What does this transistor do?

[cybermaus]

So this is a controller for a old MIG welder.
The 555 circuit on the right, together with the two darlingtons, is providing a 1KHz PWM control for the speed of the welding wire feed.
The 555 circuit on the left is for pulsed welding, so a pulse of 2 to 0.5Hz, so it switches on-and off the welder (and the PWM)

I cannot for the life of me figure out what the function of the top transistor is supposed to be.
It looks like a standard half H bridge, but if you look better, it does not fit, because there is no need to feed +30V into both sides of the motor.
Under ideal conditions, it turns on when the bottom turns off, like a H bridge, so not a problem. But what is it supposed to do?
Some snubber for the motor? The diode already does this. It does not drive anything? Well, in fact, it drives a short when the bottom transistor also turns on.

It is a problem, because everything is not ideal, the reset signal of the 555 takes some time to propagate, and during the transistor swapover it briefly shorts.
This eventually burns out the the transistors.  If a simply leave out the top transistor, everything simply works.
Which is BTW also how I found the problem: After decades of working well, the welder stopped, and I discovered both transistors burned. I replaced them both, and it all worked a few days, and then burned again

So again, if I leave the top transistor off, it all works.
And now I suspect that is how it all worked all those decades. Top transistor was probably broken all this time, I should only have fixed only the bottom one

But what is the top transistor even supposed to do ?!?

See attachment

Also I have been playing with falstad simulator.
https://tinyurl.com/2f8poj3o

[Andy Watson]

I cannot for the life of me figure out what the function of the top transistor is supposed to be.

I think it is supposed to halt or reduce the wire feed by stopping the motor in proportion to the "pulse off" period. You wouldn't want to be feeding wire into a cooling melt puddle. It's a somewhat brutal method of stopping the motor!

It is a problem, because everything is not ideal, the reset signal of the 555 takes some time to propagate, and during the transistor swapover it briefly shorts.
This eventually burns out the the transistors. 

Indeed - brutal
Have you considered that there may be some other component failure that is allowing too much overlap of conduction between the two transistors? Are V21 and/or V31 functioning correctly?

If a simply leave out the top transistor, everything simply works.
Which is BTW also how I found the problem: After decades of working well, the welder stopped, and I discovered both transistors burned. I replaced them both, and it all worked a few days, and then burned again

So again, if I leave the top transistor off, it all works.
And now I suspect that is how it all worked all those decades. Top transistor was probably broken all this time, I should only have fixed only the bottom one

Perhaps it's a design error! Perhaps the top transistor blows before it leaves the factory and most users just compensate by turning the wire-feed down a notch?

[magic]

It looks like a standard half H bridge, but if you look better, it does not fit, because there is no need to feed +30V into both sides of the motor.

I wrote a different answer initially as if the PNP were driven by PWM rather than the pulse controller, this is deleted now

Andy is right; short circuiting the two ends of a motor stops it.

[cybermaus]

I think it is supposed to halt or reduce the wire feed by stopping the motor in proportion to the "pulse off" period. You wouldn't want to be feeding wire into a cooling melt puddle. It's a somewhat brutal method of stopping the motor!

You mean like an active brake right. Dropping V+ on both side with some oemph behind it to stop it by force rather then by freewheeling?
Does that even work? Is such an concept normal?
Never seen that done before, but this is not my area of expertise.

Indeed - brutal
Have you considered that there may be some other component failure that is allowing too much overlap of conduction between the two transistors? Are V21 and/or V31 functioning correctly?

Considered? Yes. Figured out which one? No. The 555 (actually, they used a 7855 clone) seems to work well.

Perhaps it's a design error! Perhaps the top transistor blows before it leaves the factory and most users just compensate by turning the wire-feed down a notch?

Tempting thought
But no. I seen variants schematics, all using the same. So it would be a very widespread mistake.

[magic]

You mean like an active brake right. Dropping V+ on both side with some oemph behind it to stop it by force rather then by freewheeling?
Does that even work? Is such an concept normal?
Never seen that done before, but this is not my area of expertise.

It's not about applying V+ because the motor doesn't care about common mode voltage. It simply is about shorting the two sides together.

