Regulator output voltage without input voltage?

[zapta]

Hi all,



Consider the circuit above with the 5V regulator being MC78M05BDTRKG.  The circuit is powered by other 5V in *or* 12V in but never both  (the other one is left open).

When the circuit is powered by 5V in, the regulator has 5V on its output without input voltage. Is it safe for the regulator? Will it consume power from the 5V input? I could not find in the datasheet any reference to this scenario.

Thanks

Z.

[Simon]

best thing is to test it as it could vary from reg to reg. You could switch it or put a blocking diode after the reg if you can afford the voltage loss

[Simon]

or use an LM317 and take the feedback from after a blocking diode

[Zero999]

It's not a good idea because the output transistor would be subject to reverse voltage, although most transistors can normally handle 5V with no problems.

To make sure nothing bad happens, connect a diode between the input to output with the cathode on the input.

[zapta]

I am using only the 5V input but the board (Digispark) came with the regulator pre-installed. My dilemma is whether to leave it as is or remove it. The board's  developer says that it is ok but I could not see any reference to it in the chip's  datasheet.

[Simon]

so was the board designed to do what you are proposing ?

If you want to supply 5V in can you put a switch in to disconnect the reg ? or maybe just short the input to output (with no voltage in)

[zapta]

so was the board designed to do what you are proposing ?

The developer says yes but I am not sure based on what (it's not that he is getting a special version of those regulators).

[IanB]

You could presumably circumvent any problems by connecting a diode between the 5 V input and the 12 V input to the board? When using the 5 V supply that would feed ~5 V to the input side of the regulator and prevent it seeing any significant reverse voltage.

[Stonent]

What's the maximum current consumption we expect on the 5V line?

[Stonent]

Ok how about this? Use a transistor to break the regulator ground connection when there is no +12V coming in. Then at least there shouldn't be a complete circuit.



EDIT: Just thought, maybe there should be a diode between the transistor base and the regulator input on the off chance any electricity that somehow gets through the regulator doesn't activate the transistor.

[edavid]

I am using only the 5V input but the board (Digispark) came with the regulator pre-installed. My dilemma is whether to leave it as is or remove it. The board's  developer says that it is ok but I could not see any reference to it in the chip's  datasheet.

I have blown up regulators by connecting them backward, but that was with the output (mis-)connected to the load.   The board's developer says it won't happen in this case, so why are you worrying?  Test it yourself, and if there is any problem, just ask for a refund.

[Simon]

Use the above circuit but use a mosfet instead as most BJT's have a 0.2+V drop on them which will lift the output by 0.2V

[Stonent]

Use the above circuit but use a mosfet instead as most BJT's have a 0.2+V drop on them which will lift the output by 0.2V

Ok I think I see what you mean. The .2 drop will be lost to the ground which could affect the final output.

I had considered a mosfet but was concerned about leaving the gate floating. I thought about maybe a resistor back to ground from the gate but wasn't sure what affect it might have on the regulator.

Oh well at least my logic was sound. I thought about several ideas before the transistor. But they all seemed to be parasitic to the 5V output. The one that I liked the most before this one was a diode with a resistor across it. When the voltage was coming out of the regulator it would go through the diode but when reversed it would go through the resistor which would cut the current down hopefully significantly enough that the regulator wouldn't care, but again it was parasitic to the 5V output.

[Simon]

Yes just put a 100K resistor from gate to ground, it will have no effect on the regulator it will just be a smal;l load on the input supply. DO NOT out a reverse diode across the reg if you do this as it will feed power back to the input and switch the mosfet on.

[Zero999]

The transistor has no base resistor so will blow up.

A MOSFET won't be much good either since the output transistor of the regulator will still be subject to reverse voltage.

The best option is a diode connected between the input and output of the regulator.

[Simon]

A MOSFET won't be much good either since the output transistor of the regulator will still be subject to reverse voltage.

Would removing the GND ref shut the regulator down ? with no GND ref the whole thing should go open circuit, I'm sure it could cope with 5V

[Zero999]

It wouldn't work using a MOSFET or BJT.

There would still be a feed turning the MOSFET or BJT on through the regulator's output transistors.

[Stonent]

Please excuse the crudity of this drawing, someone decided to draw an animation of a man jumping off a cliff with my copy of DaveCAD and I ran out of concurrent licenses.

[Zero999]

That's better, although, it's easier and more effective to just replace the transistor and resistor with a diode between the output and input of the regulator as has been mentioned numerous times.

[Stonent]

That's better, although, it's easier and more effective to just replace the transistor and resistor with a diode between the output and input of the regulator as has been mentioned numerous times.

Perhaps but I enjoy the challenge of coming up with a different solution.

[tszaboo]

Or you might even consider to use a proper solution.
http://www.ti.com/lit/ds/symlink/lm5050-1.pdf