IANAEE but I would think that measuring the voltage across the terminals won’t tell you how much capacity is left and whether it can do useful work. To test a battery you would have to put a load on it and measure the current.
But I don’t know what current you should see for what load. Is there a rule of thumb I could use to put a resistor between the terminal of a 1.5v battery and the probe and measure current?
each battery type, size and design will have a different current rating.
I’ll take a stab…
Volts x Amps (Current) = Watts
So… If you took an Amp probe and measured current while the device was under load…
…and multiplied that current times the rated voltage…
you’d get the actual watts of the device under load…
and compare that actual watts vs the rated watts of the device…
Is that right, or effective?
I dunno. But it’s my SWAG.
Close.
The only way to do this accurately is to measure the total energy that the battery has delivered, and to compare that to the battery’s rating. This is what all modern laptop battery do - they have a “smart” charge controller built-in, which manages both the charge and charge remaining functions.
There is no accurate way to use a meter and a load to measure the capacity remaining in a random AA cell. There are “battery testers” out there which claim to do it, but they are making all kinds of assumptions on the battery chemistry, the rise in internal resistance, the original capacity, and the discharge curve.
In practical, real-world terms, yes, the voltage you read is an indication of how much life remains in the battery. As the amp-hours of charge in the battery decay there is a voltage depression which moves along a curve (with different shapes to the curve depending upon the type of battery - in dayes of olde NiCad batteries had a very flat voltage curve, with a sudden drop-off at the end, which meant people like cavers were cautioned to expect a rather quick progression from “light” to “lights out” using NiCads, something I witnessed on more than one occasion whilst caving).
I am not an EE and I do not have references to explain why this is so. Nor am I claiming that “exactly X% depression means Y% of battery life remains.”
Here Discharge tests of Alkaline AA batteries 100mA to 2A are some easy to interpret graphs which demonstrate voltage versus consumption curves.
Two points to notice:
So to convert any given voltage measurement into a capacity-remaining estimate, you must know the particular behavior of the make & model (& batch and age and temperature?) of the cell under test.
So to convert any given voltage measurement into a capacity-remaining estimate, you must correct for both the history of the cell and its intended use going forward.
Much as it pains me to disagree with **Una **…
In practical, real world terms, no. You can’t use a multimeter to give a useful indication of remaining cell/battery capacity.
There are various “levels of effort” for testing a cell. A simple test will make assumptions which may or may not be accurate. A complex test obviously take more time and effort, but will be more accurate.
The simplest test is to measure the open circuit (no load) voltage of the cell using a digital voltmeter. For a given chemistry, the nominal open circuit voltage is a function of the cell’s charge. This is a surprisingly good test, given that it’s so simple. Just make sure your DVM is accurate to 10 mV or better. The primary deficiency is that it doesn’t give you any idea on what the cell’s internal resistance is. It also assumes the battery is in good condition, and the temperature is ambient.
A better test, then, is to measure the cell voltage under two different loads. As an example, you could measure the cell voltage with no load (I = 0), and then measure the cell voltage with a load that draws 30 mA (or whatever). In addition to the percentage of charge on the battery (which is determined by the open circuit voltage), you can calculate the cell’s internal resistance using this method.
The previous procedure does have a problem, though: it calculates the internal resistance for constant-current loads. Many devices “pulse” current from a battery, thus it would be better to measure the cell’s internal resistance using a load with a “pulsing” duty cycle. This is beyond the scope of Joe-homeowner.
And then it can get even more complicated… sophisticated battery testing systems evaluate a battery over temperature using programmable loads with complex waveforms.
For most people, simply measuring the open circuit voltage is sufficient. Just keep in mind that correlating the open circuit voltage with percentage of charge is a function of the battery chemistry. For best results, measure the open circuit voltage when a particular cell is new, and use this value as a baseline for subsequent measurements of this cell.
