old BOC machine that the HF could be pulled out to an arc nearly 2 meters long and would creep all around the work shop if there was a bad earth on the work piece.

[Chris Wilson]

Hi G7PSK (sorry, can't find you at qrz.com to call you by a real name). The chap I am speaking with worked for Migatronic, he says most technical stuff on older machines was binned, and it's only a few old time service engineers who may have kept information.

I have learned since blowing up my trusty 15 year old meter (not an expensive one, but like an old pair of shoes, one i was comfortable with0. I have the fuse for the HF pulled. I can sometimes get it to scratch start, but I think the issues are resolvable without actually having an arc running.

If I could understand the circuit better around Page 1 top left quarter I would be able to make more sense of my voltage readings.  M1-18 and M1-19 go to the welding torch trigger switch. Normally open, closed when you want an arc.  As far as I can tell IC2  pins 9 and 10 work as a comparator to set current level, voltage across the main current shunt in the + output lead is fed into (page 4 top left), IC5 pins 5 and 6.  I believe I should see a 1V to 10V output on changing the current adjustment pot, on pin 8 of IC2 (page 1, just above centre right). I don't. But I don't understand the circuit....


As for  the original problem, no trigger signal to SCR T2, I attach screen shots of scope patterns from the input pin 3 on IC8, which is similar when I scope input pin 5.

I also attach screenshots for the waveforms on the output pins 7 and 1. Then onto The T12 and T15 sides of the resistors R177, and R180.


Finally I scoped the transformer primary sides of R177 and R180. All relative to PCB ground, although perhaps in hindsight i should have scoped across the primaries of the output transformers TR1 and TR2?

Once I got to the transformer sides of R177 and R180 the patterns became intermittent, and of various amplitude. I am wondering if that was my measurement method 9to PCB ground) though?


I'll try the other firm you suggest after the Bank Holiday. Thanks for that.

[tautech]

As for  the original problem, no trigger signal to SCR T2, I attach screen shots of scope patterns from the input pin 3 on IC8, which is similar when I scope input pin 5.

I also attach screenshots for the waveforms on the output pins 7 and 1. Then onto The T12 and T15 sides of the resistors R177, and R180.

Finally I scoped the transformer primary sides of R177 and R180. All relative to PCB ground, although perhaps in hindsight i should have scoped across the primaries of the output transformers TR1 and TR2?

Once I got to the transformer sides of R177 and R180 the patterns became intermittent, and of various amplitude. I am wondering if that was my measurement method 9to PCB ground) though?

Yes you should connect scope across pulse transformer primaries. Just make sure your scope and computer is not grounded in any way or better still use 2 probes, 2 Channels and no Gnd leads for a differential setup.
I can't see in the images where your trigger level is set. Just make sure it is well within the waveform amplitude.

It seems you have output for both SCR gates now whereas previously no -15 V rail was the problem?

As for your previous probe slip, we will just have to fix it, bit by bit.

[Chris Wilson]

OK, thanks for taking a look at them! I have read of ground loop issues, so I have isolated the Variac with an isolation transformer. Hopefully that should do the trick. The control box sees no direct mains, only 24V AC into it, on the bench, and in the machine. I have now run out of hard drive space, my proper drive failed a few months ago and this was was a stop gap. the gaps just closed

Just tried Acronis True Image with my new 1TB drive as a USB connected device and it hangs... If I am not back you know why Trying with it connected internally, I never have much luck with USB. In fact, under Windows, USB and Sound is a nightmare. Thanks again, I may be some time...


EDIT, a painless transfer of image once I installed the drive I wanted to clone to INSIDE the machine, instead of hung off a USB boxatricks. I now have nearly 2 TB of drive space, so verbose replies not a problem

[Chris Wilson]

Sorry to have to return to this subject again. I borrowed a supposedly good and working control box from the Migatronic guy, and although the machine now welds in AC and DC I believe there is still an issue. I renewed the unobtainable R1 on the "schematic.jpg" link 100 Watt 10 Ohm centre tapped rheostat above the fan, that blew when the choke shorted, with 2 off 10 Ohm metal clad 100W resistors in parallel, with another pair similarly connected in series. I then centre tapped the two pairs, giving me again 5 Ohms per side, centre tapped. These big boys are also on big cast heat sinks, below the fan, so should be able to dissipate far more heat than the original. There's a label on the machine saying the original rheostat should be set to give an equal 5 Ohms per side, centre tapped. My metal clads measure nears as damn it 5 Ohms per side, too, measured properly with Kelvin clips on a decent low ohmage meter.  In DC welding or just squeezing the trigger and not striking an arc, these stay cool. But in AC mode actually welding the half connected to M3 pins 10 and 13 gets mad hot. If I stay in AC mode and just pull the gun trigger but DO NOT strike an arc, both get mad hot. They then get very mad hot, fast, if I don't actually strike an arc. I really,, really need someone to tell me what this rheostat does and how it interacts in the grand scheme of things. When actually welding in AC mode it's almost as if the arc is splashy, and it seems to pulse  rapidly.  The Migatronic guy has gone to earth, and does not seem keen to involve himself further, but it's not through my unwillingness to pay, him or anyone else for that matter. Other TIG specialists do not seem to want to know at all. I now have a lot of money tied up in this damned thing, what with a new choke and whatever Mr Migatronic will want for his supposedly good control box...

I will link to the schematics again, unfortunately they span different documents, thanks, if anyone can find the time and will to have a look I'd be eternally grateful:

 http://www.gatesgarth.com/71616531.pdf

and


http://www.gatesgarth.com/schematic.jpg

The original rheostat that blew when the choke shorted is at:

http://www.gatesgarth.com/R1.jpg

R1 is on the "schematic.jpg"

Thanks again!

[megajocke]

You may have to redraw the schematic of the parts around that resistor to make sense of it without going crazy. From what I can see it's part of the circuit involving C1 which is described in the functional description.

The functional description says the open circuit voltage should be DC even when in AC mode. Is this the case? It seems at least one of the main thyristors shouldn't be receiving triggering pulses until the arc is struck. If there is AC even when unloaded, this could explain excessive dissipation in the resistor.

But anyways, how hot do they get and what kind of heatsink do they have? The aluminium clad style resistor power rating is usually given when mounted to an infinite heatsink, while the ceramic ones are usually rated for natural convection. Also remember that the resistors aren't made of flesh nor silicon so being hot may not be a problem unless excessive. "too hot to touch" wouldn't even make them sweat 

[tautech]

Chris, a few simple measurements can answer your question.
First measure voltage across each 5 Ohm resistor.
V²/R=W
The answer will show if you are within the W rating of the resistor.

Ref:
http://www.electronics-tutorials.ws/resistor/res_7.html

[Chris Wilson]

Thanks for the replies, worse case scenario was 30V AC across one half of the "rheostat", so I make that
30 X 30 = 900      900/5 = 180W which sounds a hell of a lot. This was not a steady voltage though, but seemed to pulse...

I am still unclear what the rheostat does though, and why it has such loadings on it.