I started this project with the idea of building a high current regulated power supply. I didn't have an immediate need for such a device, although most of the major components were to hand, and it is useful to have a range of different power supplies available when working with electronics. While working on it, another project required some parts to be welded together, although suitable welding equipment was not available. At this stage, the only completed part of the power supply was one of the transformers. This transformer was tried on an experimental basis as a current source for arc welding. Initial results were poor, due to both the limited output voltage (~15-20V), and my limited welding experience. However the utility of having an arc welder was immediately recognised, and therefore I decided to redirect the project into building a dedicated welding power supply.
The requirements of a basic arc welder are fairly simple - a power source of about 40V open circuit, able to supply around 100A on an intermittent basis, with a fairly poor regulation characteristic in order to survive short circuit conditions. This is easily met in principle with a suitable transformer, however such a component would be relatively expensive. Instead I used a rewound microwave oven transformer (or 'MOT' as they are known).
When selecting transformers for this application, it is best to use ones out of large, high power microwave ovens. (See the Microwave Oven FAQ for information on safely dismantling microwave ovens.) The high voltage winding can be removed by cutting off the exposed portions with a hacksaw. The remaining section can be removed from the core with a hammer and punch. The filament winding can be unwound by hand, as it is only a few turns.
Secondary winding cutting details
I have seen microwave oven transformes rewound using ordinary PVC insulated cable. This is not ideal, both due to the space taken up in the core by the insulation, and the inability of the PVC to withstand high temperatures. I decided to use a strip of sheet aluminium, insulated with adhesive tape. This can be purchased at hardware shops as aluminium flashing. The short, wide roll of flashing can be cut with tin snips and folded into a long, narrow strip as shown in the picture above, without the need for any joins.
For the insulation, polyimide (Kapton) tape would be the best choice, but I did not have a source of this at the time, so I used paper masking tape instead. Although not particularly suitable for high temperatures, it should withstand heat better than PVC tape. High insulation strength is not a requirement, as there is only a volt or so between turns. (Polyimide tape can be sourced from Dealextreme.) Only one side of the strip requires insulation, but you need to make sure both edges are well insulated. The new secondary was installed by simply feeding the strip through the window in the core until it was full. This took around 16 full turns. The transformer gives around one volt per turn, so a second rewound transformer was connected in series to obtain sufficient voltage.
New transformer secondary winding
Amongst my collection of junk I had what appeared to be a data terminal server of some sort. Though the precise function of the unit was unknown, it had a heavy metal case with ample room to house the components of the welder, as well as a nice clip on lid. (Look closely at the pictures, and you will see it is what is probably the world's only IBM-compatible welder!) I installed the transformers in the case, together with some cooling fans, and a 10A circuit breaker in the primary circuit. The handpiece was fabricated from an offcut of firewood, together with a heavy duty brass terminal strip to hold the electrode.
I decided to bond the return terminal to the metal case, and hence to mains earth. Although the primaries and secondaries of the transformers are physically separated, the secondary insulation is less than optimal, and it was considered desirable to protect against possible insulation breakdown by earthing the secondary. However, when using this arrangement, care must be taken never to weld on an earthed object, as a portion of the welding current could flow through the mains earth wiring, potentially overheating it.
Initial testing of the welder revealed difficulties in controlling the magnitude of the welding current. I had a large bridge rectifier assembly out of some industrial equipment, complete with SCRs in two of the arms. I connected this to the the secondary circuit, and built a simple phase-control circuit to trigger the SCRs. I tried to make the circuit regulate the firing angles in order to provide a constant current characteristic, but could not get this to operate satisfactorily. Therefore, a simpler voltage-based control strategy was used. A front panel knob was provided for adjustment.
The rectifier of course provides a DC output, which is said to provide better performance. Usually, the electrode is made negative, and the work positive, in order to direct more heat into the workpiece. I also experimented with connecting some filter capacitors across the output, but found the intial striking of the arc to be too violent, and so the original configuration was reverted to.
Internal view of welder. The four rectifier heatsinks are at bottom left. The two transformers are to the right. The capacitors at the top are not currently in circuit.
It is possible to construct an arc welder at virtually no cost from scrap components. The unit I constructed is a little underpowered, but it has still proved useful for a number of welding jobs. If you are wanting to build a unit for heavier duty work, it might be worth using more transformers, with thicker secondaries. Dan Hartman has built a welder using a total of 8 MOTs, which looks like quite a capable unit.
This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.