Nintendo Switch Fan and Joycon issues

So here’s an odd one for anyone interested in NS repairs. I have a V1 NS which was mildly water damaged. Disassembled and cleaned, probed around and no shorts, booted fine. The fan wouldn’t turn off, if the battery was connected- the fan was on. I replaced the 9pin LDO BGA on the back of the board and bingo! Fan now runs- but only when plugged in & receiving 15v (not 5v) from the USB. Hmm, maybe my reflow was bad? Try again- same issue, bad chip? Replace again - same issue. Well, maybe it’s the M92T36? So I bust out the parts cannon and replace it. Great! The fan is working perfectly - for about 10 minutes and then it starts to only work while receiving 15v from the USB. I am at a loss, I have no shorts around the usual suspects, and this board doesn’t have a diode near the fan port. The voltage injection method, I feel doesn’t apply to this situation. My next shot was going to be the BQ chip.
A few points which may be relevant:

*The joycons connect and work fine, however they only charge while the NS is receiving 15v

*I’ve tested the board with a known good fan

*I’m using atmosphere with fan parameters set to always run for testing, so it’s not likely a thermal sensing issue

*I reflowed the M92T36 thinking it was a bad connection - no change

*The fan has 1v on the (signal?) pin and 0v on the pos pin while in handheld mode. When plugged in it has 0.5v sig and 5v pos.

If there are any NS Oracles out there and have come across similar before I’d love to hear your thoughts! Or if anyone can see something I’m missing I’d really appreciate the input.

On side-A on the right side is a circuit for powering the fan and Joycons. Depending if VBUS is active (powered by usb) the input for VFanJoyCon is switched.

From your description I assume that the VSYS powered side is not working or not switched on.

Hi Calvin, thanks for your comment and diagram. I’ll pull the board out again later today and take voltage drop readings. Next to the component marked vsys is the LDO I replaced(twice) is that a likely suspect again or do you think I’m looking at an issue further upstream on the vsys line? Maybe M92 again? I found it strange that it worked as designed for a short time. Cheers.