pulse transformer

I think you'd still need some form of clamp (probably a Zener + normal diode series combo) across the primary side of the traffo to limit the voltage less than the breakdown voltage of the SCR. What you're concerned about here is the primary leakage inductance - that energy isn't coupled to the secondary and therefore can't be dumped into the load, so you have to make provision to dissipate it on the secondary side.

You may also need a way to inhibit the charging circuit until the SCR has turned off (probably just for a set time after the GO button is pushed) so that it can't get held on by the charging current - or just make sure that R1 and R2 both limit the charging current to less then the holding current of the SCRs.

tomizett said:

The catch could arise in that if the weld current did not flow, due to a high-resistance joint or something, then the transformer would try to support the full maximum input voltage and quickly saturate - probably blowing the primary. So you'd have to have a way of disconnecting the primary after a set period so that the core could reset.

Click to expand...

That is not going to happen. With ideal coupling, any secondary magnetic field created by the secondary current is opposed and cancelled by the primary magnetic field that the primary curent creates.
Saturation is only due to mangetizing inductance of the transformer (primary inductance measured with secondary open) and is only dependent on the voltage x time product on the primary.

With not ideal coupling I think that for any secondary current there would have to be primary curent provided and then some more for the field lost in the leakage inductance, so primary field will not be completely cancelled by secondary filed and flux density would be higher under load. But I am not entirely sure about this and will try to investigate some more tomorrow, but anyway a core with no secondary at all should not be worse than a core with a loaded secondary.

Hi throbscottle:

Here is a link I found particularly useful in design of the trigger circuit: https://www.onsemi.com/pub_link/Collateral/HBD855-D.PDF It is really a compendium circa 2007.

Unfortunately, it is 240 pages and on the order of 2.4 MB, so I can't post it here, and that link is dead. I tried. Maybe, if you have time, Onsemi has a revised version on its site, or I can send it to you by e-mail via a PM exchange.

John

I'm not quite sure what you mean there tomizett. Are you talking about a multi-hundred volt zener here and connecting them as a HV catch diode? Why won't a plain catch-diode do the job?
It does occur to me that because it's driving an inductor the back-emf might keep the SCR turned on and begin to charge the capacitor in reverse. Could be nasty.
Do you think this is overkill: https://www.ebay.co.uk/itm/Stud-Pha...095868&hash=item41d3be6edd:g:4fQAAOSw3fZaAlw2 ? I can't find info about it's surge current but I guess it's plenty for this project.
How the hell do I work out what that's going to be?

Oooh oooh more replies whilst I was writing my reply! *wave*
Found it John, thanks
It's intended to be able to detect if the secondary is disconnected, so there shouldn't be any issues due to that.
Kubeek I look forward to your investigation...

A normal catch diode would probably work fine - it's just that a diode/Zener combination seems to be more common in the circuits I've seen.
The difference is just between applying a low voltage (just the forward drop of a diode) for a long period, or a high voltage (the Zener drop) for a short period.

Hope I'm not muddying the waters here.

Ok that makes sense. But surely after the zener has stopped conducting, there is still some emf being produced which has to go somewhere. I suppose you just make sure it's at a level that doesn't matter.

The primary will have some leakage inductance and this will resonate with the capacitance of the thyristors at switch off, the common way to deal with this is to use a rc or drc network, the r is valued so that the current through it is the same as the primary current at switch off, and the c is valued so that it'll hold the energy stored.
Its also important to ensure the thyristors trigger symmetrically this way no dc is produced which will overheat the trans and pull loads of current.

I agree that some RC or DRC snubbing would be a good idea, too. The DRC could be another alternative to using a simple diode or Zener clamp.
Remember that an SCR can trigger spontailiously of subjected to too high a dv/dt, so there is a theoretical risk of the back EMF spike re-triggering the SCR. In reality, it shouldn't matter in this application because with the capacitor discharged there would be nothing to keep the SCR conducting even if this did occur. I suppose it might lead to excessive heating, though.

I'm not sure that I follow Dr Pepper's point about symetrical triggering and not generating any DC through the transformer. As I understand it, we're discussing a forward converter, in which the currents are always in the same direction through the transformer - ie, pulsed DC. So long as the core is correctly sized and is allowed to reset between welds (which should be no trouble as they will probably be seconds apart) then there should be no danger of the core saturating.
We did discuss the possibility of this saturation if the secondary was open circuit, but decided that the transformer should be able to absorb that quantity of energy quate happily - as long as it was not done repeatedly, of course.
I may have you wrong - perhaps you could elaborate?

throbscottle said:

But surely after the zener has stopped conducting, there is still some emf being produced which has to go somewhere

Click to expand...

Regarding this, I think you may have misunderstood how the back EMF is produced. When the primary switch is opened, the core contains a certain amount of energy that needs to be removed from it. At that instant, the primary current remains the same and is diverted through the clamp (diode, zener, whatever) and generates a voltage accross it. Because the transformer is now seeing a "reverse" voltage, the current begins to ramp down towards zero - the rate of decrease depends on the inductance and the magnitude of the voltage (which is set by the clamp). The voltage, of course, barely changes untill the current reaches zero. At this point, the area under the volts x amps x seconds curve equals the energy initially stored in the core and the core is now "empty" (reset). So the instant the current reaches zero, the voltage on the Zener (say) will suddenly drop to nothing, and all the back EMFis gone. The higher the clamp voltage the quicker the current drops amd the quicker reset is achieved (more volts, same amps = fewer seconds). I imagine this is why we see higher-voltage clamps in flybacks etc, because a faster reset will allow higher duty cycles (I think?).

All this said, we may be getting a bit hung up on this back-EMF business. Given that we're talking about a capacitor discharge through an SCR, there won't be a sudden turn-off so it may not be any problem. Like me, you are probably used to looking at more typical "hard switching" circuits with FETs etc.

