The commercial welder
The views above show the electronics box, electrodes and welding table I'd made for the commercial prototype.  I'd gotten the cost of the unit greatly reduced - this would have retailed for $400 US, but for one slight problem...  More on that later.  The big step forward was finding a cheaper way to control the inlet power to the welding transformer.

I'd thought initially that a variable reluctance transformer could be the way - that is, the welding transformer itself had a varaiable air gap in its magnetic circuit.  Mechanical control would be required.  After spending a while fooling around with this and getting nowhere - it had to be a reasonable cost, jam-proof, and resist wear and tear while also controlling an air gap of less that 0.5mm - I had a better idea.

A 500W lamp dimmer switch.

This has worked nicely to replace the variac used previously.  It works by rapid on-off switching of power into the welding transformer.  The transformer only gets a portion of a full mains supply cycle, controllable by the dial visible on the top left photo.

The rest of the welder would largely have been commercial parts:  the plastic box, microswitch, fan, inlet, power cord, fuse and holder, and so on.  The transformer would have been made in a limited run by my local motor rewinder, I had a carpenter making the welding tables, and I could have got an engineering shop to produce the electrodes without much trouble.  The whole thing was to be produced to Class II insulation, which effectively means two seperate layers of insulation between the user and any live part.

You're probably asking:  It looks good, and after all this effort, why aren't these being sold?

Two reasons:  a basic arc welder can probably be used without much trouble, and these sell for much less than I could have sold the unit shown above.
The other reason is that any electrical unit to be sold commercially must - MUST - be safety tested.  This is a product type test.  The testing laboratory take one sample and effectively destroy it by abuse testing.  The test costs between $3000 to $8000 here in NZ, and the results are only applicable to the Australian and local markets.  For the US and UK, it'd have to be tested by a lab that had the appropriate license, in other words, there'd have to be another test and another fee.

So, for a product run of about six units for Australia and NZ, I'd have to come up with the testing fee up front, and then try to recover that in the unit's price - i.e., at least another $500 (possibly up to $1400) put onto the price of an item already priced at about $800.  That is, if it passed the test first go.

Anyway, here's a look at the inside of the welder and the transformer used with it:
The transformer uses two completely seperate coils.  The one shown is a prototype.  This was for safety - for a mains voltage to reach the exposed electrodes, the insulation would have to fail on both bobbins.
The circuit used is just like the original welder, except that the dimmer switch is used in place of the variac.  Some things to point out:  The holes at the rear of the box for the fan's outlet, the use of plastic screws and bolts, and that the spade lugs for connection to the electrodes had not yet been fitted to the welding cables.  The frame of the transformer was earthed.  I'll repeat the safety warning:  If you're going to build one of these for yourself, PLEASE DON'T GET ELECTROCUTED.  Get it checked before letting it anywhere near mains power.
Next page:  The story of the armoury