My devices seem to have batteries that are never in the <20% bracket for any time, they're all good then dead. Any general tips?
I use device watchdog but never get warned on low batteries, I just get told inactive after x days.
Battery reporting is hit or miss at best. Some devices report voltage, some report %age. Few are the same and even less are accurate.
I use Device Activity Checker. I've set it for 24 hours so any device that has not responded in 24 hours will raise a flag. In my case it is a SMS message.
Lithium, alkaline, or rechargeable? Lithium will always have a steep drop like that near the end - just the way they discharge.
1 Like
Set the alert at 40%, or go by activity timeout alone. I have a dozen different devices that are hit or miss if they ever report < 33% before they are dead. Usually I'll get something around 33% or so (again depending on the device), but anything past that is not reliable.
2 Likes
Hue devices only seem to report 100%
I agree with the comments that battery reporting (at least in Z-wave devices) is notoriously unreliable, and this is not confined to Hubitat. And, I find that the same model of device can have the same settings and batteries across multiple devices, yet use batteries at dramatically different rates despite similar patterns of use. This is probably the single biggest effect on reliability of my system, as some devices show near 100%, then just die with no warning, even with alkalines vs. lithium (I expect that in lithium ion batteries, but formerly thought alkalines would at least report declining battery % for a few days/weeks. I no longer think that's entirely reliable). Tracking device activity seems to be the only solution to catch a dead device within a day or so.
Most of my sensors use some type of button battery. Although there are alkaline and silver oxide button batteries, most of the button cells used in sensors are lithium batteries providing 3 volts. Because of the chemistry of the lithium batteries, the voltage stays very close to 3 Volts until the battery is nearly dead. Then the voltage plummets. This behavior makes it difficult to predict the end of life. If the device is of critical importance, you either need to replace batteries on a regular schedule or replace them as soon as the voltage starts to drop.
If your devices are powered by standard alkaline AA or AAA cells then the end of life prediction will be far more accurate. The voltage of alkaline cells drops more gradually making it easier to predict the useful life.
1 Like
Good point.
Perhaps a more useful approach would be a method of simply counting the number of days since the last battery replacement in each device, and an averaged value of the number of days between the last few changes of battery in that same device.
There are "created", "last update", and "last activity" date/time stamps for each device, so one would think that a user-settable parameter could be tracked where, as a battery is replaced, the user (manually) clicks "I just replaced the battery" on a device, and the system then tracks the elapsed time.
When the elapsed time approaches the "last known battery interval", a warning is generated to check batteries.
If data were shared, one could also pre-load the "battery interval" for a new device with an average of everyone else's prior findings on battery life, kept on a public list.
The tracking of voltage is presupposed to be a waste of time for reasons stated clearly by rwclements228. I agree - button cells, RC2s, AA, AAA, will each present different "voltage death spirals".
On the other hand, fire departments try to convince everyone to replace batteries in smoke alarms "when you set the clocks back" each fall. While this blind annual replacement effort prompted the creation of "10-Year Smoke Alarm Batteries", and even entire "10-year sealed (disposable!) smoke alarms", one could take a similar strict-calendar based approach to all the batteries in all devices in any one configuration.
1 Like
One other point:
I've had the same issue with CR2032's and CR2450's, and I've decided that they are just not worth my time. Wherever possible, I've replaced those devices (primarily Zigbee motions sensors, humidity sensors, etc.) with Zigbee devices powered by a CR2. Those batteries seem to last longer.
The CR2 batteries last longer because they are bigger than the CR2032 and CR2450 batteries, but the chemistry is similar. . I have some motion detectors powered by CR 2477 batteries which are bigger than the CR 2450 batteries. Thus, they last longer.
In the CR2477 name, the CR refers to the battery type. It means it is a lithium battery that also contains chromium. However, depending upon the manufacturer, other initials may be used. The 24 represents the diameter of the button (24 mm). The 77 refers to the thickness of the button (7.7 mm). The CR2477 is the largest button battery having a capacity of 800-1000 mAh.
The CR2 battery has a different shape, but the capacity is similar to the CR2477 batteries. Thus, some of the higher drain sensors use CR2 while others use CR2477. The battery life will be similar with either type.
1 Like
Every one has to find their comfort zone when it comes to replacing batteries. I am more diligent with water sensors than my information only temperature sensors.
I do log every battery change in Excel with the date battery type etc. However with batteries lasting 18 months or more it takes a long time to collect meaningful data. That is if there is meaningful data. Its even worse in that I cannot make any predictions on the performance of the battery that was received with the device (no history on freshness etc).
Another confounding factor is the mfg battery data is always at some substantial current (at least compared to the ultra low sensor requirements).
Just for information.
I recently changed out all my Sonoff motion sensors as I was getting false motion detection.
I bought some Tuya sensors which work well but on all of them (10) I've had to replace the batteries. From memory they were RTF CR2450 batteries. They all failed within a month. I have replaced the batteries with Panasonic batteries.
Either the sensors have been sitting about for some time, there was no 'film' to remove so the batteries were live in the sensors, or the RTF batteries are cheap and have a short battery life.
I never received any low battery notification on any of them. I think the lowest was around 70%.
Fortunately 'Device Watchdog' notified me of device inactivity and I could see that the devices were no longer working.
Just thought I would pass on this useless info. 
1 Like
That is far from useless information.
I keep a stock of Panasonic batteries in various sizes. I order them ten at a time.
1 Like
Hey bobbles
Can you link to the Tuya sensors you used and do they work well with the HE. which drivers do you use.
Thanks
I am also following the thread for Tuya info. It is my understanding that Tuya is now part of the Zigbee Alliance and is coming out with Zigbee 3.0 compatible devices. I think most of the earlier devices were WiFI and required flashing with Tasmota. I am waiting for the Zigbee certified devices as they will a lot more user friendly.
Hi @greglsh
These are the ones.
I didn't pay the amount shown though. Think it was £12 - £13 each.
Driver being used is the inbuilt "Generic Zigbee Motion Sensor (no temp)".
I also got one of these to try.
I quite like these but they are from China so take a while to turn up.
They are small and are mounted with double sided tape on the back so can be fitted on walls, ceilings etc.
If I need to order more motion sensors I think I will go with these. They are zigbee 3.0.
It took some messing about to pair it and I'm using a custom driver which works well with it.
Do you have a good example of how to do this? Something in RM or an app? Am I missing a setting for this somewhere? Thank you!
