Cooling of overloaded transformer

Yes, the capacitor is between the inductors as coupling. It's optional, but the inductor is smaller with.

Incidentally, the ground return of a Cuk carries the sum of input and output currents. Duh, but since the voltages are opposite, the current is the sum of the magnitudes, i.e., a 5V to -5V, 1A converter delivers 2A through its ground node.

All three nodes (in, out and GND) carry continuous current (just inductor ripple).

You might then suppose, well, let's ground the output and allow the GND return to lift. Well, after some rearrangement, you can reduce this to a single winding inductor, and you've just invented a buck converter (and full switching ripple is back, because of the sum of input and "output" currents). :-)

I wouldn't think there would be anything wrong with DC for spot welding.

Just build a bigass multiphase buck, dump all the outputs together, don't even bother with output capacitors. (You need a current mode controller to do this.) A CPU VCORE controller might do the job. You still need a shitton of 12V to get there, of course; if the controller doesn't mind what the supply voltage is, you might do 160VDC (rectified 120V, or higher for that matter), with a fairly low switching frequency and awfully low duty cycle.

Bonus points for tacking on PFC, so you can draw several kW in the seconds before the breaker pops. Actually, just run it constant duty (with a fixed peak current limit to maintain protection) -- pulsating power is no worse than usual.

Alternately, use supercaps (and maybe a 24V bus instead) to deliver much more power than available (over the same time frame) from the outlet, and also enable "portable" applications. Hmm, you'd only need about a dozen Boostcaps, that's not all that horrible. It could actually be luggably portable (if not handheld).

Might also be good to use a current multiplier for the output:

Just the series-parallel transformation of a Cockroft-Walton ladder. (Incidentally, the inductors have to be separate, you can't really share any on a common core.) Doesn't seem to be any reference to it on the internet; perhaps I'll call it the Williams ladder.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

he's got a wife ;)

Toast her first.

That's all about arc welding. IMLE ac gives a lot more spatter. Spot welding is another matter altogether.

NT

no splatter with TIG, but yes spot welding another matter altogether

I'm having trouble imagining a wife who doesn't want pan-toasted bread, but is okay with a 220-to-110V transformer with intercoolers, capacitors, and who knows what else sitting next to the toaster.

Michael

assuming you mean Benefit of DC.

You can store it in capacitors.

so you can parallel a bunch of ultracaps and run the welder off a wall-wart.

--
This email has not been checked by half-arsed antivirus software

yeah, you can't have fewer than 1 turns.

--
This email has not been checked by half-arsed antivirus software

Sort of. You can use a multi-leg core as a turns divider. You still have to have one turn on something, but the main winding can be effectively two or more turns per turn of that secondary.

The trick is to put a balancing winding between all the legs, to enforce equal flux. Without, the legs act independently (given that the sum equals the flux through the center leg, the main winding).

For a spot welder, I suppose you'd have one turn on each limb, and wire them all in parallel; this would affect flux balancing, so you don't need anything extra.

Leakage inductance is worse, because the windings aren't on top of each other. The textbook schematic of a transformer with windings on either side of a core, is a pretty awful transformer indeed.

It's interesting, for any low-voltage application: you only need as much core cross section as a single turn secondary requires, at the operating frequency. You might design an isolated Vcore forward converter this way, a one-turn secondary that needs very little flux indeed, at a fairly high operating frequency. The ideal core has a wide winding area with a thin center limb, allowing ample space for windings. This would be a good application for a planar transformer, but available cores aren't usually so exaggerated in proportion -- they might be good for 5V or more on a single turn.

Tim

--
Seven Transistor Labs, LLC 
Electrical Engineering Consultation and Contract Design 
Website: https://www.seventransistorlabs.com/

Can you take it apart and rather than havong the elements for each slice in parallel rewire them to be in series.

I expect dome of the desktop unit work that way

spot welding works by resistive heating of the metal, so there's no ion effects. DC is going to create a mgnetic field, I can't see that being a good thing.

this guy seems to be using regular low-ESR electrolytics:

4V so the turns ratio would be 55:1 assming 220V in. have you measured the primary current?

there's probably a magnetic shunt, but I don't think that will effect open-circuit voltage.

yeah they're impressive.

--
This email has not been checked by half-arsed antivirus software