3x 5W through hole vs 10x 2512 surface mount resistors?

[incf]

I have a design where I need to dissipate 2.7 watts continuously through a bank of resistors (actually, I have 3 identical banks with similar loads). I have two choices:

• Use 3x 5W through hole resistors. At 900mW per element (They are physically very big).
• Use 10x 1W 2512 surface mount resistors. At 270mW per element.

(I have more than two choices, but I think the above are pretty similar)

I currently use the through hole resistors, but they are pretty bulky. This PCB (which is primarily SMT) currently has banks of through hole resistors for various functions. I am trying to evaluate if SMT 2512 resistors might be a viable alternative.

I want to keep the temperature rise low enough to not melt or soften cables, tape, etc. that may incidentally touch the PCB.

For a resistor bank like this which is preferable (SMT vs TH) to you and why?
Which can be made more reliable?

I think this device might be occasionally dropped down a concrete staircase. I wonder if the SMT resistors might crack.

SMT resistors tend to occupy a smaller area. They may have more trouble dissipating the heat. I suppose they could be spread out over the same area and alternately placed on both sides of the PCB. I almost feel that I might need to calculate surface areas.

I have trouble estimating temperature rise. I hope to achieve something roughly equivalent to the through holes.

They will be mounted on the edge of a PCB and allowed natural convection cooling inside of a large enclosure.

Image showing the size difference between the two schemes, and the approximate width of the PCB.


A possible way of arranging the 3 banks of SMT resistors

[thm_w]

Since you are well below the ratings of the resistor the temperature rise shouldn't be as much of an issue.

Through hole is usually going to be easier for dissipation, you take the heat element and bring it away from the board and allow more surface area for airflow. Its just a question of slightly higher assembly cost, and higher board height.

"Dropped down a concrete staircase" would be a problem for either SMD or through hole. You'd want a way to mitigate that, potting the board in and enclosure, mounting the board in a way that decouples it from the case, etc.

https://electronics.stackexchange.com/questions/416083/calculating-smd-resistor-temperature-rise

[mariush]

There's also the option of DPAK / D2PAK  (to-252 , to-263)  25-35w resistors ... bigger surface area to solder to circuit board and spread the heat directly into the copper area.  See https://www.digikey.com/short/3dpp43h8

Even to-220 resistors could work, could screw them down to a long narrow piece of aluminum (I'm thinking ram stick sized) and have the aluminum screwed down or locked to pcb using clips / springs
  ,, ex https://www.digikey.com/short/0trppdt1

of course it's more expensive...

[Niklas]

I would argue against the 2512, mainly for the higher risk of cracking. Location and orientation on the board are sensitive to bending stress. There is also a mismatch in thermal expansion coefficient between ceramic and FR4 that will be sensitive to fatigue cracks due to thermal cycling.

Through hole mounted resistors with pigtail bend to add flexibility and mechanical fixation of the body would be a better option.

[coromonadalix]

same here  to-220 or else  resistors  or standard thru holes ones with elevation over the pcb to help air flow around them


what do you need for tolerances  1%  0.5%  0.1% ?

[Hydron]

The other thing to bear in mind is that having lots of hot resistors on the PCB will reduce their individual dissipation abilities.

[incf]

I would argue against the 2512, mainly for the higher risk of cracking. Location and orientation on the board are sensitive to bending stress. There is also a mismatch in thermal expansion coefficient between ceramic and FR4 that will be sensitive to fatigue cracks due to thermal cycling.

Through hole mounted resistors with pigtail bend to add flexibility and mechanical fixation of the body would be a better option.

Would a higher number of smaller ceramic chip resistors be preferable? (eventually the number of solder joints becomes rather large)

TO220/DPAK resistors are pretty specialized. It's not a matter of cost, but rather, "is it a commodity part that is well stocked everywhere?"

[Solder_Junkie]

SMD components dissipating 2 or 3 Watts will heat the print board significantly. I would opt for through hole resistors mounted clear of the board.

SJ