AC-fed SMPS topologies or the lack thereof

The inrush is only charging the cap to the un-boosted voltage, which isn't much energy... in the USA anyhow. An NTC themistor can help there, if it's a problem. The boost topology needs a relatively small cap to store the required energy, and it tolerates a lot of ripple on the boosted side.

Higher voltage electrolytics are, per unit of stored energy, more volume and cost efficient than lower voltage caps.

It's amazing how much power and quality you can get per dollar from a MeanWell PFC supply.

The ancient GE SCR manual has a bunch of topologies, including the cycloconverter.

I don't recall any having tolerable input current waveforms, by modern standards.

The more I think about AC feeding, the more sense it makes. But I was clearly barking up the wrong tree. I'm thinking of an AC-fed *isolated* PFC stage, say 1:1 ratio to start with something, and then stepping down using the multiphase flying capacitor topology or another weirdo. Maybe it will not make much sense ultimately, but it works great for the neuron decalcification purposes. ;-)

Best regards, Piotr

Boost converter working in BCM as the PFC stage can have a very high efficiency without a large bulk filtering cap. Then for isolation/step down you put it into a "transformer" or more like a push-pull half bridge eg:

Then for regulation/filtering it goes into a buck with a fast control loop but you can significantly reduce the size of the output filter cap there too into non-electrolytic size by noting the "transformer" can have multiple outputs phased any way you like, so the catch diode can be connected up to one of those instead of ground so the filter cap never "sees" the full peak voltage between your output voltage and ground.

It's not that crazy it's been done before for LED lighting and stuff.

Well, it does. Wheel invented, case closed. :-/ Best regards, Piotr

Nm, I misunderstood the thrust of your post cap size reduction doesn't appear to be a requirement ;)

I'm not sure what you mean. Post a scribble and we can play with it.

I'm into resonant power stages lately, but that's just me. Why snub, and throw away, inductor energy when you can reuse it?

Because:

1. Resonant topologies usually require more VAR than delivered power.
2. Quasi-resonant and snubbed conventional topologies incur less (or much less, respectively) VAR, but the demands on inductor Q, Cp and device Cjo / Coss are severe, often making the approaches impractical, or altogether worse to attempt.
3. Controlling any but snubbed-conventional ones is increasingly difficult. Doesn't need to be a big deal, all that logic can be thrown into an IC, and there are many ICs for specific topologies already. But there's no general purpose method (of course), so you're stuck with that.

Tim

mandag den 3. september 2018 kl. 19.30.20 UTC+2 skrev John Larkin:

here's some big ones,

You are absolutely correct. Daunting.

Best regards, Piotr