Electronic welding machine in short circuit

[coppercone2]

you can also probe the opto coupler in circuit carefully to see if it works by hooking up a power supply current limited to one side and dmm ohm meter to other side, might be a good idea to try that before you reinstall parts

[DisasterNow]

those bases suck their  easy to melt i doubt it damaged the pad you can probe it with a LCR meter near it if you wanna see if its still connected with low ESR but I think its OK

Maybe it's just all the lacquer that was covering the chip, I was blowing hot air in that direction and it probably got under the protective aluminum. First I want to try to clean this lacquer from the board, with isopropyl alcohol I don't solve anything, I read that acetone is supposed to clean the lacquer. Then I'll check that capacitor (but I don't think anything happened to it).

[DisasterNow]

you can also probe the opto coupler in circuit carefully to see if it works by hooking up a power supply current limited to one side and dmm ohm meter to other side, might be a good idea to try that before you reinstall parts

Which optocoupler are you talking about? I don't think I have an optocoupler in my board (both hybrid and main).
Can you explain this check to me better? Thanks

[coppercone2]

on the lower left of the close up picture I thought i saw a 4 leg opto coupler chip (dip -4)


I see its a IRF 110 now. never mind. mosfet

99% of the time, dip 4 chip is optocoupler . welding machines seem to have 4 pin mosfets

[DisasterNow]

I imagined you were confused, yes it is a mosfet and I imagine that it amplifies the square wave signal that should come out of pin 14 QE of the TDA1060.
I checked that dark part better after desoldering, in reality in person it is much less evident. It is just a part with a lot of lacquer, tried with acetone but it does not clean. Oh well it does not matter.

[coppercone2]

nah you need special solvents.

THF works good. Flux remover (like the rosin flux remover from the can) works good, its some kinda freon.

It basically makes it "constrict" and lose adhesion, then you can just peel it off with a stiff brush (ESD SAFE ONLY)


ALcohol has little effect, Acetone I am not sure.

The best stuff is here
https://mgchemicals.com/products/conformal-coatings/conformal-coating-stripper/conformal-coating-remover/

It has acetone, combined with 2 other solvents. Solvent combination is necessary. Often when you mix solvents, its better then just pure solvent for dissolving. They do experiments to see which mixtures effect the glues the best.

Alcohol is used for rework, so they pick a glue that does not get effected by alcohol too much.


Now, there are stronger coatings, like paralyene and stuff... that are a bitch. Those you need to use heat, i.e. a pace lap-flow tool, which is basically a scraper that is resistance heated. You heat it up to a depolymerization temperature of the glue, and then scrap. But that is a expensive vapor deposited coating (vacuum oven required). Usually only found in government equipment, like space parts. I don't think I ever came across it, but there is plenty of info about it.

Also, epoxy is kinda hard.. but its not the best coating anyway.


I found that basically MG chemical conformal coating remover worked on anything I ever come across. If I wanna do a 'neat' job, i.e. a repair on a nice board, where I don't want to risk peeling the coating, I use the lap-flow scraper tool at low temperature to remove the coating, or a abrasive rubber burr, followed by damp q-tip cleaning of the residue.



Being able to work on conformal coating nicely, is a skill IMO... I am sure there is people that are really good at it, at the repairs department of NASA or something. I find it completely obnoxious. That is the same skill as a beauty salon worker, work with smelly chemicals and skin peels.  . They use the same tools too. I imagine a slightly overweight woman with streaks in hair and 1/2 inch long nails doing this job "nicely" while on phone. I just recommend dousing and scrubbing with a ESD safe brush when its all deformed, followed up by ultrasonic cleaning (its really hard to get the board "clean" at a magnifier level, there is always little curls of coating left behind.

[DisasterNow]

MG chemical makes excellent products, I have the varnisch 4226A. However, I saw that with hot air (well, I was at 370°, but I think it works even at much less), the lacquer melted and then when it was cold, all you had to do was scrape it off easily. I always remove the flux with isopropyl alcohol, but I found this lacquer to be really tenacious. But as I said, for now I only care about cleaning the pads well.
Now I'll see about ordering the two chips (I think from the usual Chinese mega shop), they usually arrive after 3 weeks. I'll leave this topic waiting, I'll come back to it after replacing the two components.

