h
te
That metal's pretty thick. That's a 1x2 square piece of 1 oz copper (34um), for example. Nichrome would be 60x thicker (2mm).
A few other guys and I linked data sheets, some with pulse curves.
If he went to 1milliohm then off-the shelf resistors work, guaranteed, no fuss.
James
There is a problem of hot spots. There is no way to figure out power unless the resistor is made of a solid piece of carbon or metal. If in doubt, use a pulse rated mosfet instead.
Vladimir Vassilevsky DSP and Mixed Signal Design Consultant
t
Is there such a thing as a 5 milliohm carbon resistor? All the SMD units I've seen are metal film.
-- Cheers, James Arthur
Sometimes, a carbon composition resistor was better than modern parts.
-- You can't have a sense of humor, if you have no sense.
I used to make 51 ohm dummy loads from 2W carbon comp resistors and a PL259 plug. I had to drill out the internal threads to get the resistor inside. I would solder the four holes for the shield shut, then insert the resistor and solder the center pin. I would fill the gap around the resistor with silicon grease, then solder a brass washer to the other end of the resistor and the end of the connector. They would handle 5W continuous, and 25 W for a minute at a time.
-- You can't have a sense of humor, if you have no sense.
ut
Interesting. The volumetric heat capacity of most 'stuff' is close* to 2 Joules/(cm^3 *K) I measured a 1206 at ~ 1.5mm X 3mm X 0.5mm, call it 2mm^3. So a heat capacity of maybe 4mJ/K, 65W for 1 ms is 65mJ. A predicted temperature rise of less than 20 K. (Assuming I punched all the number correctly). Maybe 1 millisecond is not enough time for the heat to get out of the thick film material and into the substrate? Did you every try lower power? How long would it last at say 6.5 Watts?
George H.
*close meaning a factor of two or so.
Yes, they are bulk conduction devices, where most of the heat is dumped into most of the mass. A carbon film resistor concentrates the heat into the low-mass thin spiral film, which transfers heat poorly to the ceramic substrate. Imaging Joergish pfff-thh sort of noises.
John
m
out
Right, the problem is that the heat starts off confined to a small part of the resistor volume. To figure the instantaneous temp rise we'd need to know the thermal tine constant from the center of the element to the environment.
John mentioned that the tau of a 1206 to big copper pours is about 2 seconds.
If we assume the metal path is the main heat route (which might not be true), then 1 millisecond is 1/2000th of a time constant(!), so substantially all the heat would be confined to the volume of the film. If we knew the value of the exploding resistor we could estimate volume, temp rise, etc.
-- Cheers, James Arthur
The gross thermal time constant of a 1206 soldered to the average PCB pads is a couple of seconds [1]. In milliseconds, most of the heat is trapped in the film and doesn't spread through the alumina substrate. That math would be interesting to do. The trim cuts cause hot spots, too.
If your calc was multiplied by, wild guess, 50:1 to take all that into account, we'd blow the element off.
This is yet another case - they pop up often here - where someone with time and interest (not naming any names, of course) could do some simple experiments, learn stuff, post it here, or put it on their web site and demonstrate competance.
As if!
John
[1] measured using a 1206 platimum thinfilm RTD, self-heating self-measuring mode.ftp://jjlarkin.lmi.net/RTD_on_board.JPG
ftp://jjlarkin.lmi.net/RTD_in_air.JPG
There are some solid manganin resistors, in Digikey I think. Hard to damage.
We make our own current shunts from sheet manganin, punched or chemically milled.
John
I've seen carbon film SMT resistors flash when a component failed and exceeded their power rating.
You'd think Joerge would have run out of them by now. :)
-- You can't have a sense of humor, if you have no sense.
I don't think you can look at a single time constant ratio and conclude anything useful about heat distribution. A bulk resistor might have the same gross tau.
I think you could analyse a thinfilm RTD to see the temperature change in the first microseconds and milliseconds after power is applied. The temp/time waveform should have a fast step component related to the film heating before the heat soaks into the alumina, then the slow bulk time constant. These should be distinguishable, to derive sort of a SOAR curve for a resistor.
Of course, you could just blow up a bunch of resistors and plot the points.
A good resistor might be film over copper, with a thin (diamond!) insulating layer between. RF resistors are sometimes film on AlN and they have astounding power ratings.
John
If you have a uniform cylinder of resistive material, with metal end caps, you'll have constant current density everywhere in it (at least until the skin effect becomes important, which in carbon is never).
The power dissipation density is rho*J**2, which is also constant. Thus no heat has to flow in order to distribute the temperature rise uniformly throughout the resistive element, and theoretically the only speed limit is the thermalization time of the electrons, which is very very fast. On my infrared tunnel junction devices, I've measured a bolometric response faster than 30 ps when you dissipate power in a metal.
So since the heat is dissipated uniformly in the volume, it's reasonable to take the instantaneous temperature rise caused by a current pulse to be
Delta T = I**2 R / M_thermal.
That's sort of fun, since the thermal mass approximation is supposed to apply only at low speed.
Cheers
Phil Hobbs
ohm
re out
nI figured the 2 seconds was the time for the heat to get from the 1206 substrate to the copper, and not the time constant from the thick film to the substrate.
The main heat path is most likely NOT the metal. Unless it was a really small value resistor... maybe milli-ohms. The thermal resistance of a one ohm resistor is about 10^-5 degree K/Watt.
I started on this project to 'do' the Wiedemann-Franz law. I've been meaning to post some question here.. but haven't gotten to it. To much other stuff to do, and now that project has gone onto the back burner.
George H.
There should be a factor of t in the right hand side, but you get the gist.
Cheers
Phil Hobbs
-- Dr Philip C D Hobbs Principal Consultant ElectroOptical Innovations LLC Optics, Electro-optics, Photonics, Analog Electronics 160 North State Road #203 Briarcliff Manor NY 10510 845-480-2058 hobbs at electrooptical dot net http://electrooptical.net
hm
e out
Thanks, A few years ago I tried to measure how much of the heat from a through hole 1 k ohm resistor comes through the leads. (versus into the air.) The answer as best as I could determine was none!
George H.
m-ohm
gure out
in
*ed
Yes.
I was figuring a metal film, and that the metal path would be faster (lower thermal resistance). That might not be true. Still, over 1mS, whatever the substrate, it's clear the temp rise originates uniformly in the film volume, and will be strictly the dissipation / (mass * heat capacity).
I'm assuming a metal-foil resistor (because Tim spec'd 5 milliohms).
To the environment? That can't be right? I'd expect something like
50-300K/Watt, depending on the size, mounting, and air flow.-- Cheers, James Arthur
e
Some resistors do have pulse power ratings/info e.g.
5m-ohm
figure out
or in
se*
,cted
e , eOK then you might be correct. I was thinking of JL's 1206 resistors.
Opps.. sorry I realized at lunch that I got that upside down. 10^5 K/ watt. Only off by ten orders of magnitude. :^) That's the thermal resistance from one side of the resistor to the other.... assuming that all you have is some resistive material and no substrate, or air, just electrons.
George H.
Heck, just solder them across a line cord and plug it in.
Some values make a little sharp PIP noise and a spark and open up. Some burst into flames.
John
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.