Who knows, maybe changing so many of the default config settings triggered an issue/bug in the firmware. As I've been fighting with a few zwave 800 firmware issues on a sensor I was working on, I know how stupidly (and largely unnecessarily, in my opinion) complicated zwave firmware can be.
I'll stick it on my power profiler if I get time to see if the device is actually going fully to sleep or not.
While I'm going to replace them either way, it would still be nice to figure out why they are dying so fast. I did sniff them for a few days, and I'm getting a normal number of messages out of them (<50 day, 15-20 more common), that's why I'm suspecting idle power usage.
Same issues looking to see if someone else was having this problem too. I've tried different batteries and quality batteries and now I'm 4 batteries down in the first few months. That's beyond what I'm willing to replace. Like to know why others don't seem to have the same issues.
Out of curiosity, are you replacing the batteries because of a low battery warning/% report, or because the device has stopped reporting?
I ask because I have tons of these in my set up (along with some 42s and 43s that came out about the same time). The battery reporting is garbage/borderline worthless on all of them (bounces around ridiculously), but I have several that have been at 1% battery for months and they continue to report events fine.
So while I'm not making excuses for Zooz, since the battery reporting is clearly forked up (yes, I'm on the latest FW for all of them), I have found that if I ignore the percentages, I'm getting pretty decent battery life out of them.
In order to actually know if the battery is dead, I'm using the Device Activity Check app that's available in HPM.
This is working for others so something has to be amiss. It's got great signal and is connected directly to the hub. But I'm 4 batteries in now in less than a few months. Just not worth it...and because of the poor battery reporting I have no idea what's actually going on with the battery...until it just stops responding.
Due to the battery discharge curve of lithium batteries, predicting battery life is difficult at best and futile at worst. Lithium batteries maintain a voltage close to nominal throughout most of their useful life. When they near EOL, the voltage plummets and the device stops working, seemingly overnight. That discharge behavior is quite different than alkaline batteries whose voltage declines slowly over the life of the battery, making predictions of battery life reasonably accurate.
Thus, if the device is powered by lithium batteries, as many are, never trust the battery life indicator.
The discharge curve depends on power draw. Since most HA devices draw minimal power, the top curve on this chart is appropriate. The chart is for the common CR2032 lithium, magnesium battery.
I've seen this rationale posted before, but I'm a bit skeptical that this is the problem with the Zooz devices. Wouldn't that mean that they'd report 100% for a long time and then drop off suddenly right before they die vs the behavior that we see - 20% today, 75% tomorrow, 100% the next day, then 1% for literally months despite still working?
It would actually be much more useful if they reported the way you said, because then you'd actually have a real feel for when it's about to die.
I am only seeing this on devices with the CR2032 button batteries. The devices with CR123 batteries are much better. Seems like if a devices gets some heavy use the voltage must drop on the CR2032 and then if it sits for a while it recovers. That is why you see it drop then come back up. That's what I have observed anyway. The battery percent is driven by output voltage. I think the poor reporting is just a by product of the choice of batteries.
The curves show that the voltage is very much dependent on power draw. Thus, in devices that draw more power, the battery monitoring will be more inconsistent.
I’m seeing periodic erroneous readings (5500C) which ruin my charts. Is there a custom driver available, where erroneous readings could be ignored?
Thanks!
There are two custom drivers, you could add that feature to them.
In the past people have found that pairing it with S2 security solves the issue of the corrupted reports. I had it on my todo list for my driver but seems like S2 has solved the issue for everyone else with that issue thus far.
FWIW, I've been testing with a pair of the current ZSE44 (800 series) for about a month. At least one of the units has been mesh with no security the entire time. So far, I haven't seen any spurious temperature or humidity readings.
I've seen issues with battery reporting and lifetime, particularly with LR, but no temp/humidity reporting issues.
Yes, by the current firmware version as already pointed out.
Usually anything v1.x takes 1.x firmware, and 2.x takes 2.x firmware updates, etc... Unless otherwise noted by Zooz on their firmware download page. The major version number usually only changes if here is a hardware rev change.
