joule capacity of surface-mount resistors

surface-mount

Yeah, I guess the summary is that the heat has to be generated all through the bulk of the resistive material. Most insulators conduct heat slowly enough that the element gets all the energy initially, so it vaporizes. Even most wirewound resistors make the heat in a small fraction of the total volume.

I might be able to use their 234SP, standing up. It's rated 30 joules for 10 milisecond pulses, apparently more for longer pulses.

If I discharge 30,000 uF through, say, 5 ohms, that's a 150 millisecond tau. A carbon comp with that diameter+length might work too, if one existed.

Here's the board... this is a dummy layout, just pushing parts around and seeing what might fit.

That resistor might fit into the upper-left corner.

I think the Globar parts are carborundum, silicon carbide, which is hard to vaporize.

Thanks

I've had to do that. Some resistors are rated for pulse service; a few even have enough information for derating depending on the pulse length.

Were I doing this I'd start by digging around for data sheets that actually specify pulse power dissipation. Unless I had _lots_ of time to waste doing my own qualification I'd award the design win to the company that actually specifies their products thoroughly.

And yes, 200ms is quite long.

surface-mount

that

wirewound

seeing

The worry with those is probably thermal fatigue. I like the lamp idea. A Maglite bulb might be about right.

Cheers

Phil Hobbs

these say 100J :

-Lasse

I'm

surface-mount

long.

the

that

wirewound

seeing

We have had big wirewounds fail from pulsed-load cycling in NMR gradient drivers, without exceeding their peak rated power.

The caps can be charged to 50 volts, and I'd like the equivalent of a few ohms.

Just tried a 1/4 watt carbon comp, 4.7 ohms, discharging 10,000 uF. It fell apart at 40 volts, 8 joules. That's impressive.

A 22 ohm 2-watt comp was fine at 60 volts, 18 j. Need to try some bigger caps.

Looks like I need volume to absorb the energy, and on this board, that means I have to use all the height available. So I have to stand something up on end. Ugly but necessaty.

Their axial 4-watt wirewound KNP400 is rated for 40 watts for 5 seconds, which is 200 joules. The chip parts don't look promising.

SMT resistors generally have poor pulse handling, AFAIK.

You could try an SMT inductor however: find one with enough copper volume and appropriate DC resistance, connect a diode in anti-parallel to quench the ringing and there you have what amounts essentially to a big wirewound resistor :)

Unlikely to be precise, but if your goal is just to discharge some energy storage caps on switch-off in order to keep them from storing charge indefinitely and you don't care too much about the exact discharge waveform, why not.

Greetings Dimitrij

ed text -

OK it seems like you just need a certain volume of 'resistor' to soak up all the energy. So for a crazy idea, how about you dedicate a layer of the pcb to make a resistor out of copper. ~0.4 J/g-K a 100 degree rise sounds a bit extreme (for one gram of copper).

George H.

The other approach would be to mount a through-hole resistor on skinny spacers, e.g. thin wall copper or teflon tubing, and put components underneath.

Cheers

Phil Hobbs

Or dump the energy into a high voltage capacitor, discharging this secondary capacitor slowly

Cheers

Klaus

text -

OK, say 3 ohms at 500 uohm/square. I'd need 6000 squares, maybe 60 inches of 10 mil trace. Not unreasonable. The paved area would be fairly large, so it would be heating a lot of epoxy as well as copper.

That is interesting. One could estimate the mass of the copper (or unwind it and weigh it) to compute peak temperature. The metal-to-insulator mass ratio of an inductor may well be better than a typical wirewound resistor.

With 30,000 uF of electrolytics, I'd expect the Q to be so low that the diode wouldn't be necessary.

Cool. Gotta think about that one.

What are you using to switch the load?

The load is a bar laser, up to maybe 25 volts and 250 amps pulsed power, controlled by a mess of mosfets. The resistor is between the capacitor bank and a customer's DC power supply. It serves several functions, like limiting poweron current surge, accidental shorts to ground, and keeps his (switching?) supply from going nuts after every laser pulse.

El 23-05-13 19:43, John Larkin escribió:

Hello John,

I know you need lots of small chip parts to handle your pulse, so if you can live with the inductance of a wire wound resistor it may be a better choice.

How many do you need to make? Can you stack surface mount resistors on top of each other?

Mikek

Did these WW's have welded connections, or swaged/caps? I've had very reliable service from the welded types.

One example from eons ago when I was a student/tech in a University nuclear structure lab- I was asked to repair an instrument that had been left on for weeks in a known failed state (they didn't want to interrupt an experiment). The failure had caused a wirewound to be driven so hard it glowed moderately bright red in normal room light. Once turned off and disconnected from the circuit - the resistor was still within tolerance.