Throughout the year, I gather up all the loose batteries I find in the house (sloppy kids, badly packaged Costco zillion-packs which have fallen apart, etc.) and stuff them in the closet in a jar. Then, once a year or so, we test them all with the “Digital Engine Analyzer” that looks like one of these: digital engine analyzer - Google Shopping
using the DC voltage settings. New 1.5v batteries usually test at 1.6v, and used ones test lower.
Hence the irony. Very modern batteries have a very flat voltage vs capacity left discharge curve. Which makes it that much harder to tell how much is left in any given battery.
By “very modern batteries” do you mean NiMH and other rechargeables? Una’s method works for me with plain old alkalines.
You answered your own question there 
For my purposes I just need to be able to tell if a battery will work if I use it in a device, not estimate how long it will continue to work. I am basically looking for Pass/Fail. For example, I have a pile of rechargable batteries (Duracell NiMH, 2650 mAh) and when I pick one up off the shelf I need to figure out if I need to recharge it before using it.
There is no way to be sure. A cell’s open circuit terminal voltage can measure normal but then drop to an unusable value when you put a load on it. I think the best test is put it to work in a device where you can see or hear the work being done (flash light, nose hair trimmer) and see what happens.
Then colour me confused as to why it works 100% of the time with me. I have a GPS which eats batteries. As soon as I see voltage drop below 1.52V, there is very little life remaining in the battery. If I take a new battery out of a pack and test it, anything below about 1.52V will die within 4 hours, as opposed to lasting 24-48 hours. This works with all alkaline batteries. I didn’t say that one could predict “X voltage means X hours of life left” in all situations, but if you test a battery and its voltage is, say, 10% off the other batteries, I guarantee that one has less usable energy in it. 
Did you mean 1.25v? If so then I agree. A cell that can’t even maintain 1.5 volts under a 10Meg Ohm voltmeter test certainly isn’t going to be worth much when a real load is connected.
So, a low open circuit test reading means a worthless or near-worthless cell. A “normal” open circuit test reading doesn’t necessarily mean a good one.
“1.5 V” NiMH cells are usually closer to 1.2 V even when fully charged, too, which can generate a lot of false positive “low battery” warnings in some devices.
i have devices that will run for about 15 minutes with a cell at 1.2V. they were engineered in a way that makes them not useful with rechargeables. that bites.
I have a new 8-pack of “Duracell Ultra Digital” AA batteries. Their voltages are:
1.605
1.616
1.622
1.604
1.617
1.616
1.594
1.628
I also have my Garmin GPS with that same, exact brand of battery in it. The Garmin reads “GPS Low Battery” or words to that effect. The voltages of the 4 batteries in it are:
1.511
1.498
1.475
1.500
I’ve been doing this for more than 25 years. I (almost) always test my batteries before I throw them out (if I’m at home). I can’t draw a curve, I haven’t done double-blind testing, I haven’t researched it since Ronald Reagan was in office, but I am able to tell, marginally approximately, how much juice is in the AA batteries I have around the house. As in, “good, bad, ugly” tell. For example, when the GPS reads “half power remaining”, the voltages tend to run about 1.52-1.55V.
What baffles me more is that no one else is seeing the same phenomenon. It could be experimental error, selection bias, or something else. I agree that no-load testing of voltage ought not to give that much info, and yet, I seem to do pretty well by it.
**Una **- I think you and I were answering two different questions. Your experience with a single device and more or less homogenous battery brands, or at least technologies, and your good/bad/ugly qualification of them is ameanble to a simple voltage test. As you’ve so amply demonstrated. After you did the sampling to decide where your good/bad and bad/ugly cutoffs were.
Another thing to note from the graphs in my link is that for high-draw devices, the voltage/AH-remaining curve is particularly steep, implying relatively small error bars in the time axis for any particular voltage reading. Since your GPS “eats batteries”, it fits that sort of curve.
I was interpretting the OP as asking “Can I use a voltage measurement on an otherwise unfamiliar battery to acccurately (+/-10%?) determine the remaining useful life to power an arbitrary device?”
Different questions, different answers.
I understand better now, thank you.