Induction Charger Beat Into Submission

Ewwwww.

Almost as ewwwww. Hard switching into a capacitor. What's that, 60% efficient? Maybe 80 tops?

As has been mentioned, several times, what you need is a series resonant driver. Start with two transistors, half-bridge style. You can use a P channel for the top one at that voltage, piece of cake to drive. Series cap and coil goes between output and ground (preferably "ground" is halfway between +V and GND, by splitting the series cap into equal halves). You'll want to control power by reducing Q, supply voltage, pulse width (TL494 style pulse width modulation) or increasing frequency, otherwise when nothing is in the charger, current will ramp way up and cook things. No need wasting 10W when it's at idle just to deliver not even 1W when in use.

For starting points, think about CFL driver circuits. Same basic thing, but current feedback makes it self-oscillate with no need for an extra controller. Current draw is determined by Q, so you either want to limit supply current, or Q. With two transformers and a pair of 2N4401 you should have this thing pretty well licked, at 90% efficiency, and with so few components Joerg might blush.

Tim

Ewwwww.

Almost as ewwwww. Hard switching into a capacitor. What's that, 60% efficient? Maybe 80 tops?

As has been mentioned, several times, what you need is a series resonant driver. Start with two transistors, half-bridge style. You can use a P channel for the top one at that voltage, piece of cake to drive. Series cap and coil goes between output and ground (preferably "ground" is halfway between +V and GND, by splitting the series cap into equal halves). You'll want to control power by reducing Q, supply voltage, pulse width (TL494 style pulse width modulation) or increasing frequency, otherwise when nothing is in the charger, current will ramp way up and cook things. No need wasting 10W when it's at idle just to deliver not even 1W when in use.

For starting points, think about CFL driver circuits. Same basic thing, but current feedback makes it self-oscillate with no need for an extra controller. Current draw is determined by Q, so you either want to limit supply current, or Q. With two transformers and a pair of 2N4401 you should have this thing pretty well licked, at 90% efficiency, and with so few components Joerg might blush.

Tim

Push pull driver for transformers see here

The other thing about the present circuit is that there's too much inductance in the coils in relation to te resonating caps. I would expect around 0.01uF range of values, not 0.002uF. I would try 50 turns and 100 sec and adjust the resonating cap values.

A series resonant drive is still the best way to go and again, keep the turns at a minimum and wire gauge as thick as possible. Skin effect will be significant, even at 50 Khz and is why switchers usually use copper foil in the high current windings, or even litz wire on some early designs :-)...

Regards,

Chris

I've already talked to the manufacturer about my center-tapped primary in the previous iteration and he bitched about that. I can imagine what he'd say if I told him I need him to wind one with multiple isolated windings. I failed to mention this in my previous posts (didn't feel the need) but low cost is a very serious design goal. And a firmware solution for tuning is pretty much free, since I write the code and there's so much to write anyway, what's 100 more lines for a tuning algorithm?

If you have any suggestions for making this thing even simpler, or cheaper, I'd gladly accept your input.

Could it be that the diodes were just too slow? If they were truly across the primary, yeah, that would be like a short for one half-wave.

With no core you may be ok. Typically you'd want some loop to find series resonance, a calibration procedure in production is usually not so cool.

I'd still provide a diode to 30V or 40V, right now the FET has no protection against exceeding it's abs max voltage spec. Unless it's a really high-voltage version.

Just a note that this kind of stuff has a lot in common with induction heaters that use tuned coils.

Oh absolutely, tons in common:

Primary is 4T of 1/4" Cu, 2" tall, 2" dia. Secondary is abuot 500T #32, with #22 leads coming out. Source is around 70V / 220A RMS. So you're seeing a good couple of amps and volts going through that arc, well fractional amps anyway. And I'm sure most of the voltage is dropping across leakage/stray inductance.

Only difference is, you should have a tuned secondary as well, making it more of a CW Tesla coil.

Tim