Thyristor gate driver problem

[Salvador12]

Hey folks. please help me out, I am having trouble finding the problem with a thyristor driver circuit. So this circuit has been working for many years without problems. It works from 2 of the 3 phases, so 400VAC (I am in Europe). The thyristor is in series with a large coil, aka electromagnet located inside a box attached to a sand separator. The box works like a vibrator. It can be powered directly from 2 phases but in order to lower the vibration amplitude the thyristor circuit is installed.
Recently the thyristor circuit started working intermittently. It works for about 5 mins then stops for about 5 mins or so.
I took the whole circuit box home, checked every part, measured the capacitors (with capacitor tester), measured the transistors, checked the circuit board, connections everything, and it was all fine.
The problem still continues. Then today I was measuring the circuit as it was working LIVE. The thyristor gate driver is supplied from a rectified but unfiltered DC supply of 25 volts, this sags to 22.5 roughly as it begins working.
When the gate drive circuit stops working and the thyristor shuts off, I noticed there is a voltage of 25 VDC across the pulse transformer primary (measuring across the 1N4007 flyback diode) The transformer primary has a resistance of some 2 ohms or less. It is a small gate drive transformer. Clearly a voltage drop of 25 volts across the transformer isn't resulting from current flow through it but most likely I'm measuring from - terminal to which one side of the primary is attached through the unijunction transistor across the 220 ohms resistor connected to + side.

Now here is the weird part. When the gate drive circuit has stopped working, as i touch the cathode side pin of the primary flyback diode (1N4007) with a screwdriver or multimeter probe, it immediately starts working.
Clearly as I touch it even though with an isolated probe, there is a small change in capacitance introduced by my body and this is enough to get the drive going again. Then it stops after a little while again, touch it once more it works again.

It seems to me this could be some kind of a problem with a trimpot maybe changing it's originally set resistance (the two yellow ones are on board 3 pin potentiometers ) or maybe one of the capacitors is bad, although I checked all of them. I even changed the 2N2646 unijunction transistor simply because I had one and I wasn't sure whether the original measured correctly.

Any suggestions? I attached my drawing of the circuit

 

[Salvador12]

Now I got the idea that I could just bypass the bc107 transistor and connect the +ve directly to the 60k potentiometer as the UJT 2N2646 firing frequency is determined by the 0.68uF capacitor and resistance from diode bridge + to the emitter which is made from the potentiometer in series with the 1K resistor. Can anyone please explain me why that schematic has the bc107 transistor in it in the first place? I am not sure what role does it play , it seems to me it makes the circuit only weirder because now the time constant is set by competing current paths that do the same job.

 

[betwixt]

The trigger rate is set by the voltage rising on the emitter pin. The time constant is set by the 0.68uF capacitor and the components feeding current to it, these are the 1K resistor and 23K resistor and additionally the current path through the BC107 and potentiometer. The BC107 initially has no base-emitter voltage because the 1000uF capacitor holds no charge, Therefore, the BC107 doesn't contribute anything and the potentiometer is disconnected. As the 1000uF charges to about 0.7V, the BC107 conducts and effectively becomes a short circuit, allowing the potentiometer to control the pulse rate.
So it looks like a soft-start circuit but I'm not sure how well it would work when fed from rectified AC with no smoothing. In fact running any oscillator from a pulsed supply makes it difficult to predict its output.

Something looks odd about your measurement, if you measure 25V across the transformer primary, and assuming this is DC or the Diode would fry, and it has 2 Ohms resistance, ignoring reactance, something like 12.5 Amps is flowing through it. That is impossible when the source is 170V and fed through two 3K3 series resistors.

I suspect your problem is an intermittent connection on the transformer primary and applying slight pressure when taking the measurement is closing the break. My first step would be to resolder all the joints and check for hairline cracks in the wiring.

Brian.

 

[Salvador12]

The trigger rate is set by the voltage rising on the emitter pin. The time constant is set by the 0.68uF capacitor and the components feeding current to it, these are the 1K resistor and 23K resistor and additionally the current path through the BC107 and potentiometer. The BC107 initially has no base-emitter voltage because the 1000uF capacitor holds no charge, Therefore, the BC107 doesn't contribute anything and the potentiometer is disconnected. As the 1000uF charges to about 0.7V, the BC107 conducts and effectively becomes a short circuit, allowing the potentiometer to control the pulse rate.
So it looks like a soft-start circuit but I'm not sure how well it would work when fed from rectified AC with no smoothing. In fact running any oscillator from a pulsed supply makes it difficult to predict its output.

Something looks odd about your measurement, if you measure 25V across the transformer primary, and assuming this is DC or the Diode would fry, and it has 2 Ohms resistance, ignoring reactance, something like 12.5 Amps is flowing through it. That is impossible when the source is 170V and fed through two 3K3 series resistors.

I suspect your problem is an intermittent connection on the transformer primary and applying slight pressure when taking the measurement is closing the break. My first step would be to resolder all the joints and check for hairline cracks in the wiring.

Brian.

Did some tests today, the pulse transformer primary becomes open circuit as it heats up... so yeah the traffo is toast. I hoped to rewind it but it's epoxy sealed and as I took it apart it basically broke. Now I wonder can I replace it with any general purpose pulse transformer?
This one had a silicon steel laminated little core. Winding DC resistance was about 2.5 ohms, and showed roughly 4mH when measured @1Khz.
I wonder can I just as well put in a ferrite cored pulse traffo?

Mine was an old WSW 080-2-368

 

[betwixt]

I can't see the physical construction of the pulse transformer but the windings themselves are very unlikely to break, far more likely is the solder connections to its pins. You might find a prolonged touch with a soldering iron will reflow the joint.

Without a full spec it is difficult to suggest a direct replacement but I would guess one with similar physical size would work the same.

Brian.

 

[hiphei]

I don't think ferrite core replaces lamimated iron core.. powdered iron rings are closer but still not the cigar..and a pain to wind to 4mH.. The soldering iron tip may just help..been there done that succesfully, but also have had it not helping..i would try..