I’ve spent the last couple of weeks building an arc welder by winding new secondaries onto two microwave oven transformers (MOTs). The one important lesson to learn from all of this is that high power transformers are not a place where the “fuck it that’ll do” method works; I wrecked a couple of primaries so ended up having to find fresh MOTs and start again.
The first step was to find two MOTs; the bigger the better, the power will ultimately be limited by the core saturating. An arc welder is essentially a high power step down transformer; something like 20A at 240V goes in and this then gets transformed to something like 100A at 50V (minus losses). I already had the MOT cores and primaries, all I needed to do was remove the old secondaries and add new ones.
First I cut the welds holding the E pieces to the I pieces. I used an angle grinder but a hacksaw should work; the primaries were going to be re-used, so I was careful not to damage them. To help remove the coils from the cores, I softened the varnish by heating the transformers in the oven at about 200°C for about half an hour. Once the varnish had softened, I knocked the windings off using a bit of wood and a mallet; again being careful not to damage the primaries. Once the windings were off the cores I cleaned the cores up a bit, tearing out the old insulation then using a file to remove most of the varnish and smooth what was left.
Now any old wire isn’t going to take 100A so I bought rectangular section enamelled 3.81×2.54mm copper. I measured the size of the windows in the core and decided that 5 layers of 5 turns each would be the best way to fit around 25 turns on each core. Even with enamelled copper and square section, you need to give at least 20% extra to account for imperfect winding otherwise you’ll find that your lovely new coil won’t fit the core. Winding copper of this size is doable by hand, but requires a decent former, some strength and patience. I made a former from bits of MDF glued and shaved to the same width as the center leg of the core. It is longer than the core both so that the corners can be rounded and to allow better air flow around the transformer to improve cooling. The hight needs to be a bit smaller than the height of the windows minus the height of the primary as the windings will expand slightly when you slip them off the former. To make it easier to slip the winding off the former, I wrapped a layer of thin card around the former and taped it together (but not to the former) before starting winding.
Starting the coil was a matter of poking the end of the copper through a hole in the former and bending it until it laid flat along the end of the former. The copper is rigid enough that no further retention is necessary.
In order to avoid bowing I had to apply a reverse bend to the copper as I laid down each pass. I had to be careful to get each pass to lay nicely in place and keep all the slightly messy beginnings of each overlap where they would be outside the core; my first few attempts with a bodged former and rushing rather than taking my time wasted both time and copper (once it’s been work-hardened once, it can’t be formed again without lots of trouble).
Once all 5 layers were complete I applied a temporary wrapping of PVC tape to protect the enamel while I made final adjustments to the size; I squashed the coil slightly using a vice with wood to protect the coil.
At this point I cut the excess to length, soldered on tab connectors, removed the tape and fitted the transformers together, using OHP transparency as a temporary insulator film.
I G-clamped the cores together, attached some 135A auto battery cable with a ground clip on one piece and a rod holder on the other. I wired the primaries in parallel and the secondaries in series, being careful to keep the outputs in phase. Initially I tried feeding it from a single 13A plug, this blew the fuse as soon as I struck an arc. After feeding each transformer from a separate 13A socket, it worked! While the welds weren’t fantastic, they were limited mainly by my terrible welding technique; there was definitely enough heat to get reasonable penetration and melt the rods. I got a little carried away and welded until the vibration damaged the insulation on one primary enough to short irreparably through the core. I tried using the remaining transformer on its own; it could just about melt the rod, but didn’t get any real penetration and was very difficult to strike an arc. It also overheated the transformer to the point of the transparency catching fire; this killed the other primary.
Luckily I managed to get a couple of transformers about a week later which had cores only slightly bigger than the originals. I expanded the secondaries slightly by putting a couple of bits of wood inside and hammering screwdrivers in between as wedges. This time one transformer had a thermal cutout in series with the primary; after looking at the burnt out transformer and confirming my hunch that the primary winding (that was being driven at something like 4 times its intended current) had been the point of ignition, I attached the cutout to the inside surface of the coil. I also wrapped the cores and the coils in high temperature kapton insulating tape to avoid short-circuit problems without causing another fire hazard.
I assembled the transformers (again holding the cores in place with G-clamps), placed them on a workmate, wired them as before and added a microwave oven fan.
I fired it up and was able to easily strike a nice hot arc; I gave it a quick test by welding together some old steel buckles I had lying about. The weld isn’t very pretty, but is nice and strong – I’ve whacked it with a hammer and it won’t break.
Next I wanted to see how delicate a workpiece I could weld; for this I devised the idea of the Swiss Army Teapot: a teapot with an assortment of screwdrivers/allen keys/small tools attached. This quickly proved that without current limiting my welder is far too powerful to weld thin sheet steel without blowing holes.
As it stands, it is a pile of components attached to a workmate with no control past which plug sockets I turn on. This is obviously not a finished product, but is a perfectly functional welder. The to-do list currently stands at:
- Weld the cores together; this will reduce core vibration and make the transformers easier to handle.
- Seal the transformers with varnish; this will improve the insulation and cut down on the vibration of windings.
- Adjustable current limiter; I’m thinking saturable core inductor at this stage, built from another MOT core.
- Case; it can’t live clamped to a workmate for ever.
- Single 13A feed; once I have a current limiter, it should be possible to run at lower powers from a single 13A plug. I could have the option to add a kettle lead to give extra power for higher power operation.
- Enforced duty cycle; currently the welder can be used until the thermal cutout cuts power to one primary when it reaches 160°C. It would be nice to have an enforced duty cycle with an LED readout to display how close the limit it is and how long it’ll need to cool.