a few electronic fuse circuits

I don't think I follow - shouldn't the comparator inputs be reversed? Short circuit on the output causes ground current to increase against the MOSFET's on resistance, turning the MOSFET off?

And the average value of some clock is used as an expedient reference voltage?

It's hard to believe that an LM317 in quantity is cheaper than a single comparator and a MOSFET in quantity but I bet it's true! I think it's possible to build the world's cheapest 12 watt stereo power amp using LM317s for all the functional parts.

I don't think I follow - shouldn't the comparator inputs be reversed? Short circuit on the output causes ground current to increase against the MOSFET's on resistance, turning the MOSFET off?

And the average value of some clock is used as an expedient reference voltage?

It's hard to believe that an LM317 in quantity is cheaper than a single comparator and a MOSFET in quantity but I bet it's true! I think it's possible to build the world's cheapest 12 watt stereo power amp using LM317s for all the functional parts.

Oops, right, the comparator inputs are switched (note clever use of passive voice.) I'll fix that. I'll blame Rob for not catching that... that's what minions are for.

The clock periodically forces the fet on, which resets the circuit if it's tripped. Something like 50 milliseconds high, and a second or two low, might be right here. The force-on interval lets load caps charge up and such. One clock generator, a uP port pin maybe, would drive all

We're paying 13 cents for an LM317L in TO-92, which is what we'll use here. It current+thermal limits really nicely. The dc/dc sips are only $4, so it's still better than rolling our own, custom magnetics and all.

John

Have you ever checked the input-output capacitance on those things?

Best regards, Spehro Pefhany

No, but there are some pretty small ceramics inside. My external caps are likely a lot bigger.

ftp://jjlarkin.lmi.net/VASD1_sch.JPG

ftp://jjlarkin.lmi.net/VASD1_top.JPG

John

...maybe it depends how long the short is? Sort of like a standard fuse may take hours at a given current before it finally sags then breaks? I have seen curves of beak time VS current apparently published by the makers (try Littlefuse).

If the ground is continuous to the DC-DC loads, you may not be doing what you think you are.

RL

What converter is it that specs short circuit tolerance as 1Sec?

Why would they market it? Why would you buy it?

RL

You snipped the part number.

Most of the non-regulating sip dc/dc converters have a short-circuit duration spec, typically 1 second. This CUI part can't manage that.

Because all the other specs are good, and it's cheap and easy. We use

John

Good point. But ... I'd run that "switcher in a can" through the paces. Heat, cold, tons of power cycles, the works. Then I'd crack one open to see which parts they used in there. Probably no 'lytics, but if they did then all bets are off when it comes to long term reliability.

If they had pinned out the center tap on the left then life would probably be much easier. Then you could pull that down to disable the converter.

I tested it pretty hard. It's rated at 1 watt and baraly gets warm at

1. Noise is pretty low. I unpotted one, pic links are elsewhere in this thread. They're really nice for , and the sip package uses board footprint well and gets good air flow cooling.

Too bad about the short-circuit thing. I'd pay another 50 cents if they'd fix that. It might cost them three.

John

Short circuit proofing is pretty tough in that small space. At least they could have pinned out the center tap on the oscillator side, let the marketeers call it a disable feature or whatever. Probably cost them under 2c. Then all you need is 2-3 resistors and a NPN transistor. They'd have to go to 0402 stuff if they wanted to cram that in there.

Pity they won't sell non-potted ones, else you could add a pin at that vacant position on the far right in your photo and run a jumper over. Pretty cheap rework and you could have that done overseas if needed.

I suggest you breadboard a few.

One of the reasons that short circuit is difficult in low-cost isolated converters is the adoption of flyback-like topologies with no output inductor. This prevents primary peak current sensd controls from reducing current into a short, even when hiccoughing.

This kind of overload doesn't show up as the linear input current stress that a 317 would react to.

RL

I figured out the likely schematic:

ftp://jjlarkin.lmi.net/VASD1_sch.JPG

from this

ftp://jjlarkin.lmi.net/VASD1_top.JPG

(I ripped out the toroidal transformer when I unpotted it.)

The probelm is that the base resistor is low enough that the transistors fry if it's not oscillating. There's a simple fix, but it needs two more parts.

The 317L does protect it nicely. It limits at about 300 mA, and thermal cycles on/off for a sustained load. It's perfect, except for the voltage drop.

John

that looks like a killer circuit that would need a supervisor circuit to protect it.. Yes, that will kill the transistors..

A resetable TOD (Time off delay) to supply the bias and reset from one of the collectors pulse to keep continuous supply to the bias of the trannies..

I did this years ago using a 555 chip.. very simply thing to do.. Also can be don't with a comparator, where the comparator senses the pulse through a RC network and outputs the bias voltage to the transistors.

Jamie.

I think

I tried that for a day or so and was spectacularly unsuccessful. Got a pointer to somebody who made it work?

Thanks,

Jim

The output stage could be a 317 follower (input to adj pin) with a 317 constant-current sink. Previous stages can get voltage gain by applying signal to adj and taking the output from the "input" pin.

Transformer coupling could be interesting, too.

Sort of silly, though.

John