resistor survival wild guess

Right. The failure mechanism will be exploding the cermet element.

My expected time scale is under a millisecond.

The joule capacity of a resistor seems to drop at shorter time scales.

Looks like roughly 20 mJ for a 1206 at 1 millisecond.

Nice.. 10-20 watts for 10 msec. At some point you might have to worry about how fast the heat gets out of the resistor material. But my WAG would be it goes (the heat pulse) as something like the speed of sound... (~4000m/s) so 1 mm in microseconds... there's probably surface effects that screw that up.

George H.

They may have different failure criteria too.

I'll blow some up. 30 volts into 50 ohms is an easy test. Just apply the 30 volts, scope the current, and let it fail.

Don't keep us guessing! Make those 1206s pay for their crimes

The heat generation is reasonably uniform throughout the resistive element, so no heat flow is required to get that amount of volume. (I have some old nickel bolometer data showing that effect working down to about 30-ps pulses.)

Cheers

Phil Hobbs

Also check out the Multicomp MCPWR06 series. The datasheet curve shows 10ms for a 20 watt single pulse.

If 8 msec. is within a factor of 2, I think I win a beer. (so I only owe JL 99 now. :) And if it is right closer investigation will revile I made compensating factors of two-three errors in my guesstimate.

George H.

At some point, skin effect will take over, and reduce volume further (also a diffusion characteristic, incidentally!). Did you not see that, or was the thingy dimensioned such that 30ps didn't experience it?

Tim

It was optical--one of my Ni-NiO-Ni tunnel junctions shorted out and turned into a bolometer. Skin depth isn't a good description of that, because of the small dimensions and because the antennas were made of gold, which doesn't look like a normal metal in the IR.

Normal metals have an effective epsilon of j sigma/Omega, whereas in the IR, the coinage metals exhibit free-electron behaviour, i.e. epsilon is a large _negative_ real number, or very close to it. Makes for some entertaining effects such as surface plasmons.

Cheers

Phil Hobbs

Ah, kinda suspected it was some of your arcanum. :) Skin depth becomes penetration depth at those wavelengths, and I'm guessing it was thinner than that, among other wave-mechanical things helping out?

Tim

es

an

To derive the skin depth formula, you start with a so-called normal conduct or, i.e. epsilon = j sigma/omega, which in the limit of large (sigma/omeg a) leads to a k vector inside the material of approximately (1+j) nhat, wh ere nhat is the unit normal vector. That is, the field is exponentially dec aying inaide the metal surface.

If the material's behaviour isn't dominated by conduction, that derivation doesn't apply. At 200 THz gold is a lowish-loss dielectric for my purposes, albeit one with weird properties. The fields occupy the full cross-section of the metal.

Cheers

Phil Hobbs

What physical conditions of the resistor is the accepted demarcation between a resistor which is in a "surviving" state and a gone, deceased, pining-for-the-fjords resistor?

A 50 ohm 1206 thickfilm resistor is a good terminator up into the GHz, so the skin depth must be a lot less than the element thickness. Resistivity is high for this stuff.

Certainly you've been witness to the release of the magic smoke. :^)

For the above I had to pick a temperature. 100K (rise) seemed to low, so I picked the 200K. That everything (solid) has about 3J/cc*K of heat capacity is a useful factoid. See 2nd last column here,

Aside: Why does pining for the fjords have anything to do with death?

George H.

It is possible to go about it systematically.

Monty Python's "Dead Parrot" sketch. One of a multitude of British euphemisms.

Jeroen Belleman