0
\$\begingroup\$

The question applies to a buck converter design on my board: Schematic section

Here is a 3D view of this section: 3D of the section

Layout of this section: L1: L1

L2:

L2

Layer 3 is power and layer 4 is again a full ground plane.

Here is the Webbench link to the design i was trying to follow: https://webench.ti.com/appinfo/webench/scripts/SDP.cgi?ID=D3257B665103C7EC

The design was intended for 8A current to drive a long strip of LED's.

However, as i connected it across the load the power supply seems to turn off (5V led turns off) around the 5A mark. The diode and FET appear to be getting very hot and my copper pour is clearly not doing a good job. I believe it might be hitting a thermal shutdown because as i reduce the current it turns back on. In hindsight the FET looks awfully tiny to support 8A.

The issue is i don't have enough time to get another design sent out and need a 'quick fix' to work for now buying me time till the next iteration.

Is there something i can do to make it work at a slightly higher current? Some kind of heat sink or if i could put a perf board on top of the inductor with a higher rated mosfet?

\$\endgroup\$
6
  • \$\begingroup\$ The SiA441 is not rated for that much current unless you can keep its case temperature below 70 °C. Maybe you can, but I agree with your assessment that the FET looks too small for that much current. Do you know what the case temperature is? Can you measure it with e.g. a thermal camera or something? \$\endgroup\$
    – Hearth
    Jan 13, 2022 at 18:55
  • \$\begingroup\$ I was looking for a thermal camera earlier and couldn't find it. I will do that on Tuesday at some point. Its definitely over 70C. Is there a way to make a makeshift heatsink with the information i have shared? \$\endgroup\$
    – Hasman404
    Jan 13, 2022 at 19:01
  • \$\begingroup\$ Blow compressed air directly on it and see if that extends the usable current range. If it does, you know it's a thermal issue. But that part seems too small for such a large current. Expanding the island size of L2 might help slightly, but you'll probably end up using a bigger part. Could make a dual-sided adapter PCB, solder that to the board, then solder a bigger part to it. But this introduces a host of other thermal issues. Even if you try thermally-conductive epoxy, this won't be as good as a purpose-built board. \$\endgroup\$
    – rdtsc
    Jan 13, 2022 at 19:19
  • \$\begingroup\$ Could try a thermally-epoxied TO-220 package, and "dead-bug" the pins. Just to test. \$\endgroup\$
    – rdtsc
    Jan 13, 2022 at 19:22
  • \$\begingroup\$ Could i also add, is it not weird that as soon as i reduce the current from 5A where it seems to turn off, and bring it back to 4.6A the circuit restarts. Almost as if its the buck that is turning it off. Surely if it's the FET temperature than it should blow up when it reaches those max temps and the circuit shouldn't be resetted. The buck doesn't seem to get that hot so it's not it going into thermal shutdown. \$\endgroup\$
    – Hasman404
    Jan 13, 2022 at 21:11

1 Answer 1

1
\$\begingroup\$

What you can try :

  • add a fan blowing on your FET
  • add a heat sink on top of your FET (can be combined with fan)
  • if the PCB itself is hot (at least under the FET), then IF PLANE FOR IS NOTHING BUT GROUND, you can put your PCB on a big metal plate acting as a big heat sink for the PCB itself. Be carefull : if you have anything else than ground on the bottom side, the metal plate will cause a short circuit. If you can't put the plate, you might put another fan on the bottom
  • if you can, just run your PCB outside or in the fridge : you gain already 20°C
  • replace your FET by one supporting higher current (with heatsink if needed), and just run wires from your PCB to your new FET
\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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