I have a curiosity. I noticed that some batteries have more than 1.5v when full. Using 4 of them in series would exceed 6 volts (the absolute maximum specified in the atmega series of microcontrollers)
My idea was to use a diode in series with the batteries, to lower the voltage when the batteries are full. To then detect that the voltage has gone to safe levels (by using the internal voltage reference and a voltage divider on vcc) and engage a P mosfet to bypass the diode.
Is this a good idea? Or doesn't even matter if the voltage exceeds 6v by 200mv or so.

 

Is this a good idea?

No. Operating any electronic component at or above its Absolute Maximum Ratings is a really, REALLY bad idea. Don't do it.

Use a low-dropout voltage regulator to give the Atmega its recommended operating voltage.

 

No. Operating any electronic component at or above its Absolute Maximum Ratings is a really, REALLY bad idea. Don't do it.

Use a low-dropout voltage regulator to give the Atmega its recommended operating voltage.

But it won't be above its absolute maximum ratings, the vcc will come through the diode, at a safe voltage. Only when the voltage drops enough the microcontroller will pull the gate of the P mosfet low and thus bypass the diode.

 

But it won't be above its absolute maximum ratings, the vcc will come through the diode, at a safe voltage. Only when the voltage drops enough the microcontroller will pull the gate of the P mosfet low and thus bypass the diode.

Suit yourself. I've been designing electronics professionally for a long time, and learned-- the hard way-- many years ago to stay well below components' Absolute Maximum Ratings.

You asked, "Is this a good idea?" and I gave you my professional opinion: NO. But you go right ahead and do whatever you want.

 

Suit yourself. I've been designing electronics professionally for a long time, and learned-- the hard way-- many years ago to stay well below components' Absolute Maximum Ratings.

You asked, "Is this a good idea?" and I gave you my professional opinion: NO. But you go right ahead and do whatever you want.

I wasn't questioning your experience. I agree it's not a good idea to go beyond the limits. I was simply stating that in this case, the voltage cannot go over the limits, since it has to go through a diode first. Only when safe voltage is sensed is the diode bypassed. Sorry if I seemed to dismiss your opinion, it wasn't the case, my experience with electronics is limited.

 

Use a buck/boost switching converter then you wouldn't have to worry about over or under battery voltage.

 

If you're going to sense the voltage and decide whether to turn a FET on, you're giving yourself a fair amount of trouble to turn a quick and dirty idea into something complicated. A voltage regulator looks better than that. I'm also wondering whether it matters what the processor's input voltage is, as the battery output declines. Maybe you could live with the diode, and then as the batteries die, just keep living with it. By the time you get to the point where the processor stops, you wouldn't have much more life available even with the diode out of the circuit. That's because when 4 cells in series are failing to run the processor (assuming it can work from 3V or so) they're losing their remaining power very fast anyway.

 

Most ATMEGAS can run at much lower voltages. OBW0549's advice is well worth heeding, especially if you are going to make more than one of whatever you are designing.

I would try to stay at 5V or lower if your chosen clock rate will allow it, 3.3V would be a good choice.