Hot resistor - why?

[pwnell]

I built this circuit:

https://i.sstatic.net/tgbFM.png

From here: https://electronics.stackexchange.com/questions/646876/how-can-i-decrease-the-relay-release-time-when-the-current-on-relay-coil-is-redu

I measure 30mA through R1 and approx 9.5V across it with V1 = 12V. I used a 1/4w resistor and it got smoking hot. So I replaced it with a 2W resistor of the same value and it still got up to 60C. Why? (My load was a different relay coil but that is immaterial as I measured the actual current).

P=I^2R = 0.030^2*330 = 0.297W. A 2W resistor is almost 7 times more than this value. Should it be getting this hot? What am I missing? R2 also gets hot - up to 60C.

Edit: it is a carbon resistor.

[floobydust]

Carefully check your TL431 pinout, they come in all permutations.
I don't think your circuit will work as expected? The TL431 is not regulating anything without feedback, I don't see any.
To speed up a relay's release time, you allow the back-EMF to happen (i.e. no D1 but instead a zener+diode).

My math the TL431 turns on above 10.9V and then Q1 and the relay are activated.
R1 and R2 should run hot, they are pretty low value, current is too high for what you need. What is the max. or typical Vin?
Even R2=10k, R1=5k I would consider.

[pwnell]

The circuit works fine, it switches off the relay below 10.9V and turns it on above that.

TL431 is connected correctly - I verified with the data sheet from mouser.

But my question is - how is 30mA through a 330 ohm resistor with 9.5V across it a lot? 0.297W through a 2W resistor is not a lot, is it? I am more curious as to why 0.297W through a 2W resistor heats it up to 60C than the specifics of this circuit (for now). I will definitely take your input into consideration for improving this circuit, for now I am just curious.

[mikerj]

I am more curious as to why 0.297W through a 2W resistor heats it up to 60C

How are you measuring the temperature of the resistor?  The operating temperature is not only dependant on the power being dissipated, but also ambient temperature, mounting arrangements and airflow.   If you have this in e.g. a solder-less breadboard with warm ambient temperatures this seems entirely possible.

[Siwastaja]

curious as to why 0.297W through a 2W resistor heats it up to 60C

It isn't that weird. Assuming your room temperature is 25degC, this is just 35degC temperature increase. Unless you have a fan blowing at the resistor, it's cooled by natural convection (hot air starting to rise, and new colder air replacing it), which is a highly nonlinear process: so when the temperature further increases, then cooling power more than linearly increases; thus even if you see 35degC increase at 0.3W, try to increase the power to 0.6W and I'm sure the temperature increase is not 70degC but significantly less.

At rated maximum power of 2W, the resistor will be very hot, close to what its materials can withstand (which could be like 150degC or so). Resistor datasheets do not literally lie, but running parts at maximum ratings requires some care and understanding. And whenever you produce heat within a small space - even just 0.3W is significant if the resistor is small - you have to think how that heat is spread out.

[pwnell]

I used a thermocouple on the resistor body. Ambient was 25C and it was on a breadboard in open air, reaching said temperature in about a minute.

[ledtester]

Here's a past post on using a TL431 as an overvoltage detector which triggers a relay:

Also incorporates hysteresis via R3.

Note it looks like you can get by using much larger resistors - hence less power dissipation.

Designed threshold voltage is 13V - just modify the input voltage divider to suit your need.

https://www.eevblog.com/forum/chat/relay-trigger-when-voltage-above-13v/msg5476018/#msg5476018

[pwnell]

Thanks - did not see that post when I searched, good info…

[PGPG]

P=I^2R = 0.030^2*330 = 0.297W. A 2W resistor is almost 7 times more than this value. Should it be getting this hot?

I have opened first datasheet I found:
https://www.vishay.com/docs/20128/wkxwrx.pdf

At page 4 you have 'Temperature Rise' chart. WR4 is 2W resistor. When you dissipate 2W temperature rise will be 200°C.

[TimFox]

Proper resistor power rating is based on an acceptable temperature rise, as in the Vishay document quoted above.
Resistor materials can handle higher temperatures than typical semiconductors, so the specifications are accurate, but you may not want to run the resistor at that maximum rating for other reasons, such as neighboring components on the circuit board, or the board material itself, or the change in resistance due to the part's temperature co-efficient.
In that example, the WR4 2-watt resistor will have a temperature rise of 50 K at 0.5 W dissipation, reaching about 75^o C.

[floobydust]

Those are the "mini" power resistors I hate, with a passion. Materials science allows their 155-200°C spec but realistically did you notice the maximum power dissipation is pretty much 1/2 rated and its still crazy hot.
The "4W" part WK8: "For values < 10Ω the upper limiting temperature is 155 °C. The power rating is correspondingly lower and can be calculated by R_th."
"E.g. for WK8 - R_th is 60 K/W; to calculate power P for < 10R; T_ambient = 25 °C"
"P = T_{Upper Limiting} - T_Ambiant/R_th = (155 - 25) / 60 = 2.16 W"... for the "4W" part at room temp 

OP, consider adding a latch function (move a wire) or hysteresis so it doesn't chatter and oscillate when it trips, because disconnecting load on a battery always causes its terminal voltage to jump up, which closes the relay, then V drops, relay opens etc. and repeat. If you go below 10.9V the party is over, latch off until a pushbutton is pressed or the battery has some charge recovery say 13V before connecting the load. I've had this drama many times with solar powered systems. It has to be done right.
If the battery is always connected to the UV circuit, you can lean up the circuit to waste less power in both states, activated or off i.e. if it disconnects load below 10.9V yet sits there continuing to drain the battery at several mA it might be bad.

