Smoking power resistor wattage overhead

[John123]

How much overhead do you guys leave when choosing power resistors? For example i've got a 22 ohm 3w resistor dissipating about 2.5w continuous, its green in colour and about 15mm long by 4-5mm wide. As expected it runs red hot, but something weird happens after a little while in smoke starts coming from it.

I checked with the scope and it definitely doesn't go above 3w RMS, so I assume it's just my finger oils or there's some kind of derating thing going on. The 2.5w dissipation needs to be continuous without smoke. 

I don't have a datasheet for this random resistor I found in my parts bin, but do they often have deratings for when they run hot? How much headroom do you guys leave for selecting power resistors for continuous duty? Obviously the higher the wattage the larger they get which is why I went for a smaller size.

Also the resistance seems pretty stable with temperature, so its not drifting in resistance and causing more power dissipation.

[johnf]

I would be using a 5 watt resistor for 2.5 watts of dissipation. Its not only the resistor you have to consider.
there is also the pcb solder joints and nearby components to take into account

[MRMILSTAR]

Do you have it installed on a metal plate for heat dissipation?

[John123]

I would be using a 5 watt resistor for 2.5 watts of dissipation. Its not only the resistor you have to consider.
there is also the pcb solder joints and nearby components to take into account

Good point, so it's not really standard practice to continually push power resistors near their max ratings? There's space left around it for dissipation.

Do you have it installed on a metal plate for heat dissipation?

No, but it never had one to begin with (it was pulled from something and still has its original lead formings). There's about 1cm gap underneath between it and the pcb and spaced away from other components.

It looks pretty much the same as these ones except rated for 3w, but I'm not sure if they're just designed for short term 3w dissipation. Like I said the temperature coefficient seems pretty stable. Smoke starts coming out after a few minutes.


I mean it's pretty trivial to swap out a higher wattage resistor, I was just wondering if there's some method to the rating madness. On another note I've got a 5w 1 ohm resistor passing 1.3 amps (1.69 Watts) continuously and it also gets too hot to touch, but no smoke lol. In a commercial product design would the designers over spec resistors like this?

Would this be a good choice? 75ppm seems pretty stable and I've seen the low ohm versions used for current sensing and emitter balancing applications in old linear power supplies. https://cpc.farnell.com/welwyn/w22-22r-ji/resistor-ww-7w-5-22r/dp/RE04106

My frequency is about 40khz max.

[davekni]

When you say "As expected it runs red hot", is that literal?  If it's actually glowing, I suspect it's dissipating more than 2.5 watts.

Power resistors, especially wire-wound ones, are generally designed to run quite hot at rated power.  If you get organic contamination on the surface, smoke isn't surprising.  But that should burn off and the smoking stop after a few minutes.  I like to have some margin below spec limits, but regularly use resistors above 50% of rated power.

Does your design mind the inductance of a wire-wound resistor as in the 7 watt link you shared?

Your image of 1.8 ohm 2 watt resistors looks extremely close to bags of resistors I purchased surplus years ago - same values and appearance, with only slight differences in marking.  I think these are film resistors, not wire-wound.  The inductance is low for wire-wound, around 15nH if I recall my measurements correctly.

[John123]

Well not literally red hot, just hot enough that it burns my finger to touch!

Is the inductance of that resistor I linked likely to be large (into the uH range etc)? It's 22mm x 8mm, or 0.866142 inch x 0.314961 inch.

With the virus stocks of metal oxide versions at my usual sources for parts are all awaiting delivery or the prices have gone up, for example https://cpc.farnell.com/multicomp/mor07wj0220b09/resistor-7w-5-22r/dp/RE05144

It's in series with a feedback winding, so I guess it'll just add a form of leakage inductance.

[davekni]

I doubt it would get into the uH range at 22 ohms.  Doesn't take that much wire unless resistance is well higher.  If I had access to the nice impedance bridge at work, I could quickly measure a few wire-wound resistors near that resistance.  But, social-distancing measures prevent that for now.

It's not quite the same as leakage inductance, as it's included in the sense voltage.  Will just produce a bit of phase-lead.

[John123]

Thanks, well I've got one of those Chinese all purpose testers which can go down to 10uH and to measure really low inductances, I've found I can just put very low value inductors in series with a known 10uH and it adds to the value in 1uH increments.

Wouldn't you know it I've just found a non inductive type right next to the other one I purchased https://cpc.farnell.com/multicomp/mcknp05sj0220aa9/resistor-wirewound-22r-5-5ws/dp/RE07467 shame they have a minimum order price of about $10 and I don't need anything else right now.  I'm kicking myself for not noticing it when I placed my order last night.

Its being used in a nasty switching transformer driver as base bias for an NPN transistor and also adjusts the switching cycle duration, so no current sensing going on. I wonder if bypassing it with a capacitor could help with any self inductance effects.

[klugesmith]

The normal operating temperature for power resistors, properly installed and dissipating 100% of rating, is high enough to seriously burn you.
Like real transformers and electric motors spec'd for 100 degree temperature rise.
If they don't get pretty hot, they are too big and heavy and expensive.

It would be exceptional for a circuit to pass design review with 3W resistors operating at 2.5 W.

Re. inductance: I've looked at some wirewound resistors with R values and sizes in the same ballpark as yours.
They turned the corner at around 1 MHz.   SO L/R = about 1 microsecond / 2 pi.  YMMV.

[John123]

Thanks!

[petespaco]

Just a comment about what's hot (to the touch, etc..)and what's not:
   At work many years ago, we used a UL standard for making sure that hot surfaces were not too hot for users to touch.  Like the cabinet of a business machine.     IIRC,  the upper limit for being able hold your fingers on a surface for any amount of time was 137°F, or about 58°C.

So I just found this:
"Typical maximum temperatures for carbon composition resistors would be around 100 to 120°C and for metal and oxide film types, about 150°C. Wirewound resistors can operate at higher temperatures up to around 300°C."

My point is----- just because it's too hot to touch doesn't mean its over its maximum operating temp.

I suppose one could actually measure the temperature of the device, but don't use one of those non contact IR meters unless you understand how their field averaging works.  Many of them measure the AVERAGE temperature at an 8:1 ratio.  This means that, if the meter is 8 inches away from the surface to be measured, it will average the temps within a one inch circle, not just your resistor!

Pete Stanaitis
---------------