Does anyone have any advice as to why a Zigbee door or window sensor won’t trigger all the time and is there a way to fix it? I’ve tried two different brands and they work fine when testing, but after a couple of hours I’ll go try it and it won’t trigger. I’ll then notice this and go back and open say, a door, and it triggers fine. Running on a Windows computer with Virtual Machine. Frustrating.
That sounds indeed rough. 
It might be a weak signal which is completely shielded 
ones a human (mainly consist of liquids) is doing the test to trigger and standing (more or less) in line-of-sight between sensor and receiver.
This can be expect for all low(er) power architectures and while 433mhz reed contacts for example can theoretically overcome longer distances than the zigbee equivalent ones they suffer from the same limitations and the signal can be solely blocked by a human being 
No it does not. 433MHz suffers almost 10 times less water absorption losses than 2.4GHz:
What’s your LQI for one of the problematic devices? I’m with the above, sounds like a weak signal. When you test, are you testing the device in place or testing near your HA setup and then moving the sensor?
Same limitations as other low power architectures I should have said 
I make heavy use of 433MHz and 2.4GHz sensors and while I’m able to cover the reception of 433Mhz reed sensors with my body I’m not able to disrupt the communication of any of my 2.4GHZ (esp, wifi) devices
That’s obviously because my 2.4GHz devices are not in the low power (
) category but using full fledged (and very resistant) 802.11b/g/n 
I’m testing with the sensor actually installed. I have 4 door sensors. The closest one to the server indicates an LQI of 255 and RSSI of -72. The one further from the server has an LQI of 255 (must be manufacturer based linkind though aqara did the same thing) and RSSI of -53 (which is weird that it’s better given that it’s further from the hub).
That’s because your 433MHz devices are crap then. None of my 433MHz, nor my 868MHz zwave devices stop functioning just because a human body is infront of them. That would be a bit ridiculous, especially for motion sensors…
It doesn’t stop functioning because a human body is in front of it. It just doesn’t work every time. Like I said, I can go to the door and open and close it in a testing environment and the sensor is tripped. But for some reason, if some time goes by, someone opens the door it won’t always trigger it. It’s almost like a state is not being checked or it’s going to sleep.
What do you that it works again? Reset the device? Take out the battery? 
Also have you tried testing the sensors closer to a receiver or on a different location to rule out possible interferences? 
Maybe also tell us the brands and models so people can tell about their mileage…
No reset necessary. Just open HA on my phone and look at the state. Keep opening and closing and it eventually triggers.
I tried Linkind and Aqara. Same issue with both. I’m going to try a Zwave sensor to see if there’s any difference.
I guess we disagree on this. I actually did a research before to find the best device for my needs. At the time (2019) there was no other device available which offered a battery lifetime in the same ball park. Without taking a battery self discharge into account @1technophile calculated a estimated runtime of 11.439 days based on his measurements. 
My units entering the fifth year of service now (around 1800 days) and still going strong with the (for my convenience) included battery
Regarding the downside of this long distance runner - the low range - that wasn’t any concern for me. As I have a minimum of one esphome node deployed per room I just added the 433MHz receivers (superheterodyne ones for ~$2) on the right spots and called it a day (that was five years ago) 
Next time maybe think twice before claiming other people devices (that you don’t even know) are crap
They’ll use a different controller, protocol stack, driver and frequency. So you would be changing all parameters at once. There’s a good chance they will not react the same way, unless the problem is with your underlying hardware (or the VM in this case).
The symptoms you described could be anything, from weak signal over software issues (you didn’t mention which Zigbee software stack you are using) to a problem with the coordinator (you didn’t mention which one you use), the VM usb passthrough, etc.
If they stop working when your body shields them, then they are crap. There’s no arguing around this.
Thank you for the responses. You’ve been at least trying to help me. I’m using ZHA integration with the GoControl Nortek hub. I have about 15 Enbrighten zigbee switches and about 10 plugs and all of them work fine.
Ok thanks for the info. I don’t use Zigbee myself (I’m on zwave), so I am not too familiar with the specifics, but the fact that your other devices work fine rules out VM and coordinator issues.
The next step would be to rule out RF signal issues and interference. I assume that with hub you mean the HUSBZB ? Are you using a USB extension cable to connect it to your server ? If not, add one. The USB controller on the mainboard can create a lot of interference.
If that doesn’t help, try the problematic device while it is physically close to the stick, line of sight. If it still causes problems, it’s not an RF isssue.
I have read a lot about issues with ZHA and recent HA versions lately, it seems that people experience a lot of weird effects. I don’t know if that’s related, but it could be another thing to consider.
Take a look at my Zigbee visualization. The highlighted devices are 4 door sensors and 2 water leak sensors. Is this why I may be having a reporting issue and, if so, what to do?
