5V regulator gets hot when pulling lots of current

[amcewen]

Testing with no current limit in the WLED software, and 186 LEDs connected (and showing orange) the area around the 5V regulator on the board gets hot.

[amcewen]

Could be the fuse—that seems to be the hottest component when checking things with an IR thermometer

[amcewen]

Using my finger though, the inductor seems hotter than the fuse.

[amcewen]

Swapped the inductor out for one that should cope with 4.2A (peak, and 3.6A saturation), which also works, but doesn't run any cooler :disappointed:

With a 2A limit in WLED, the inductor (well, the highest spot I can record with the IR thermometer—it's tricky to check exactly which component it's measuring) runs at ~80°C. Dropping the current limit to 1.5A it sits at ~64°C. At a 1A limit it sits at 41°C.

[amcewen]

Looking at the resistance of the inductor, even the replacement ones I got are ~130mOhms, so if there's 2A running through that's ~0.5W of power.

Checking Farnell for options with lower DCR values:

• https://uk.farnell.com/tdk/slf12575t-220m4r0-pf/inductor-22uh-20-shielded/dp/2215751 is 12.5x12.5x7.5mm, with 0.0263ohm DCR, and £0.699/each for 100, with 4A current capability

Forward voltage for the diode is 0.55V, so that's dropping 1W at 2A. From this datasheet, giving the cathode a 1cm² pad almost halves its heating.

• https://uk.farnell.com/nexperia/pmeg3030ep-115/diode-schottky-rec-30v-3a-sod128/dp/1829195 has a voltage drop of 0.35V, might even fit on the existing footprint.

I should design a couple of test circuits to try out different options for the 5V boost, at the least one each of the current diode and this possible alternative with the bigger (physical) inductor and a larger cathode pad.

[amcewen]

Using a more precise temperature probe from my multimeter, with the current limit set to 2A I was reading:

• Diode ~76°C
• Inductor ~75°C
• Chip: ~84°C

So it's the chip that's getting hottest... probably needs a redesign of the copper to add more heatsinking...

[amcewen]

Chatting to Andrew Armstrong he suggested looking at the peak current (vs. the average current) given that the power dissipated is I²R, so rises more quickly as current rises.

[amcewen]

Reading through the datasheet for the existing boost chip it seems that those sort of temperatures are what's expected for that current (although the graphs also don't go quite as high as 2A...)

So I had a dig around for an alternative to try, and have ordered a couple of Feeling Tech FP6276B to try instead.

[amcewen]

When we rework the PCB to fix this it would be good to see if we can trim down the size a little too.

[amcewen]

Looking at one of the potential pouches, if we can make the edge with the connectors on 100mm or less that would be good.