Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

pulse transformer

Status
Not open for further replies.

throbscottle

Well-Known Member
I have been re-visiting the idea of building a capacitive discharge welder such as used for battery tabs, for which 1F car audio "stiffening" capacitors are popular, but seem to have got rather more expensive since I last looked. One of these should store about 72 joules at 12v. I've seen a project for one that uses 220,000uF at just over 20v, so that stores 44 joules and looks to be the minimum needed. So aiming for the higher amount of energy within that voltage range, no matter what value/voltage combination I use I can't build a suitable bank of capacitors for less than £25-£30
I did read somewhere that the commercial ones use a smaller, higher voltage capacitor and a pulse transformer to drop the voltage, and found that I can build a smaller, higher voltage (eg 100v, 200v) bank that will store the same amount of energy, much cheaper. So I would just need a transformer to drop the voltage to a suitable level.
Any thoughts on how big a core I would need? Bearing in mind it's only to weld thin metal like battery tabs and emi shields.
 
Etd series transformer kits might be a good start, usually the dominant size factor is operating freq, however wire size in this application need to be taken into account, you'll want as low a resistance as possible, therefore heavy guage copper conductors, Etd44 would be one to look at.
Dumping all the energy from the cap/s in one pulse would require a hooge transformer, probably better to use a high freq smps chip to dump the energy from the cap to the electrodes.
Commercial spotters have settings for weld current profile, maybe you could do something like this with a microcontroller and current transformer.
Car audio caps might not give good Di/Dt characteristics, but ought to store a fair amount of energy, watch out for cheap junk.
 
Here's another approach: **broken link removed**

He uses LiPo batteries sorta like capacitors and gets high current+ high joules without expensive capacitors. It is offered as a kit. There is a very long thread here that describes its development: https://www.eevblog.com/forum/projects/guesses-on-what-i-am-attempting-here/msg1232857/#msg1232857

The welder I built several years ago uses about 600,000 uF and goes to >18V. I use it for battery tabs and fixturing thin-wall SS tubing prior to brazing.

I think the LiPo approach is clever, but that requires the ability to turn off the current. Of course, a simple CD welder does not need that function.

John

PS: I have not compared the cost of the LiPo batteries to what I paid for my name brand capacitors on eBay. I don't think there would be much savings, but I haven't checked capacitor prices lately. One big advantage of the LiPo is portability.
 
Last edited:
Ferroxcube ETC44/22/15 - I have a transformer from a Panasonic "inverter" microwave oven which is a similar size - I wonder if the core of that would do? I don't know how to treat the specification of ferrite in a meaningful way so I'm blind either way...
That's quite a remarkable system there, John!
I wondered about switching it with a spark gap. Save having to have expensive semiconductors...
 
Here's the unit I built about 10 years ago, after quite a bit of development:
Completed welder.jpg


Except for what I learned doing that 10 years ago and the fun of doing a little machining, I would buy KWeld's device in a blink today. Only the electrodes we developed independently are almost the same, and doing it again, I would do that differently. Namely, I would make both electrodes from tig welding electrodes in copper holders and put them in an over-center clamp for battery tabs.

BTW: The author tested various LiPo brands, and there are serious differences. The brand he chose is relatively cheap, and I would follow his recommendation.

John
 

Attachments

  • Electro Online post_011508.txt
    21.2 KB · Views: 117
Very nice piece of work John. No way could I afford the K-weld kit, even if I could I'd never use it often enough to justify the cost!
By upping the voltage I can get the cost of capacitors down to under £10 now since Multicomp make a 470u 400v cap. I can rectify mains directly and get about 350v, and use 2 or 3 in parallel (or a voltage doubler and 2p2s caps for more juice)
Did you use an SCR in the end?
 
Yes on the scr. It is on the ground/low side. The main thing I paid attention to was getting a good turn on. There are some papers (Semikron as I recall). Here's a snippet from my write up:
jpa said:
Almost any high current, high pulse-current SCR should work. These can be obtained at relatively low prices on ebay or from other industrial used/surplus dealers. Circuit development was done with a Semikron SKKT 92/16E SCR. The final design used a Powerex T500108005AQ ( ).

OOPs, neither link is good anymore. Sorry. That's why I insist on paper from my financial institutions.

John
 
Where I used to work, we had a projection welder with a maximum energy of around 2.5 kJ. We used it for welding the lids onto DIL-14 oscillators, like this:-**broken link removed**

The capacitor voltage was around 2500 V. That works out at 800 μF.

The capacitor was discharged into the primary of a transformer using an SCR. The secondary was a single turn, and the wire from that to the welding dies was about 500 mm2, and the peak current was somewhere around 100 kA

So it is possible to have a high power pulse transformer.
 
Now I'm no expert on magnetics, but the idea seems feasible - especially given Diver's post above.

