AMS1117 bypass

[nadavb]

Hi,

I am connecting an ESP32 dev board to my custom PCB.
I have on my PCB a 5V AMS1117 LDO connected to the Vin pin of the ESP (The ESP board has another 5V-->3.3V LDO)
When I want to program the ESP, I use the dev board's USB for power source. My LDO has now 5V at the output and no input voltage (I disconnect My PCB power source when programming).
Could this setup harm the LDO in any way? Basically the LDO pass device will have 5V at the Drain and floating at the input.
Should I add a protection diode across my LDO? or maybe a bleeding resistor?

Thanks

[moffy]

From the data sheet it sounds like it is generally safe not to have a reverse protection diode except under certain circumstances:

[nadavb]

OK. but even if I add such a diode, the LDO input voltage pin is floating. There will be no current path from output to input. the output pass device will handle 5V?

[mikerj]

OK. but even if I add such a diode, the LDO input voltage pin is floating.

It is not floating, the LDO input voltage will be pulled to the output voltage less one diode drop.

[nadavb]

Assuming this is true, will it damage the LOO?

[ejeffrey]

You should  be fine.  The place you would potentially run into trouble is if the input was shorted or ended up back feeding power to some other circuit.  The internal diode can take a large current pulse from discharging  capacitor but if you have a continuous power supply it could potentially be a problem.

[nadavb]

That is good to hear....
But.... this gets complicated - The main PCB has two more AMS1117 5V LDOs for the other parts of the circuit. They are connected to the same main power supply (9V)
So basically this reverse leak voltage at the 1st LDO (~4.5V) will feed the other LDOs?
How will the other AMS1117-5 LDOs respond to this low voltage at the input?

[magic]

The others will work normally and output 4.5V minus their dropout voltage. Depending on their loads, some meaningful current may start to flow there. Now the question is if the reverse biased regulator can cope with such current flow, unfortunately the datasheet doesn't give any maximum continuous reverse current rating so we are a bit in the dark. If you want to play it safe, you may add a Schottky diode D1 (see moffy's post) or a series diode at Vin to block any reverse current flow completely - this should be OK if normal input is 9V. BTW, with such high input you could use ordinary LM317 or LM7805 too.

[Peabody]

You can leave the 1117's input open, and of course no current will flow out that way.  But its ground is still connected, and voltage applied to the output pin could potentially result in current flow to ground.  In fact we know a little current would flow because of the resistor divider on the output that regulates the output voltage normally.

I suspect very little current would flow back through a 1117.  But if you can isolate it, you can measure that by applying the new output voltage to that pin, with the input open, and seeing how much current flows.  You might start using a 1k series resistor to prevent very large current flows, but when it turns out there are none, you could just connect directly and measure the current flow.

I don't have a 1117 handy to test with, but someone here does, and could do that test.  In my experience in general, if very little current flows back through the 1117's ground, with input open, then no damage will result, and a bypass reverse diode isn't needed.

[magic]

1117 is a floating regulator like 317, so its ground pin is only for the feedback divider. Not much current flowing there.

But if there is other stuff on the regulator's input rail, then you've got a problem.

[Siwastaja]

Yeah, watch for everything else connected to Vin. Including large capacitance, which can pull significant enough current to destroy the small integrated protection diodes. And obviously input short circuit will be disastrous.