4.5v as a power source?

Following on from this thread
http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1163418637/0
I was investigating how to run a 328 from as simple mobile supply as possible.

According to spec, 4.5v (from 3 AAs) should be too low for 16Mhz operation, but my project "works" (I put it together before discovering this limitation!).

So I've just been doing a little experiment - basically I've been letting the batteries run flat to see when in conks out.

Last night it was down to 3.4v and still going strong - which should be WAY out of spec on the speed grade!

Unfortunately I forgot to turn it off when I went to bed and at some point it passed the threshold. But I got about 3 days continuous use which is not bad at all, and perfect for the application.

Obviously this is a statictical sample of one, but it's got me wondering if a 4.5v battery pack is actually a viable power source in the real world...?

What is the actual effect of running too low? Does the timing just become unreliable (which s not an issue in this case), or is there a more serious consequence?

Cheers

The Atmega328 is rated for 20MHz operation at 4.5volts, so you are well within spec running at 16MHz at that voltage.

I have tested a number of Arduino 16MHz boards running at 3.3 volts and have not seen a problem. I don't recommend relying on that, but 3.8 volts or more should be within spec for 16MHz operation

Even though it works, the specs are there for a reason. Maybe it works fine at 3.4V, but will it work if the voltage is low and it's cold? Or does your current device work, and maybe the next three out of the tube do, but the one after that flakes out?

Besides reliability issues, the ADC will return garbage unless referenced against a known source. You'll need to take steps to create a reliable voltage reference if you need an ADC while operating at sub-spec voltages.

-j

It depends on the application.

The battery does not supply a regulated voltage. If you draw a large
currents the voltage will dip. If you dip below the brownout voltage you reset.
Different battery chemistry's have different series resistances and will have
different drops.

If you have a properly designed linear regulator circuit in series with the battery you will not see a dip. 4.5V would be a very good size for a 3.3V system. For a linear
regulator you are at 75% efficiency (for a fully charged battery). There is enough
of a voltage difference to make selection of an LDO easy. You can get cap-less
LDOs which only require small filtering capacitors.

(* jcl *)


www: http://www.wiblocks.com
twitter: http://twitter.com/wiblocks
blog: http://luciani.org

Something like the LP2988 would provide 3.8 volts (spec for 16MHz) at battery voltages down to 4 volts.

I forgot to mention that for a 3.3V system I would use a XTAL frequency of 12MHz (or less). This meets the specifications shown in the graph.

(* jcl *)

Thanks for the input. I appreciate the circuit can be designed to run at a lower voltage, but this requires additional components/cost/space and also possibly modifying the bootloader for lower crystal frequencies, as the other thread covered...