[TimFox]

I concur about modern resistors' power rating in small physical sizes.
I originally learned about resistor power ratings and sizes with the now-discontinued Allen-Bradley carbon composition units.
Looking at their RC07 data sheet:  https://www.datasheets360.com/pdf/5949656958192319054
1/2 W and 2 W units were rated at 70^o C, and derated to zero power at 150^o C (the datasheet only mentions ambient temperature).
I assume the materials used in the RC07 carbon resistors could tolerate considerably lower temperatures than later ceramic-based units.

[SeanB]

Those are carbon composition, so are a rod of carbon blended with clay, and then encapsulated in some filled Bakelite, and then vacuum impregnated after the bake in wax, after being measured, and binned into the various tolerance bins of 20%, 10% or 5%, with those within 2% being the ones selected for being Mil Spec ones. Bakelite does tend to disintegrate past 150C, so that was the limiting thing, plus the wax would long tern in operation boil off at high temperature, and then the resistors would drift with humidity, and also tended to drift up in value as well, if over around 5k in value, and those lower in value would tend to drift down.

[floobydust]

Datasheet Holy Grail number is power dissipation. Resistor manufacturers will stretch that any way possible to compete, they have to. Once one company starts the bullshit others follow, same for power mosfets and MLCC's. It's a trap, what you think you are getting.

Resistor thermal design is quite complicated, I got trashed in thread Resistor Sizes Vary Greatly? before it was all hammered out - you have at least say 20 factors to consider and looking only a few leads to a mess. I decided to avoid the asshole.
Also encountered here https://www.eevblog.com/forum/beginners/is-digikey-marketplace-no-different-from-aliexpress 

OP's 35°C rise with 0.3W giving 117°C/W, for many resistors this is reasonable... for 1/2W parts.
We don't know the p/n or size etc. but I would say it is doing poorly (hot) compared to most other "2W" parts.

Carbon comp is superior for impulse overload handling. But not for much else, maybe in making smoke lol.
A long time ago, rule of thumb you automatically derated a resistor by 50% of rated. Today, I would say 1/3 to 1/4 of rated is a reasonable start.

[Zero999]

R1 is far too lower value. Increase it to something more sensible such as 10k.

R2 could also be higher, say 2k2. It doesn't take that much current to saturate the transistor with such a light load and even if it doesn't go deep into saturation, it'll be enough to turn the relay on.

EDIT:
Here's the correct way to do this. The pull-up resistor should go between the base and emitter and there should be a feedback resistor to introduce some hysteresis to avoid oscillation and relay chatter.


EDIT:
R5 could also be 9k1, simply to use fewer component values, which makes it easier to source parts, which can reduce costs.

[Siwastaja]

OP's 35°C rise with 0.3W giving 117°C/W

... which is, as I explained above, a meaningless number because it linearizes a problem which in real life is highly non-linear. Thermal resistances are simple to use in thermal interfaces between heatsinks because there is no convection. But component-to-air resistances have to be defined with a dT value, and in OP's measurement, the dT value is small. At higher dT, convection cooling starts operating, and thermal resistance value will be better, too.

But I'm not surprised at all you don't know this.

Guys, remember to ignore floobydust. I was considering giving a warning beforehand, but chose not to, and look, he's here again. He is well known to pollute the forum with his resistor power rating conspiracy theory bullcrap. We other engineers have no problems with resistors, but as always, do remember that resistors, like all other electronic components, are made for professionals, and designing components in requires understanding and design skills. You just don't choose any component based on some headline figure but have to understand what every number in datasheet means and how components interact with each other. This has not changed, and older / larger resistors need as much care to design in.

[dietert1]

Why? Heat is caused by electrical power: P = U * I.
Different parts have different temperature limits.
- Hot resistors can kill semiconductors.
- Old PCBs often show burnt areas around power resistors, so the PCB suffers as well.
- Hot semiconductors can kill electrolytic capacitors on the long run. The typical hot rectifier bridge next to the capacitor.
If you mix different parts on a PCB close to each other, the most sensitive part determines the maximum temperature. Derating of other parts is the consequence.

Regards, Dieter

[Doctorandus_P]

Why are you using such low value resistors?

In this application the TL431 is used as a comparator, and it's output is either on (2V5) or off(12V). R1 is only to overcome effects for residual leakage through the TL431. Something between 10k and 100k would be more logical.

For R2, assume 200mA relay current, and an absolute minimum Hfe of 200 of the transistor. To be sure drive current is plenty to put the thing in saturation, we'll use an Hfe of 100 in the calculation. So base current needed is 200mA / 100 = 2mA. This results in R2 = (9V5 - 600mV)/2mA = 4k5

With such resistors, they will not get as hot anymore.