Rip apart my "bulletproof" Breadboard/Arduino power supply design

[1276-2449-1-ND]

Made an overengineered power supply for my kids after watching them fumble around with a breadboard and frying an Arduino. Okay it's for me too.

It uses a TPS259621 efuse to shut off the power when a power rail is shorted, has an overvoltage (ie by plugging power supplies into each other), or overcurrent. It can also be configured to shut down all the power rails at once if any of these conditions happens on any rail. The TPS continually checks if the condition clears and then powers everything up when it's okay. If any of the power regulators (MIC29302) goes into shutdown all of them shut off. Startup is sequenced, and each TPS has inrush current management.

It's powered from a notebook power supply. The LED on the affected rail goes out if there's a problem. The current consumption on each rail can be monitored via a TPS pin broken out to the side of the PCB -- It was meant to vary the brightness of an LED depending on how much current is being drawn, but ran out of room and time. Maybe later.

I'm not a power guy -- just followed the datasheets. See anything that could/should be done better let me know. 



Direct link to schematic:

(Not evident on the schematic, but ENABLE are all tied together.)

[westfw]

Doesn't have a way to mount the heatsinks you'll need to handle the power you need to dissipate.

[I wanted a rude username]

Unless the intention is to glue them on with thermal epoxy? But TO-220s would seem like a safer choice.

Also important to consider case design, especially if fanless. With appropriate vents, mounting the PCB vertically will provide better convection than horizontally, due to the chimney effect.

[1276-2449-1-ND]

Thermal dissipation was my concern too. There's vias stitching the bottom of the regulators to the back of the PCB, which is a ground plane. The LM1117 has a separated area underneath since the tab isn't connected to ground -- but it never got too hot at 500mA anyway. The case is designed to have convection pull air from the bottom underneath the PCB.

[wizard69]

Actually the first thing that stood out was the low current capacity for the 3.3 VDC rail.   Are you really sure you want such low power capacity on 3.3VDC.

As for the regulators I'd prefer heat sinks.    On the other hand your electronic fuses are limiting current and thus heat.   

[mariush]

You could have found a better regulator than 1117 series, which is known for instability if the esr on the output capacitor is outside acceptable range...
Most datasheets recommend using electrolytic capacitors on output, as the esr must be between 0.3 ohm .. 1 ohm ... other datasheets say minimum 0.1 ohm...

[twospoons]

+1 to inadequate heatsinking. No way are you going tot be able to dump 10W+ into a pcb through just 9 vias.
Either change  to TO-220 or similar, or redesign using buck switchers to reduce the waste heat.

[1276-2449-1-ND]

Thanks guys.

[Prehistoricman]

I don't suppose you intend all of these to be loaded at the same time? That would be (worst case) 48W dissipated in the regulators. Toasty.

Also we have a really nice image attachment thing on the forum, so you can use that instead of imgur.

[250266]

Anyone knows where I can source TPS259621?

It seems to be out of stock anywhere.

[bsodmike]

Do you mind sharing the Eagle schematic/board? I'm keen on toying with this as well.

This is an interesting alternative although backorder estimate is April 2022:
https://export.farnell.com/texas-instruments/tps25944lrvct/efuse-power-mux-2-7v-18v-wqfn/dp/3123442?st=efuse

[Renate]

Speaking of out of stock...
1276-2449-1-ND is out of stock!