Thermal resistance figures for SMD resistors

[HwAoRrDk]

Does have anyone have any links to resources they can share that give any kind of figures for thermal resistance (R_th) for surface-mount resistors?

I am considering using a 2512-size SMD resistor as a minimum load on an otherwise-unused power supply rail (my board will only load one rail, and I'm concerned the PSU may not regulate well if there's zero load on the other rail), but I'm not sure whether thermally it's going to work well. And of course, to ascertain that I need some kind of idea what the thermal resistance of a 2512 resistor may be.

I plan to dissipate about 1.5W from the resistor. I think that might be okay for a 2W-rated resistor - given that rating is typically at 70C ambient, and I'm only concerned with room-temperature operation. Although, I guess suitability depends on how hot it's going to get. Most 2512 that I see are rated for 155C operating temp, but I don't want to get anywhere near that, if at least for the sake of PCB material rating (e.g. common TG130).

The reason why I want to use an SMD resistor is because of space constraints. I thought about using 3W through-hole resistor, but I don't have enough board space in the layout. Can't mount one vertically either due to height constraints.

[srb1954]

You are going to struggle safely dissipating 1.5W in a 2512 resistor even at moderate ambient temperatures.

For the higher power SMD resistors the key factor is the thermal resistance of the PCB, which ultimately dissipates most of the heat from the resistor. The PCB thermal resistance can be improved by increasing the area of Cu attached to the resistor or using thicker Cu on the PCB. Some power SMD resistors require both options by specifying a PCB with 2oz (70um) pads and internal 2oz Cu ground planes to transfer the heat away. However, using heavier Cu is going limit the availability of PCBs and/or restrict the use of very fine tracks on the etch.

I seem to remember Vishay/Sfernice did some application notes on heatsinking higher power SMD resistors so that might be a useful starting point for your research.

[Vovk_Z]

If it is your own DIY device you may solder two or three SMD cases onto each other.

[Someone]

Power rating of resistors, like mosfets, is usually under unrealistic conditions of perfect infinite heat sinking. Good resistor brands/series will have a datasheet showing the derating by ambient temperature for some given footprints (traces or small fills).

[HwAoRrDk]

I seem to remember Vishay/Sfernice did some application notes on heatsinking higher power SMD resistors so that might be a useful starting point for your research.

This one? https://www.vishay.com/docs/53048/pprachp.pdf

Thanks, this is very informative.

Judging by the information therein, I think you're right, it's going to be a tough ask to get a 2512 resistor to dissipate that much power, or even if I reduce it to, say, 1.2W. Either way, the resistor is going to be over 125C.

I think what I'm going to do is leave it up to the user to provide a dummy load to the unused rail on the PSU (should it prove necessary). After all, the user will be the one providing the PSU, so it will be their problem if their choice of PSU doesn't regulate well with unbalanced loading. Then they can use whatever form of load resistor is most convenient for them - wire-wound, cement, aluminium-cased panel-mount, etc.

[wraper]

If it is your own DIY device you may solder two or three SMD cases onto each other.

It's pointless because there will be almost no improvement in heat dissipation and they will run almost as hot.

[Vovk_Z]

It's pointless because there will be almost no improvement in heat dissipation and they will run almost as hot.

I was waiting for this response. Yes, that's no improvement in heat dissipation, but 50% less power dissipates in each resistor case. So, at least you'll have some margin.
You typically have to derate the resistor when it is too hot (and it will be too hot, with 1.5W for 2W rated). A typical derating curve is in that Application Note in the attachment two posts earlier.

[thm_w]

As long as the temperature is not so high to melt the solder or delaminate the PCB.

For one off, probably some sort of heatsink with non-conductive thermal epoxy would work as well.

[Someone]

It's pointless because there will be almost no improvement in heat dissipation and they will run almost as hot.

I was waiting for this response. Yes, that's no improvement in heat dissipation, but 50% less power dissipates in each resistor case. So, at least you'll have some margin.
You typically have to derate the resistor when it is too hot (and it will be too hot, with 1.5W for 2W rated). A typical derating curve is in that Application Note in the attachment two posts earlier.

Baloney! The linked application note:

This one? https://www.vishay.com/docs/53048/pprachp.pdf

Goes into explicit detail about how the termination/copper area is dominating the performance, almost no difference from the case size! Adding two or three packages on the same footprint will do almost nothing to improve power dissipation, vertically finned off the board a minor increase in convection (or a bigger increase under airflow). But because the paralleled parts are still tightly thermally coupled to each other, they aren't sharing the temperature between them, they reach the same temperature.

[SeanB]

how about 3 or 4 SMD resistors on a large copper pour on the board, you can use areas that are not in use for this, routing just the power rail there, and connecting the pour to your common ground. Spreads the heat out, and you can have the pads there, but not populated, for adding the resistors on for those power supplies that fail to regulate correctly.