[DisasterNow]

Update: I replaced the only two chips of the small logic control board (TDA1060 and LM324) to see if anything changed.
Before I never had a voltage at the output (maybe only when I had those occasional short circuits). Now the situation is as before ONLY if I keep the output current potentiometer at maximum; but if I lower the potentiometer even a little (and then to the minimum current), the short circuit (I see it from the light bulb in series) is continuous, and now I hear a constant 'tic' noise (its frequency decreases with the mains voltage, and increases if I lower the voltage with the variac). I have not understood exactly where the tic comes from, but analyzing with a microphone I seem to understand that it comes from the power transformer (TR2 in the diagram). During this short circuit that is now continuous, the only components that I see heating up are R11 (180°) and C11 (not too much, but it heats up only with the short circuit). But now if I see the output signal of the welder, I have a voltage 'dirtied' by the short circuit as you can see in the image; the frequency of the 'tic' that I hear, I think it is like the time interval where the signal returns to zero and then rises again (about 500ms). In short, now with the replaced chips something has changed, but I don't understand where the short circuit comes from.

[coppercone2]

im too tired but it made me think of a active snubber

my power supply had a snubber / crowbar whatever that kept turning on, it was over voltage protection. it made a terrible clicking noise



What I would do however is try to determine why those chips broke and what other chips might be broken. it could have spread through the power rail. So if two low voltage silicon chips died, chances are more are dead. its like lightning struck a house with those welding machines, from what I see on my dynasty

Just a good bet might be to keep replacing things connected to that power supply


I would guess the IRF and the BD transistor are suspect, I think their gates/bases could be messed up if the chips nearby were messed up

clearly you found the right area. little transistors can break in power supplies that are near big transistors, and the opto coupler might be messed up too. you have no way of knowing when it makes a trigger signal


it is progress

if those chips got damaged they started sinking energy, so hopefully the damage is localized to that area.


the fact that you need to turn it up all the way to do something, it might mean the gain of the error amplifier or whatever you wanna call it is messed up. so if a op-amp is trying to drive a transistor, and the gain on the transistor is bad because its damaged, it won't give the response until signals are very strong. things get "skewed"



imagine the handle on a bicycle break was super badly bent and also loose. when you push it 90% of the way, you finally get out of the 'loose' region and you hit the break a little bit, but it bearly moves the break line even when its fully pushed in. its not 100% disconnected, but your not getting the correct amount of leverage because its deformed and loose, so the input you give it does not bias the system properly. damaged transistors may end up doing something like this, sometimes you can think of this as 'low amplification' or 'poor response' or 'lost sensitivity' or 'low gain' in a control loop


lets say you some how burn the metalization off a mosfet and the capacitance is less at the gate, then if you give it 5V, it does not have enough charge to properly trigger the fet, if you give it more then that, you punch through the oxide layer and break it, so its effectively broken, the response parameters are wrong. with BJT its similar, that is the current you give it does not create the correct amount of current flow. in either case, it still has a transistor like response, but its all wrong for the system its in and in a stressed mechanical state . MOSFET are charge devices, charge is proportional to voltage and capacitance, if you have less capacitance (missing metal), the normal voltage does not contain enough charge  on the transistor

[DisasterNow]

As soon as I have time I will repeat the tick recording with a better software, so I can see the exact point better, it should still be near TR2 (power transformer, I have two images below).
Looking at the output signal in fact it would seem that something at each tick enables/disables the output voltage for a fraction of a second, and that when it is enabled there is still a short circuit in progress.
On the control board, in addition to the two chips already changed, the only interesting thing would be that IRF110 marked in an image with a red arrow, but since it is covered by alcca, the acronym is not readable (I deduced IRF110 from the hybrid diagram found online, but it is not entirely identical to my board).
As already said during this short circuit the only components that heat up are R11 and C11 (I have the images below, also of the part of the circuit that concerns TR2-C11 and R11).
Other interesting silicon components could be the two transistors (T2 and T3) that are in the part of the circuit that goes to the igbt gates, I could also replace them, but checked on the board they do not seem to be faulty.
For many components, a replacement problem could also be that their leads are drowned on the large rear tracks, they have tinned everything together (as in the image some examples circled in red).