As far as I can see, you'd have to design it as follows:
1. Size the primary turns and core to stand the maximum input voltage (from the charged cap bank) for the expected maximum duration of your weld.
2. Select secondary turns for a sensible maximum output voltage, and wind using the biggest wire you can find.

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.

It's interesting that the kWeld makes a point of applying a known energy to the weld, which of course results in a known temperature rise in any given weld area. But, of course, a capacitor charged to a known voltage also contains a known amount of energy so, to a first aproximation, you'd think the same could be acheieved by simply altering the charge voltage (as I presume John's unit does).
The difference would be due to the time over which the energy was spread and the rate at which the surrounding metal could conduct it away. I wonder how significant this would be?

Also, I would have thoguth you'd want to keep the output cables loomed closely to minimise inductance and potential harm from the generated magnetic field (you'll have seen mikeselectricstuff and the destruct-o-tron?).
 
I think that it would be very difficult to have a situation where an open-circuit load would blow the transformer. I agree that an open circuit load would cause the transformer to saturate, so then there would be little back emf, and the primary current would be large.

My point is that the capacitor would then soon be discharged, so there is little point trying to turn off the primary current as it will stop of it's own accord very soon. The energy in the capacitor could end up entirely as heat in the primary winding, but how hot would that get the winding? According to https://www.engineeringtoolbox.com/specific-heat-capacity-d_391.html, it takes 385 J to raise 1 kg of copper by 1 °C. So if the welder that I was using had 500 g of copper in it, (which I suspect is a gross under-estimate), 2500 J would raise the temperature by 13 °C. That's hardly going to burn the transformer out.

If you have a very fast repetition rate, you could get a problem.
 
Yea, you're right there, thanks for checking the numbers - I was probably worrying unduely.
You'd just need to make sure you took care of the back EMF then.
 
Woo-hoo! I just ordered a Kemet 4700uF 400v cap from a seller on fleabay. Looks genuine, though the part number is arranged differently to how the datasheets have it, looks like they are left over from a manufacturing order so I guess that's why, also very cheap only £10!
So now I need to find a big SCR. I realised the spark gap isn't such a good idea as the back-emf would also flow through it...
This is now getting officially slightly scary...
 
Nice piece of work Jp.

Its academic now however your microwave trans, it probably has an air gapped core, you can find out by pulling the core sections out and holding them togther, see light? its air gapped, this type of trans or coupled inductor is for a flyback converter, for something like what you need either a push pull or a forward converter.
 
Ok now your got me reading about air gaps. So far it looks a desirable thing to have. Are you saying I should just use a bigger core instead?
 
Air gaps are popular in flyback supplies as it means you can use a smaller core and possibly less copper, therefore less £££'s.
They say the energy is stored in the airgap, however there are parasitics, and you need to be carefull in design to stop them being dominant.
Theres a russian geyser who's done some calc software, the push pull calc is called excellent It, and the flyback is called, well flyback, I think I found them on the esp audio forum.
 
I'll just have to experiment. Probably after Xmas now. Interesting video. I'd wondered about using a car battery, never had a spare one to try. I have a feeling this core will be too small in any case.
It occurred to me a way to get extra pulses (cleaning, pre-heat) could be to use an extra winding driven by a smaller cap.
I also thought with another extra winding I could put a signal on it, and detect the level of that so it won't fire if the welding contacts are open circuit.
It occurred to me a method to limit the charging voltage, but that means the output voltage will also be changed. Then I thought about firing through different taps on the transformer so the output voltage doesn't suffer too much (perhaps having an SCR connected to each tap). Or would it be better to connect a bank of mosfets in series with it and trigger those with an adjustable-width pulse?
 
It occurred to me a method to limit the charging voltage, but that means the output voltage will also be changed.
My gut feeling is that this doesn't matter, as long as there is still plenty of voltage to drive a good current through the weld (or, conversely, as long as the resistance at the weld is not too high). I think it's the overall energy that should matter, and as long as the weld resistance is low enough that the charge left in your capacitor is insignificant, the starting voltage won't really matter.
However, maybe it's not the simple?
 
Hi tomizett - thanks for that, I was hoping it would be the case. I'm thinking I should be able to vary the input voltage over a range roughly 150 to 350 volts. Suppose my transformer has a ratio of 1:10, that gives a range of output 15-35 volts. I think 35 is too high and will spark! On the other hand, I suppose the voltage will get pulled down by the near short-circuit condition of the weld anyway. Maybe I will have a "high" and "low" setting...
What happens if I don't have a catch diode for the back emf? Will all the energy just get dumped into the weld anyway? It seems that way in my head. Maybe use a snubber instead? Or as well?
 
Something like this. I took a leaf out of John's book and put the SCR's on the low side.
 

Attachments

  • Schematic_Design__cdw.pdf
    19.5 KB · Views: 215
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top