discharging caps

As a comparison, look at the size of your 5J and 10J MOV parts.

Is the MOSFET die about that size?

What's the rule of thumb for abuse on a resistor? Isn't it more than

Hmmm..how about an incandescent 60w bulb?

I think MOVs have some non-thermal long-term degradation mechanism. I can certainly get a mosfet rated for 100, or even 300 watts, but lots of mosfets die if you try to run up to their rtaed dissipation at higher voltages. You can see that in the SOAR curves. And the heat does have to go somewhere, and heat sinks are expensive.

I tried some 6 watt wirewounds that we have in stock. They get warm at

That would be neat electrically, and fun to watch, but messy mechanically. Where do I get a pcb-mount 60 watt lightbulb?

Why not use a little PCB-mount power relay (powered from the unfiltered supply)?

That way you don't need the LED or the pushbutton- it's always discharged to a safe condition shortly after the power is turned off.

Or use a MOSFET and a bright blinkie-LED if you want solid state, automation and confirmation, but the control gets a bit more complex.

Discharging from + to - might warrant some analysis to make sure there are not any conditions under which Bad Things might happen of the polarities of one of the supplies relative to ground flips.

The supplies are commercial switchers, so I don't have an unfiltered version. And if anything went wrong, something would catch fire. My test people are smart enough to turn off power before they push the DISCHARGE button. And I'd need a relay rated for 1 amp, 100 VDC, 48 volt coil.

Right. I think the diodes in the switchers will take care of that, but I'll check.

John

I assume you have antireversal diodes on the rails, right?

Cheers

Phil Hobbs

I figure the switching power supplies will take care of that. But I suppose I should verify that.

Thems is incendiary words...pun intended. My

Think that's a false assumption. Who among us has never done a stupid thing?

I'm paranoid when it comes to product liability. If you adhere to standard safety practices, there are lots of places to place blame. Once you add a button that says, "press this to be safe", YOU become the primary target. The injured party testifies that he pressed the discharge button before the explosion. Maybe your discharge circuit failed...maybe it didn't. Doesn't matter. You're screwed!

I think the legal ramifications of this decision far outweigh the technical issues.

And I'd need a relay rated for 1 amp, 100 VDC, 48

I'd keep the resistor. I have no equations to back this up. I've got a little few watt heater. (120 C operating temp) I put both the transistor and resistor on the thing I'm heating. Each disipates about 1/2 the power at normal operating conditions. That seems safest.

George H.

I agree, that's a poor assumption to make. Even if it's 99.99% correct, if you do this operation 10 times a day, you'll blow up something about once per three years if you have only a single machine.

I'm a software guy... and I don't trust software for safety interlocks. If you've got a processor driving the gate of something that can only sustain a 10% duty cycle without burning up, then it's a

I don't trust "wetware" any more than I trust software, for this sort of safety. Human brains simply aren't reliable enough... and that's just as true of smart people as unsmart people.

That's why electrical-worker safety regs require lockout and tagging protocols (with one lock per worker!) on circuit breakers. It avoids those embarrassing "Oh, I just did a thinko... and now I have to go apologize to the grieving widow" moments.

In this situation, I'd stick a relay in the switched mains circuit, and wire it up to disable the DISCHARGE button... make the discharge circuit a no-op if the mains are still live.

Put in more screws. If it takes a hundred seconds to discharge, 50 screws should be about right.

Tim

-- Deep Friar: a very philosophical monk. Website:

That's a gross SOA violation--a FET-buster.

@ 50Vds it's good for 2A x 10mS =3D 1J.

With the resistors it looks great.

-- Cheers, James Arthur

Oh, I wouldn't use that fet. It would have to be some monster.

Yeah, I should play it safe and stay low-tech on this one.

"John Larkin"

It requires a button to be pressed and held down by an operator - Murphy's Law will apply.

If one electro is not charged, faulty or has a broken connection, all hell may break loose.

At the very least, hefty reverse diodes across each electro are essential.

... Phil

On a sunny day (Tue, 17 Apr 2012 14:30:44 -0700) it happened John Larkin wrote in :

tium experiment

That is exactly what I do in the tritium experiment with the power MOSFET that has as only purpose heating the hot plate. Small source resistor, CMOS opamp and feedback, Tim added a capacitor. For a fixed input voltage the current is constant. If you want power limiting during dischange add a 2$ PIC and also monitor voltage across the MOSFET, and drive the MOSFET from filtered PWM [1]. Maybe you need a 3V button cell to keep that computah running to the last volt (Vgs you need). Oh what fun... :-)

can you use burn the energy in the ESR of the caps, perhps using an SCR.

Where have you been for the last 100 years? There have been surface mount incandescent lamps since the first workable light bulb.

There will be a bright orange LED on to remind the test tech that the board is hot. They turn off power and push the button until the LED goes off. I've done this before on other, bigger amps, and it worked fine, no problems, but I used a chassis-mount power rocker switch and a big aluminum-case bolt-down resistor.

Passive bleeders burn a lot of power and are slow and expensive.

Across each cap? They are soldered to a PC board. If the connection to the board is bad - unlikely - the connection to a diode will be bad, too.

I checked my switching power supplies. Each, when off, seems to have about a schottky drop in the reverse direction, so there's no worry about reversing a supply through the discharge resistors.

.highlandtechnology.com=A0 jlarkin at highlandtechnology dot com

it on the limit but: it will barely do 500mA @ 100V, and the voltage will quickly drop abs. max Pd 60W, so that will also be just under the limit, and again the voltage will drop start to drop. looks like anything over ~100ms is same as DC so the 2.5C/W tjc should count so with a block of alu big enough to eat the 150J without getting more than a few deg hotter even that little fet might just survive

-Lasse