Just to close the loop on this thread, I purchased a third sensor and it works flawlessly. The
Aeotec SmartThings Multipurpose Sensor - Zigbee - Door and Window Sensor, Temperature Sensor.
It’s a little pricier than the other two I tried, but not only are these sensors always being appropriately triggered, you’re allowed a whole inch between the sensor and the magnet. Perfect for doors with molding or that aren’t on the same plane.
1 Like
That’s good! Questions still stands why your last two purchases are failing at the same location with the same symptoms 
Maybe they actually were RF interference and your newest device performs better in that terms?
Aren’t there any options to activate some debug logging on the devices or coordinator to get an idea what’s failing? I’m used to just open the (ota) logs 
of my esphome nodes if something needs insights. 
On the other hand did you ever check if the two (failing) devices perform normal/good on other locations (specially close to a receiver)? 
Sorry I need to tell you this but the air is a shared medium. That means it’s always possible that additional obstacles in the line-of-sight can degrade the link quality - even up to the point that a connection is lost 
To make a simple test you can just take one (or two) of your superior devices and have them placed on the edge were they still catch a signal. Now you can try to place different obstacles between sender and receiver and monitor the link/signal quality. Does it stay the same? Does it in/de-crease? You tell us!
As another note for @ecchodun: Are your glass doors/windows coated? While glass itself (even double and tripple layer) isn’t that big problem some metal coating applied on glass (often branded as “heat shield” 
etc.) on the other hand can interfere a lot with RF. For my 433 reed sensors it was necessary to have the receivers within 5.5m (they can reach up to 13m line-of-sight without obstacles) to have 100% receive rate of the open/close actions 
Let’s consider an average transmitter (+10dBm, even though actually good ones go higher, depending on legal regulations for the 433MHz band) and an average receiver at -80dBm (good receivers usually are much more sensitive than that). Line of sight, dry air, with a typical link margin of 15dBm, which is plenty for OOK modulation as normally used by cheap 433MHz devices. No antenna gain, since the antennas are typically omnidirectional.
free space path loss (FSPL) is thus
FSPL(dB) = 10 + (-80) - 15 = 75
So, plug that into free space path loss equation (*):
FSPL(dB) = 20log10(d) 20log10(f) + 20log10(4π / c)
f being 433MHz,
d = 10((FSPL(dB) - 32.45 - 20log(433)) / 20) = 310 meters
In practice that figure will be lower due to air never being entirely dry, due to losses on the PCB traces carrying the RF signal and due to enclosures. Let’s half it for good measure, around 150 meters. Coincidently that’s pretty much what I get with my what you called “superior devices” on line of sight conditions (actually entirely non-superior normal Oregon Scientific thermometers and some DIY with good transmitters and an RFXCOM).
In conclusion, your 433MHz devices are crap. You’re welcome.
(* more info for people interested in the math behind RF distance estimates and link budgets, see this paper here).
You should by now already be aware that I don’t use a average transmitter but one that almost has no power draw (the self discharge of the battery is probably higher than the energy this device uses 
). Obviously this thing doesn’t blast with full power but conserves energy as much as possible
Power consumption estimation
When sending signal (open or close) 12mA during 1,6s
Sleep current 0.47uA continuously
The findings from @1technophile are well documented, fell free to widen your horizont a little bit @HeyImAlex
In synthesis the main criterias that differentiate both sensors are the autonomy and the range, in terms of range the GS-WDS07 is a lot more powerfull compared to the DIGOO. At the opposite it results in an impressive autonomy for the DIGOO.
Your choice will depend to the distance between your windows/doors and the gateway, if you live on a small flat for example you will not need the power of the GS-WDS07 and could take the DIGOO.
For a house or a big appartment I will advise the GS-WDS07.
While it might be hard for you to accept this devices are great for many use cases. 
Beside If I would have invested in another device (for example one of your “superior” ones ) I would have probably spend more on batteries the last five years than I spend on this device with battery (still going strong after 5 years with the included battery which were around $3 shipped - reed sensor & battery 
)
virtually running on free energy 
BTW.: With LORA on 433 it’s possible to achieve coverage in the kilometer range 
Oh I think I’m good you know
55uA draw, 5 year battery life, made from parts I had lying around in one evening, and guess what, >100m line of sight range. No wifi though, sorry 'bout that It’s not hard. The transmitter isn’t even good. It’s just that the type of devices you linked to above are really really bad.
Yep. Mostly because the rx sensitivity is very high and the required SNR / link margin very low, due to the very low bandwidth and spread spectrum. But you don’t even have to go LoRa for that. You can get >1km range with 433MHz narrow spectrum without a problem on higher bandwidth links, still with very little current draw. You just need to use the right components.