# 3.3V regulator does nothing for first ~12 seconds after power-on, then outputs 5 volts

In my circuit I'm using a TPS78633DCQ low-dropout regulator to regulate my 5 volts power input down to 3.3 volts. However, when powering up the circuit board for the first time today and measuring the voltages, I was getting 5 volts on the 3.3 volt rail.

There was no smoke or anything — I was watching quite carefully for this during the first power-on — and none of the chips on the board were getting hot when I felt around with my fingers; I was also very careful to check for shorts not just to ground but from each rail to every other rail before turning it on, and there weren't any.

On probing it with my oscilloscope and examining it while turning it on, it looks like the regulator isn't actually turning on at all until 10-12 seconds after first applying power, and when it does it jumps all the way to 5 volts:

(yellow is 5 volt rail, pink is 3.3 volt rail)

Note that this is on an assembled printed circuit board, so I don't believe it's possible this could be a grounding problem; the ground and thermal pins both have vias directly to the internal ground plane as close by as I could manage.

This is the power supply section of my schematic:

U17 is the 3.3 volt regulator that appears to be behaving incorrectly; initially I suspected it may have been assembled with an incorrect chip, but the chip on the board at the very least has the correct markings:

Here's my full schematic, pcb layers, and EDA project, for reference:

Note that other 1.8 and 1.1 volt rails did come up to their correct voltages after that, so those regulators are at least functional (were able to regulate down the 5 volts rather than the 3.3 they ought to have been getting).

However having 5 volts on the 3.3 volt rail did exceed the voltage rating on the TS3USB221 muxes (datasheet says 4.6V absolute max for VCC) and TUSB4041 hub chips (datasheet says 3.8V absolute max for VDD33), so it most likely has damaged this board. Hopefully I can fix the problem with the regulator on the other boards before powering those ones up.

What is going on here? Did I not connect the voltage regulator (enable pin?) correctly, or something? Were my boards assembled with a wrong/counterfeit chip despite the markings? I have never seen a behavior like this before, and I have essentially no idea what to actually do about it.

For the moment, I can remove the regulator from my other boards and feed their 3.3 volt rails with a bench supply (in order to test the other features of these boards), but I don't even have any concept about how I would redesign the circuit to make this not happen.

• Note to OP that although the symbol pin is labelled NR/FB, per datasheet the FB bit applies only to the adjustable version at which point the pin should connect via a resistive divider, and the NR applies only to the fixed versions which should be left unconnected, or bypassed to ground via a low value capacitor. – Tom Carpenter Aug 2 at 18:05