resistor power ratings

[waspinator]

Hi,

I'm building a dummy load based on Dave's design using a similar schematic to sleepy robot .

I thought I could replace those 10 1/4W shunt/sense resistors with 1 2W resistor. The datasheet seems to suggest that it should be able to work up to the 2W in ambient temperatures of up to 70 degrees Celsius. My room is no where near that, but the resistor itself heats up a lot, enough to desolder itself from the test wires.

What's going on? Am I understanding the datasheet incorrectly? Am I expected to have some sort of heat sync and fan cooling solution for the resistor? How would I figure out what kind of datasheet power rating in a resistor I would need?

Thanks

[Neilm]

Note 3 on the data sheet states that that there are minimum pad and trace areas required to meet that rating. Page 1 first table last entry shows that it needs an area of 300mm^2 to meet the 2W power rating

[biffster]

The heat that it can dissipate to air is a lot less than if it were soldered onto a board. That's because conduction through the metal board is a much better heat transfer mechanism than the convection of the air around the resistor.

Although the data sheet says it needs to be soldered to a minimum pad size as Neilm says, you could get the same effect with a heat sink or fan or whatever as a temporary measure while you test it. The main thing is that you dissipate enough heat that the resistor temperature doesn't increase too high, it doesn't really matter how.

[waspinator]

Oh I see. I just used the recommended solder pad dimensions, but those only give 10mm^2 on both pads.

So I would have to increase the size of the A and C dimensions from 3.7 x 1.5mm to to 20 x 8mm? Guess I need to give it more board space.


Thanks

[mmagin]

As I've found, there are two kinds of power resistors.  First, the ordinary leaded axial type and the big ceramic tube wirewound type.  Those don't rely on external heat sinking (though they benefit from not being right against the board or anything you don't want getting hot.) 

Second, surface mount power resistors, power resistors with obvious heat sinking tabs (TO-220) or other attachment points.  This includes all the fancy non-inductive RF load resistors that are rated up to a GHz and stuff.  Those all need heatsinking of some form to achieve their ratings.

[uncle_bob]

Hi

For just a simple dummy load ... cheap leaded resistors are the most affordable way to do it. It's a good idea to derate any resistor about 50% power wise. The ratings are not just based on temperature. They also assume some amount of air flow, often from convection. A leaded part is often mounted in free air with clamp blocks on the leads at a specific distance. The heat sinking from the clamps and the infinite air column likely do not duplicate a leaded resistor in a typical device. The same sort of issues apply to board area specs. It's not just the board area, it's how the air can move around the board (heat's got to go somewhere).

Bob