Picked up a square foot of copper plate the other day, and...

[T3sl4co1l]

...Finished a transformer, of course.  Should be good for a few amperes, eh?

Construction is three series sections of ~15 turns each (10AWG litz, primary), three parallel sections of 2 turns (sheet, secondary).  Output terminal is parallel plates.  Insulator is fish paper.

I just barely managed to solder the secondary winding straps onto the plates, with my hot air machine turned up all the way plus using my soldering iron for local heating.  Took a good ten minutes of heating, plus fiberglass blanket draped over the assembly to hold the heat in.

The top plate has a wishbone or spanner shape to it, picking up the three outer secondary connections but not extending under the core.  The bottom plate picks up the inner secondary connections, and has a smaller hexagonal hole in it to mate with the straps (again, soft soldered).



Earlier view of the primary, and the driving circuitry -- a control board featuring phase, current and power regulation for the resonant load, as well as fault latching and rudimentary remote control (on/off, fault and power set/read); and a half-bridge inverter board good for a kilowatt or so.

It's been a long time since I did much work on this induction heater project.  Last time I did this much work, was probably back in 2009 when I built this thing:



which I ran up to 5kW at the time, though with some quirks that needed a bit more tinkering to finish up, but it was definitely proof of concept.  But then I had to move out to college.  And after that, I started working, and after THAT, I didn't have a workshop or a 240V outlet, and...

I've still got that unit; I'm finally getting around to... rearranging, or repurposing, but not necessarily scrapping it?  Here's the output network removed, fitted with a shiny new coil:



Not nearly as pretty as it was back then!  Well, the old transformer doesn't do me much good.  I did wire it up to the new inverter, and see that everything still works; got some steel rod nice and red-hot, but it's neither the right turns ratio nor low enough stray inductance (looks to be close to 0.8uH in the pipes between capacitor, transformer and coil terminals; what was I thinking?! ) to do what I want to do now.

Which is why I made the new transformer, should be very low stray inductance indeed.  I'm aiming for a tuning range of 0.2-2uH and Q = 5-50, a very practical range for smaller coils useful at these lower power levels, for applications like brazing and heat treating of small parts.

Tim

[T3sl4co1l]

And here it is in use.



Tim

[MagicSmoker]

...
Earlier view of the primary, and the driving circuitry -- a control board featuring phase, current and power regulation for the resonant load, as well as fault latching and rudimentary remote control (on/off, fault and power set/read)...

So you went the PLL route? I attempted that once before (for an ultrasonic welder) but never could get the loop time constants just right to deal with abrupt changes in loading. But I was using a crappy old 4046 and this was back in 2012, so the details are kind of hazy.

[T3sl4co1l]

Yep. The compensation still needs a bit of tweaking; sometimes it starts up and skips below resonance.  There's probably some crud with the phase reference too (I think it's timed to the wrong edge of the gate pulse, so it's delayed by a dead time).  A confidence detector would be cool (phase output doesn't assert until the inputs are of adequate amplitude).

Seems hard to tweak the compensation anyway; the error signal is extremely small, which figures as we're talking 100s Hz tweaks out of a ~100kHz total range.  I need to do some step transient testing.

It doesn't help that the frequency of the tank's dominant pole goes as the difference between driven and resonant frequencies; so the compensation needs to be very slow to account for the huge LF peak when near resonance, even if it's stable with a much faster time constant when further away from resonance, or at resonance.  Which also impacts startup time (as it ramps down from the initial frequency, slew rate is set by compensation).

That's one advantage to a variable-supply inverter, or phase shift PWM (full bridge).  Run it at resonance, so you have a single pole (the L/R = R*C time constant of the tank), and control it just like any buck converter driving an R||C load.

Tim