[coppercone2]

i heard a clicking coil before when a power rail was damaged for a active crowbar type circuit (over voltage drain transistor). it shorts out the rail and makes the transformer click like a solenoid


that whole circuit with R10, its a snubber,.

3.9k = bleed resistor (drain), C12 R10 = snubber

snubber are known to fail because they take a beating. a partially shorted capacitor could be a phenomenon


The IRF110 or IRFD110 is common in welding machines, mine had a broken one

and there is another 2 transistor, one to the left of the picture, the small footprint one, its suspect since it on that PCB, the other one is left to the IR110


and I don't know what C11 does its confusing to me


also tantalum caps are known to short out and go bad, especially cheap ones on equipment thats not derated properly (aka non-goverment)



so on your ceramic board there is 3 suspicious transistors and 2 tantalums that are just a known trouble maker in cheap equipment unless the designer happened to be smart.


the ceramic are better on ceramic PCB, they usually break for mechanical reasons, they are hopefully OK. But rule nothing out if there was a over voltage you know, its like a moose got into a china store. Usually the more complicated parts break first and sink the fault voltage and protect the basic stuff lol

and if there was a over voltage, tantalums are sensitive to that. and they explode when they fully go, it will shatter that board like a small fire cracker put on top of a soup cracker. they have a bad enough reputation that you might want to deal with that, because it can be a ticking time bomb if the circuit was stressed, which it might have been. If a tantalum really explodes it can put a literal hole in a fiberglass board, ceramic will likely get shattered


debugging the board individually is a good idea, if you can manage to hook it up, that is the right way to do it

[coppercone2]

if the 3900pF capacitor is shorted or partially shorted, a 10 ohm load on that driver WILL make it click and eventually burn up that resistor, 10 ohm load is rather low, its only 10 ohms for the frequency that is supposed to go through the capacitor, for the welding energy it should be open circuit

[coppercone2]

also if you scrape the goo off the transistor it will be easy to read, the top does not need goo, its just hard not to get it there if its dip coated. ideally they would mask all the chips and keep the goo off the top, but it would be tons of extra work only the repair man can see


keep in mind, goo can make failures worse. it makes it more difficult for a component to vent. what you want is energy going above the PCB. if you have a submerged part, it will explode harder. for electrolytics your supposed to put tubes on them if you pour sealant, so the top of capacitor is unsealed and can open up to vent smoke instead of failing to open up and explode!

if you elevate a part off the PCB with a air-gap, it will burn up and leave the board alone. if its glued to the board it will transfer more heat into the PCB and wreck it!


very high end boards, instead of trying to use the board as a heat sink, they use bigger parts, and elevate them with standoff from the PCB, that way they can happily fail while glowing white hot without destroying the circuit card. I had a ton of board repair work from over stressed transistors that perform no RF function. if they elevate them from board by 5mm with rivet standoffs and solder them there, it would be 100% fine

[DisasterNow]

i heard a clicking coil before when a power rail was damaged for a active crowbar type circuit (over voltage drain transistor). it shorts out the rail and makes the transformer click like a solenoid


that whole circuit with R10, its a snubber,.

3.9k = bleed resistor (drain), C12 R10 = snubber

snubber are known to fail because they take a beating. a partially shorted capacitor could be a phenomenon


The IRF110 or IRFD110 is common in welding machines, mine had a broken one

and there is another 2 transistor, one to the left of the picture, the small footprint one, its suspect since it on that PCB, the other one is left to the IR110


and I don't know what C11 does its confusing to me


also tantalum caps are known to short out and go bad, especially cheap ones on equipment thats not derated properly (aka non-goverment)



so on your ceramic board there is 3 suspicious transistors and 2 tantalums that are just a known trouble maker in cheap equipment unless the designer happened to be smart.


