Transistor booster design

I like the technique Genome showed us using a self-coupled step-up inductor for the boost which relaxes the requirements for inordinately high speed switching for large duty cycles: View in a fixed-width font such as Courier.

. . . . V+ . . | . ) . ) . ) . ) . |o . .-----------+ . | | . o ) . \\ ) . o ) . | ) . | |o . --- | . '----|>|---+--->Vout . | . --- . --- . | . | . --- . . . . .

John

Now that's the way to breadboard a wideband circuit...

booster

That's what I have in mind. Isolation is the primary objective. Self oscillation is a bonus. If it does not oscillation, I can always PWM it.

I could not get any reasonable parameters out of your second circuit

But I place a 1uF cap across the emitter coil in your first circuit. The output looks a lot nicer.

Your can't change this cap value yet, but you can see it in the spice model.

Thank you for your contributions. I will send you gifts and samples when it's ready.

That's the "Genome" circuit, a fet driving a center-tapped inductor, but the hot end drives a C-W multiplier for about 1 KV net DC output from 5 volts in. Tapping the inductor allows a higher flyback ratio mostly because the fet drain capacitance has to charge less.

The unused DIP is a VMI C-W package, an 8x multiplier, probably a hybrid. I didn't use it because the caps are very small (100 pF or something like that) and because they want $48 for it!

I like to breadboard this way because it has a good ground, and because you can comment the circuit with a Sharpie and keep it forever, and because I can really "see" the circuit. I hate those plastic-block breadboard things.

I've also - a very long time ago - done a "reverse Genome" buck converter, with a center-tapped inductor...

in-----------switch---------///////////--------+---out | | | | catch k C diode a | | | | gnd | gnd

I did this because I couldn't get a power Schottky that would stand the full input voltage. The bad news is that the ripple current into the cap is high.

This was used in a state-of-the-art digital engine control console for the LHA ships. A few years later, the Navy decided they couldn't maintain it, ripped it all out, and replace it all with pneumatics.

John

Fred Bloggs a écrit :

And that variation from Tony Williams is really great. Taken from the thread "Popped my FET -- flyback discussion"

----------- The only prob with that circuit is that even with an autotransformer there will still be a transient loss of transformer action during the initial collapsing of the flux. This still makes the Vd-s spike slightly indeterminate.

========================== 12V +-----//////--+ +--///////--|>|--+-->Vout | s 20T f | | s 132T f | | | D | | | |--+--|>|--+ Vc | 10uF=== ---|| | === 0.22uF min | |--+ ===C | | | | | 0V -+-------------+-------+----------------+---

The circuit above is a variant, with D+C added. They look redundant, but what they do is provide a guaranteed close-coupled snubbing of Vd-s. Any transient primary flyback energy is still being recovered and added to E-out. C= 1uF, or so?

Vd-s(snubbing) = Vc = (Vout + 79)/7.6

For a 200V Vout, Vc drops to 37V.

Which means that the transformer turns-ratio could be optimised to get closer to the FET's Vd-s(max), (and consequent reduction in Ipk).

Speaking as a connoisseur of bizarre circuits, that one is especially nice.

In a single-inductor boost converter, the limit on boost ratio and efficiency tends to be the fet drain capacitance; eventually all available inductor energy gets dumped into Cd and you can't multiply any more.

John

you got that transistor upside down :-) avalanche?

Bye. Jasen

hmm, it looked that bad in suimulation for me too, but I thought maybe having the right model for Q2 would help.

Is it supposed to regulate at all, because I'm not seeing a significant amount of supply rejection...

as I see it it suffers from the slow turn-off that's typical of the eccles-jordan oscilator, that burns lots of energy in Q2

a half-bridge driving a doubler would beat this thing.

Bye. Jasen

See:

Ed