Try to short the motor with a wire (with everything powered off) and turn it by hand, it should become much harder. The motor should also be difficult to turn in reverse direction, because it's shorted by V20.

[AVGresponding]

I think it is supposed to halt or reduce the wire feed by stopping the motor in proportion to the "pulse off" period. You wouldn't want to be feeding wire into a cooling melt puddle. It's a somewhat brutal method of stopping the motor!

You mean like an active brake right. Dropping V+ on both side with some oemph behind it to stop it by force rather then by freewheeling?
Does that even work? Is such an concept normal?
Never seen that done before, but this is not my area of expertise.

Indeed - brutal
Have you considered that there may be some other component failure that is allowing too much overlap of conduction between the two transistors? Are V21 and/or V31 functioning correctly?

Considered? Yes. Figured out which one? No. The 555 (actually, they used a 7855 clone) seems to work well.

Perhaps it's a design error! Perhaps the top transistor blows before it leaves the factory and most users just compensate by turning the wire-feed down a notch?

Tempting thought
But no. I seen variants schematics, all using the same. So it would be a very widespread mistake.

It's a common way of electrically braking a motor. There are a couple of things you could try, to reduce the chance of melting more transistors.

Increase the value of R24. Mechanical friction in the system probably means you don't need much of an electrical brake anyway.

Pull up resistor on the base of the top transistor, pull down resistor on the base of the bottom one.

[cybermaus]

@magic

.

I could use some more of your assistance.
I will try to fix it again. (it burned again)
If only because I should be able to!! But in the mean time, I am also considering a more no-nonsense path.

The only sellers of this board I found are Polish. In fact, they make a modernised replacement board.
https://allegro.pl/oferta/c-2731-274-1-plytka-us25-us-25-bester-magster-nowa-7599389619

Now, they do not sell to Netherlands. I plan to call them on monday and see.
But this appears as some sort of marketplace to me, I am not sure I am able to figure out who is actually the seller or their contact. I mean, I am OK with buying it from that market place, but they'd have to make an exception. They ship other stuff to Netherlands. Maybe it's the seller himself who does not ship abroad.

Being Polish yourself, are you able to figure out who to call or email? The actual seller I mean?

BTW: board search string is: Bester C-2731-274-1

[dietert1]

The problem should be easy to fix, once you know the size of the motor and its DC resistance. Apparently there is no real brake resistor to absorb the motor kinetic energy. The 0.1 Ohm current limiting resistors R23 and R24 (if they really have that resistance) are wrong and likely responsible for failure of the transistors. I'd gues one wants at least 2 Ohms or so. These resistors will limit cross conduction and are necessary even if the motor resistance is big enough.

Regards, Dieter

[magic]

This site is an eBay of sorts. The actual seller appears to be this company:
https://plytkadomigomatu.pl/pl/p/US-25-B-3731-130-1-C-2731-274-1-US25/6
https://plytkadomigomatu.pl/pl/i/Kontakt-i-dane-firmy/9

The board appears to still be just a pair of NE555 with different power transistors. The devil is in the details, of course.

Note that they say you will need to drill three new holes in the chassis to screw the board in.

[cybermaus]

The problem should be easy to fix, once you know the size of the motor and its DC resistance. Apparently there is no real brake resistor to absorb the motor kinetic energy. The 0.1 Ohm current limiting resistors R23 and R24 (if they really have that resistance) are wrong and likely responsible for failure of the transistors. I'd gues one wants at least 2 Ohms or so. These resistors will limit cross conduction and are necessary even if the motor resistance is big enough.

Actual resistance is 0R56, so a good 5 times higher.
Also, in my board, the actual Darlingtons pairs have been replaced by FET's IRF9540 and IRF540
So it seems some improvements have already taken place

[dietert1]

When using mosfet switches, the 3V zeners V21 and V31 have to go. Or you replace them by 15V zeners. Otherwise the mosfets won't switch properly. And with mosfets the current limiting resistors should really be more like 2 Ohms or so.

Regards, Dieter

[David Hess]

The top transistor is the brake for the motor.