2
\$\begingroup\$

I'm designing a custom 8-Layer board. The main power supply is based on a 24V-12V Iso DC-DC (Tracopower TEN 40-2412WIR). I'm a little concerned about the heat generated by this device because it's gonna be operating almost at full load in a close environment (no forced air allowed).

The datasheet of Tracopower TEN 40 specifies a non-conductive FR4 material for the baseplate: this makes me think that the device was designed to dissipate most of the heat generated by the top of the device itself (because of copper case), using a heatsink.

I'm evaluating some ways to spread as much heat as possible far from this device without a heatsink on top of it.

I was thinking about using the GND pin of the DC-DC to help conduct heat away from the converter to the main PCB, using a region outside the DC-DC in contact with the ground polygon (on the layer immediately below it). In this context, I planned to place an exposed pad to hook the convection under the Traco case, as per attached figure. It should favor the dissipation along the dedicated polygon outside the component footprint.

enter image description here enter image description here

I don't know if it can be a good or a completely useless idea. Alternatively, are there other ways to spread heat away from the DC-DC converter?

\$\endgroup\$
4
  • \$\begingroup\$ What kind of enclosed area? You can't mount it to so the converter is pressed up against a metal casing to dissipate heat outside? \$\endgroup\$ – DKNguyen Jul 29 '20 at 23:13
  • 3
    \$\begingroup\$ the underside of TRACO's are typically epoxy as they typically place their small pcb into the metal lid and pour in to encapsulate. there will not be that much thermal flow from the bottom \$\endgroup\$ – JonRB Jul 29 '20 at 23:18
  • \$\begingroup\$ @DKNguyen Unfortunately, I can't mount it in such a way because there are other boards alongside before the enclosure. \$\endgroup\$ – alediben Jul 30 '20 at 6:47
  • \$\begingroup\$ "it's gonna be operating almost at full load in a close environment" - Sounds like a bad design. Any reason you can't use a higher rated converter? Why can't it be heat sinked? \$\endgroup\$ – Bruce Abbott Jul 31 '20 at 0:26
2
\$\begingroup\$

The casing maximum temperature is 105 C. (very hot)
At 90% worst case efficiency and full load it only loses 4 Watts.
At 60C Ta you can still do maximum load since with 10.8 C/W it's at 103.2 degrees celsius.

You can create a thermal pad with a via-stitching grid to all layers under the module and put some thermal transfer pads on it. Maybe transfer the heat to a plate on the bottom mounted a similar way. Or directly to the enclosure with an aluminum spacer.

Test it! Solder some wires on the thing and grab some copper clad fr4. Cut the fr4 to strip and put it on with some thermal adhesive pads.


But the difference between no heatsink and heatsink is only 0.5 K/W, so I doubt any case mounting options will be very effective.

If you need it to run 100% load, you're going to have to limit your ambient temperature to below 60 C. Or get a different series, eg: TEP 75WI.

\$\endgroup\$
2
  • \$\begingroup\$ I have a doubt about it. SInce TRACO's baseplate is realized in FR4, probably it will not dissipate so much. Nevertheless, on the GND pin, I would have a canalization of heat. My question is... If I leave this connection by adding thermal pads, is there the risk of heating up (improperly) the exposed pad, losing the control of the situation? I mean, the exposed pad, prompted by the heat on the GND pin will heat up the entire pad, heating the component base as consequence . \$\endgroup\$ – g.mezzina23 Jul 30 '20 at 9:25
  • \$\begingroup\$ @g.mezzina23 Heat transfer trough a pin will also be limited, I would not expect siginificant improvements. Only problems with manufacturing. \$\endgroup\$ – Jeroen3 Jul 30 '20 at 12:32
1
\$\begingroup\$

Unless you expect to have solderability problems, have WIDE metal around that GROUND pin.

At least 3X wider.

That foil has 70 ° C thermal resistance per watt per square.

You have about 2 square of foil, out to the region with numerous vias.

At 2 squares, you have 2 * 70 == 140 ° C per watt.

Widen that 3:1, and drop the thermal resistance to 50 ° C per watt.

You could even use a few vias right alongside the GROUND pin.

==========================================

In your initial PCB, the middle pin (the GROUND pin, right) has some diameter X and the PCB foil on either side (above and below) is about 1.5X on each side.

I would widen that, having 2.5X or 3X on each side (you have plenty of room), so the thermal resistance is much lower to move the heat to the right and those 10 vias.

You might also add a via above and below the GROUND pin, near the pin, to take heat thru the FR-4 to the underlying Ground, if there is Ground foil near the GND pin.

\$\endgroup\$
1
  • \$\begingroup\$ So, are you suggesting to realize a wide polygon closely outside the footprint (full of vias) and to add vias all around the GND pin? What do you mean for 3x, w.r.t which dimension? @analogsystemsrf \$\endgroup\$ – g.mezzina23 Jul 30 '20 at 9:18
0
\$\begingroup\$

I am having the exact same problem as yours with a DC/DC converter and at this moment I am designing your same solution on the PCB. It reached 45º on an open environment with the old PCB design and air temperature of 25º. It will be installed in a closed environment with air temperatures up to 40º, so I am a bit worried.

Another solution I will implement is to tie the ground plane to a metal box to act as a thermal conductor

You can also remove the soldermask on the ground plane if it is on top or bottom layer

\$\endgroup\$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.