AC Cap Dropper DC Power Supply

Body diode?

Jamie

I ended up with this...

Using a depletion-mode NMOS FET.

I have a version I personally liked better that used a cheapy TRIAC under DDROP3, but I could not find a way to get some "free" power for the gate... needed while pumping, goes away when overcome by the switcher coming on line.

(This is a wild high PFC, high efficiency dude, where the customer is squeezing me for milliwatts :-) ...Jim Thompson

As Spehro suggested, the fet can be turned on when the current swings negative, which shorts the body diode and improves efficiency a tiny bit more.

Yep, that's very similar to my 'heater pulse regulator' but if you want to shave the last mW out of it then you may want to add some form of hysteresis because, what happens is, after the initial charge up, bringing Vcc to 'reg', it then goes into 'series pass regulator' mode for the rest of the positive cycle. That may not be a problem for you because you're using the cap dropper, and I wasn't, but mine was regulating on the order of 9 watts and the 'series pass' bit was adding another 3.

I was using junk bin parts so rather than get too terribly fancy I just summed in a bit of the AC waveform so the MOSFET would be forced off after the charge up. That had the effect of a charge pulse twice per half cycle since it was 'in range' on the down slope as well as the up.

That's crazy. Not only is it expensive, once the 24 volt supply kicks in, it dissipates most of a watt, doing nothing.

Boy, did I just save you some embarassment.

You're so ignorant you don't know how it works. Thanks for the exhibition! ...Jim Thompson

Well, how much power do R1, R2, the TL431, and ROFF dissipate once the main supply is up?

Harder question: what's the worst-case dissipation of the fet if the main supply doesn't come up, like if the load is shorted or something?

Bwahahahahaha! You lost the thread... actually two threads... and a controller chip...

Follow your own advice... do your own searching ;-) ...Jim Thompson

...Jim Thompson

How much power do R1, R2, the TL431, and ROFF dissipate once the main supply is up?

It was "my bad" to have emulated the main supply with a pulse source... doesn't work that way at all, but I can't show the customer's circuitry.

In actual practice a PFC/PWM chip comes alive and the switcher starts. But that "20V" sags since the charge pump can only support ~24mA.

The switcher takes over at 14.5V (about 250ms into the action), the FET gate is pulled low to shut off the wasted power. The gate resistor is actually 220K, but the demo TL431 won't work (because of Iq) to show how "regulation" might be accomplished.

You are trying to amuse yourself by throwing sand in the air. But you don't understand how it works. Will some first year EE student explain it on Larkin ?:-) ...Jim Thompson

I was thinking about that, specifically where that idealized 300 volt-peak full-wave-rectified waveform came from. If it is through diodes, there's no pull-down to pump the series RC thing; if it also drives a 24-volt non-PFC switcher front-end, you won't get that waveform. If the startup circuit has to be working before a main PFC switcher starts up, there are all sorts of interesting tangles. If the switcher starts up on its own, what's this circuit for? Standby power for some controls? It would still have to work when the PFC was down.

Right. ROFF is obviously over-constrained, hence my question about power consumption, which you elected to not answer. There's no ROFF value that works sensibly. A much simpler circuit would work.

Does this rig have to work from 90 to 264 volts AC? That, and the Idss spread, gets interesting.

I understand how it doesn't work. And so far, it doesn't.

You've made a pretty quick transition from "I ended up with this..." to "my bad." But you're not done yet.

ps - the answer is just about 1 watt lost, with your values of R1, R2, the TL431, and ROFF. That is, for the record, about 1000 of your customer's milliwatts.

Keep trying.

My fault for trying to demo without telling ;-)

There's roughly 400mW consumed while bringing up the switcher. Drops to ~1mW when turned off.

If the load is shorted ?:-) 212mW in RDROP, 31.9mW in the FET. See, you clearly don't understand how these things work :-) ...Jim Thompson

You always brag about designing great circuits without "telling." You troll for ideas, then hide behind "proprietary" when you do settle on something... even if you got the ideas here.

Then you have some other circuitry that specifically turns off the startup supply, that is not shown in the circuit that you posted here. What is it?

Your circuit is complex. Something much simpler would work, and would go to very low power all by itself once the main switcher comes up.

Is that worst-case? The fet Idss is min 20 mA, typ 80, no specified max. There is some value of Idss that results in the maximum power dissipation. What is that value, and what is the dissipation? You should be able to tease Idss to dump around a watt in the fet, which is fine if you heat sink it.

What's "complex" about it? I need to support 20V at 15mA-20mA for

"should be" is not an engineering term. My wife uses that term "should be" a lot while referring to the present administration. Annoys the hell out of me. We "should be" a Representative Republic, but we're a dictatorship, the other branches of government are sitting on their hands and letting Obama do everything by executive order :-( ...Jim Thompson

It has a lot of parts. And you won't address the 1-watt dissipation issue.

You are obligated to find the maximum power dissipation point as a function of possible Idss values, and to heatsink such that nothing catches fire.

If the design is for universal, CE-tested use, AC line voltage can go to 264 RMS. That's 373 peak. In that case, the maximum fet dissipation can be teased well over 1 watt; try it. That fet is rated 1.8 watts max. If you sim one case, with nominal line voltage and typical Idss, you are taking risks. Test lab time is expensive; product recalls and fires are more expensive.

Your sim seems to assume 240 line voltage.

What 1-Watt dissipation issue? Just because you made up some bull-shit issue doesn't make it real.

Poor baby! Go sit in the corner and suck your thumb. ...Jim Thompson

It's down-right amusing, YOU lecturing me on rigorous simulation. I often do 45 PVT corners.

YOU, You post nothing but vague crap.

Once I decide on a configuration I'll beat the crap out of it to make sure it's a valid design.

Go sit in the corner and suck your thumb and contemplate how many other ways you can think of to pimp yourself >:-) ...Jim Thompson

In the PDF that you posted, when the startup circuit is back-fed to 23 volts or so by the main switcher, the parts that I've named dissipate about one watt. ROFF alone burns over 800 mW. If you have a workaround for that, show it.

Why can't you see this?

OK, you can't calculate the worst-case fet dissipation.

Hey, spin the design as many times as your customer can stand. And have lots of liability insurance, 'cause you're going to be on the painful end of the expert-witness game. And you're being this sloppy, in public, archived forever.