the ceramic are better on ceramic PCB, they usually break for mechanical reasons, they are hopefully OK. But rule nothing out if there was a over voltage you know, its like a moose got into a china store. Usually the more complicated parts break first and sink the fault voltage and protect the basic stuff lol

and if there was a over voltage, tantalums are sensitive to that. and they explode when they fully go, it will shatter that board like a small fire cracker put on top of a soup cracker. they have a bad enough reputation that you might want to deal with that, because it can be a ticking time bomb if the circuit was stressed, which it might have been. If a tantalum really explodes it can put a literal hole in a fiberglass board, ceramic will likely get shattered


debugging the board individually is a good idea, if you can manage to hook it up, that is the right way to do it

I tried to clean the lacquer on the IRF110(?), but without success. Maybe I should heat it with hot air to remove the layer.
The one next to it should be a transistor, it is on the pin circuit of the logic board that goes to the relay.
The other 3-pin on the left, from the diagram found online, should perhaps be a double diode, and on the I_sense circuit.
The tantalum capacitor C17 is on the signal that goes to the EN pin of the TDA1060.
If I knew all the values ​​(but I don't), I would want to change all the components of the logic board, or at least those most suspect for the short circuit.

C11 certainly heats up at the moment of the short circuit, from 30° to about 65°.

Do you mean to try not to power the main board with the mains voltage, and try to power the logic board only from the M2 pin (13.7dc)?

[DisasterNow]

if the 3900pF capacitor is shorted or partially shorted, a 10 ohm load on that driver WILL make it click and eventually burn up that resistor, 10 ohm load is rather low, its only 10 ohms for the frequency that is supposed to go through the capacitor, for the welding energy it should be open circuit

Maybe you mean C12 of 4700pF? You can see it well in the third image above, together with R11 and R10: the 10ohm resistor is new, I changed it (even if the old one was ok). While at C12 a lead had broken (bending it), I think I had measured it and it was ok, then I resoldered it.

[DisasterNow]

also if you scrape the goo off the transistor it will be easy to read, the top does not need goo, its just hard not to get it there if its dip coated. ideally they would mask all the chips and keep the goo off the top, but it would be tons of extra work only the repair man can see


keep in mind, goo can make failures worse. it makes it more difficult for a component to vent. what you want is energy going above the PCB. if you have a submerged part, it will explode harder. for electrolytics your supposed to put tubes on them if you pour sealant, so the top of capacitor is unsealed and can open up to vent smoke instead of failing to open up and explode!

if you elevate a part off the PCB with a air-gap, it will burn up and leave the board alone. if its glued to the board it will transfer more heat into the PCB and wreck it!


very high end boards, instead of trying to use the board as a heat sink, they use bigger parts, and elevate them with standoff from the PCB, that way they can happily fail while glowing white hot without destroying the circuit card. I had a ton of board repair work from over stressed transistors that perform no RF function. if they elevate them from board by 5mm with rivet standoffs and solder them there, it would be 100% fine

On this board they exaggerated with the protective lacquer, it could have limited the heat dissipation of many components; in addition it complicated the measurements (I had to scratch all the points to have contact with the multimeter tips).

[coppercone2]

yeah powering it from a lab supply and giving it your own signals, if you can figure it out, its a very powerful troubleshooting technique, if you have sufficent understanding of the circuit.

the 10 ohm resistor in series with a capacitor is a classic snubber circuit (RC SNUBBER). The resistor in parallel with it is just a drain circuit I think, to make sure charge drains off if you turn the machine off. The cap and 10 ohm should not get warm at all unless there is AC current flowing, you can even calculate the dissipation, that is the ESR of the cap and the current flow through 10 ohm resistor.

That component is fairly.. . high energy, being a snubber, you should replace it

[coppercone2]

for replacing those parts, I don't know honestly...

I just don't know enough about how this circuit works to understand why it might be clicking like that. Usually their in current limit with a short like you think.


A potentially bad idea might be to hook it up to a 12V car battery on the offending board, even a power component might blow up or smoke to tell you where the short is.

haha, I wonder what would happen if you put a big contactor on a car battery, hook it up to the welder where there is a short, and see what happens if you give it a momentary contact. or just spark the terminals

[coppercone2]

current limit typically behaves like this

https://www.cui.com/blog/fundamentals-of-power-supply-over-current-protection

look @ hiccup mode

Perhaps the most common implementation for current limiting now employed in power supplies is known as hiccup mode. This mode of over-current protection can be thought of as an active version of fuse protection mentioned earlier in this discussion. With hiccup mode current limiting protection, the output voltage of the power supply is shut down when an over-current situation is detected. After a specified waiting time the output voltage of the power supply is re-established. If the over-current situation still exists, then the supply repeats the shut-down and wait process. If the over-current situation no longer exists, then the power supply continues operating in the normal mode (Figure 6).

