My iris iL07_1 motion sensor battery died. Its a Panasonic CR123A. I measured the battery before changing it and found it to be ~3.076 -3.078. Low for a battery but I felt the unit should work at 3V.
Humm..
So I put my scope on the battery set to trigger at 2.99 volts. Waited a while and no trigger. Then I created a motion to trigger a message. The battery dropped from 3.0V to pretty much zero volts in 1/2 a millisecond.
So now I know, this lithium maintains a voltage but the internal resistance goes way up, enough to make the device not work.
That is exactly the reason why lithium batteries are so good... they maintain voltage almost up to the end of its charge... and that is great.
Unfortunately to measure its internal charge is quite complicated for that reason... some professional instruments are able to do it very accurately but introducing some specific parameters of each model and brand.
Because it isn't the battery that reports charge level, it's the charge circuit. Once the battery cannot accept anymore charge from the charger, the charger stops charging and reports the battery is 100% charged.
A very worn out battery will therefore stop charging pretty much instantly, causing the charger to report full charge even though the device will immediately die once the charger is disconnected.
There's technically some more to the story, but that's the high level overview version of it.
@rob9 is correct. This wasn't a load test, nor an attempt to determine what the remaining charge level was. It was a test to see why the battery from a non reporting iris iL07 that was measured at 3.078 was not powering the iris adequately.
The test was a simple " monitor the voltage while subjecting the iris to motion".
What I found is at least part of the answer as to why a battery will one day read 100% and a few days later be "dead".
I still have the subject battery, If I get time I may measure the internal resistance.
I just measured the open circuit voltage. The voltage is 3.104 Volts after being under no load for about a day.
not sure if this is true, but i read that adding a simple 100ohm resistor in the path will give a better indication of battery charge, if you only have a multimeter and a resistor and not a "fancy pants" oscilloscope
From the iris I tested, the major current only occurs for fractions of a second, would be difficult to read with a multi-meter. Not sure if a "peak hold" function would matter.
I would guess the device may be able to sync a voltage measurement with transmit action. However I don't think the mfg has battery reporting as a high priority.
Thanks, what I meant was introducing a resistor in the circuit when evaluating the charge state of the battery, outside of the device. Like check the voltage with a resistor in the path of the meter leads, versus just connecting the meter leads to battery without a resistor
I doubt that would provide a helpful measurement. If you assume your multimeter (ideally) draws zero current, there will be no voltage drop across the resistor.
What may work is to periodically connect a more substantial load on the battery and read the battery with the load connected. It could be connected for a very short time, however if performed periodically will likely significantly reduce the battery life.
Keep in mind, I've already extrapolated a lot from my one test but I have no data that supports any particular behavior.
Another factor is the self reporting aspect. The battery is expected to have enough charge to get a final last gasp report to the hub, right? It is highly unlikely to occur. Battery reporting messages are typically set to report once an hour or when X change. If the last microvolt of energy occurred 30 mins before the report, then the final report would 1) be inaccurate, and 2) probably show an adequate remainder. I look at battery reports and if they show anything less than 85%, I assume they are dead by now, because I don't look at battery reports every day, all day By the time I get around to looking, the most recent sub-85% report is days ago. It's dead by now
By the time I get around to looking, the most recent sub-85% report is days ago. It's dead by now 