I'll probaly be off-line untill the new year now, but I look forward to seeing how the project develops.

Oops I didnt realise you planned on a forward converter, of course so long as the core can reset symmetry isnt an issue as your operating from Dc

Ok that kind of makes sense. AC theory is still a bit alien to me, tbh - I only know the basics.
I think I'll give the transformer a trial winding and start with the cap at lowish voltage. I don't have an SCR to try yet so I'll have a go at triggering a spark gap. Not much difference for this purpose. Might even be better! Have to be after Xmas now.
Actually measuring the current is going to be interesting.
Right, so it all comes down to the core. So, if it has an air gap I need to worry about back emf because it's really an inductor, if it doesn't have an air gap I don't have to worry because it's really a transformer and the only back emf will be caused by leakage inductance, which will be small. That right? And according to the principles of forward converters the easiest way to deal with that is with an auxilliary winding in series with a diode connected in anti-parallel with the primary.
So what this circuit should be called I think is a "one shot forward converter".

The difference is as much in the way they are connected as they are put together.
Air core or not there will be leakage inductance, it can be an issue with air cores if not designed well.
A forward converter requires a means to reset the core, it can be done with resistor/diode, seperate winding/diode or with 2 switching devices amongst probably lots of other ways.
By one shot are you implying that you want all of the weld energy in one pulse, that would require a massive transformer.

I was beginning to think my "inverter" microwave oven core might be too small. I finally got round to looking up the relationship between joules and watts and did some very rough calculations - scary!
Is it possible for the core to explode if it's too small?
I also found a site with some pulse transformers they sell for inputs of KV's and outputs of many KV's. They're big, but not huge (well, mostly) for similar power to what I'm interested in.
I've collected a few identical clip-on mains cable interference suppressors. If I glue together a stack of these I could have a giant 2 hole ferrite bead. I think there may be enough for 6 to each hole. Is it worth a try perhaps?
Thinking some more about using a spark gap, I realised it could be used to limit the discharge as it will stop conducting when the voltage drops to a certain level. So it could be adjustable. Crude, but ok for experimentation.

I dont think you'd be able to blow up some ferrite, not without a particle accelerator.
However a chunk of ferrite big enough to spot weld in one go would be something you wouldnt be able to pick up methinks.
A few thousand pulses from a much smaller trans ought to do the job, a zvs has enough meat, this core looks like a standard Tv lopty core and is probably running at some khz, lookster:

Merry Xmas, btw!
I'm fairly sure that isn't how these welders work, though. But a cool demo, all the same.
The block diagram at the top of this page shows a full scale commercial version. I just want to build its baby cousin... https://tjsnow.com/welding-machinery/capacitor-discharge-resistance-welding/
I was wondering if an iron-cored transformer might be more suitable as it will slow things down a little?

I dont know for sure however I suspect that the system uses more than just one pulse, it makes sense to store energy at a higher voltage in caps as you dont need ultra low esr caps and hooge busbars if the voltage is higher as the required current is lower, you can then transform it down to the welding current through a converter be it smps or a iron core, the conversion doesnt need to be all in just one pulse, it could be a few hundred to a few hundred thousand pulses, allthough I havent seen inside one of those machines I suspect the method is similar to the latter.
Imagine a high current smps say 5v at 5000amps (what the spotters did for making car exhausts where I used to work), but instead of powering it direct from mains juice, power it from a cap bank, when the cap bank is discharged stop the weld, this way you can charge the caps slowly and dump 1000's of amp when needed without dimming the lights like old skool spotters do.
A couple of the dinasuars I used to maintain had trannies the size of a fridge and were controlled by ignatrons, 'orrid things.
Spotting requires a programmed current curve, if you were to just short a low esr cap into a weld you'd probably blow a hole, esp with a larger size weld.

Merry bloomin crimbo, and 'appy new year.

Yes that makes sense. I've been trying to find a more detailed architecture on the hoofernet. no luck so far...

(edit: Ahh, this is what I need... https://ewi.org/eto/wp-content/uploads/2016/10/Development-of-an-open-architecture.pdf)

(another edit: It says they use a stacked core transformer - laminated I read that as. I wonder if I could convert an old site-transformer (as discussed in a previous thread) they come in sizes from 500VA to 3KVA and up)

(yet another edit: When I started this thread I had thought of using an IGBT but jumped the SCR idea instead. Now I've just found a patent that uses an IGBT so maybe that is the way to go after all)

Couldnt really tell, but that implies all the weld current is just in one pulse, the pulse time must be very short in the order of mS and the transformer must be massive, to store enough energy in the core it must be very large, maybe thats what they mean by stacked.
Interesting weld current is controlled by a choke, one thing that would do is increase the likelyhood of sparks.

In the examples, the "choke" is given an inductance of 720pH. I think it actually represents the secondary inductance of the transformer, especially since it's labelled "Ls"
Anyway I'm going to try using an MOT since I have one lying around with the secondary already removed. It probably has about 230 to 250 turns on the primary which isn't a bad starting point though I should probably take half of them off - reluctant to do so since I may want to use it for something else.

jpanhalt said:

Hi throbscottle:

Here is a link I found particularly useful in design of the trigger circuit: https://www.onsemi.com/pub_link/Collateral/HBD855-D.PDF It is really a compendium circa 2007.

Unfortunately, it is 240 pages and on the order of 2.4 MB, so I can't post it here, and that link is dead. I tried. Maybe, if you have time, Onsemi has a revised version on its site, or I can send it to you by e-mail via a PM exchange.

John

Click to expand...

Is this it?
https://edisciplinas.usp.br/pluginf...d_Design_Considerations_Handbook_HBD855-D.pdf