[DisasterNow]

yeah powering it from a lab supply and giving it your own signals, if you can figure it out, its a very powerful troubleshooting technique, if you have sufficent understanding of the circuit.

the 10 ohm resistor in series with a capacitor is a classic snubber circuit (RC SNUBBER). The resistor in parallel with it is just a drain circuit I think, to make sure charge drains off if you turn the machine off. The cap and 10 ohm should not get warm at all unless there is AC current flowing, you can even calculate the dissipation, that is the ESR of the cap and the current flow through 10 ohm resistor.

That component is fairly.. . high energy, being a snubber, you should replace it

In the next few days I can try to power only the logic board (I will try only from M2 and from M16 gnd).
R11 from 10ohm heats up due to the short circuit; I did not have C12 at home, when I bent it it also lost part of the blue cover of a rheophore. The ideal would be to start changing various components, such as C12, C11 and the two transistors on the IGBT gate circuit.

[DisasterNow]

for replacing those parts, I don't know honestly...

I just don't know enough about how this circuit works to understand why it might be clicking like that. Usually their in current limit with a short like you think.


A potentially bad idea might be to hook it up to a 12V car battery on the offending board, even a power component might blow up or smoke to tell you where the short is.

haha, I wonder what would happen if you put a big contactor on a car battery, hook it up to the welder where there is a short, and see what happens if you give it a momentary contact. or just spark the terminals

If I wanted I could not use the bulb in series, now after replacing the chips it always stays on. The hope would be that it would burn the part in short circuit, but I would not want to burn other parts instead...

[coppercone2]

why would a short circuit heat up a resistor with a capacitor in series???

If there is a short circuit, the voltage on the rail would actually be LOWER and even less current would flow through the 4700pF capacitor.


The 10 ohm resistor would be hottest in the open circuit condition, where there is the highest voltage across it.

At least that is one simple way to look at it.

But I guess it might be getting hot from the supply turning on and off and making a big fat square wave. So the hiccup ripple is heating the capacitor.?

[coppercone2]

you can disconnect that circuit from the transformer and put DC across it. If any current flows through the 10 ohm resistor @ DC, then the capacitor is shorted

If no current flows, it means its just the hiccup heating it. I don't know much about how those supplies function


and the possibilities for hiccup mode:
1) there is a short and its responding
2) the control loop is detecting a short circuit that does not exist (bad circuit)


Disconnecting the transformer from the circuit can make probing alot easier, its mad annoying if there is a low impedance winding @ DC essentially grounding everything to miliohms of resistance from + to -

[DisasterNow]

current limit typically behaves like this

https://www.cui.com/blog/fundamentals-of-power-supply-over-current-protection

look @ hiccup mode

Perhaps the most common implementation for current limiting now employed in power supplies is known as hiccup mode. This mode of over-current protection can be thought of as an active version of fuse protection mentioned earlier in this discussion. With hiccup mode current limiting protection, the output voltage of the power supply is shut down when an over-current situation is detected. After a specified waiting time the output voltage of the power supply is re-established. If the over-current situation still exists, then the supply repeats the shut-down and wait process. If the over-current situation no longer exists, then the power supply continues operating in the normal mode (Figure 6).

It should be like this, we have this short circuit that is not easy to detect (at least for my skills), no smoke, burning smell, damaged components. Maybe as we said at the beginning of the topic it could be hidden under one of the 3 large heat sinks, who knows. And this short circuit inhibits the output.
Thanks for the input Coppercone 

[coppercone2]

actually I recommend doing that, I fixed some linear supplies when i finally decided to disconnect the transformer and it got alot less confusing and easier to measure

Disconnect transformer and give it a small low current DC signal to see what it does and if it makes sense

it felt much more mentally cleaner to me to measure and understand the circuts