#Zigbee utilization message in logs

1 messages ยท Page 1 of 1 (latest)

paper tide
#

With the channel for Zigbee stuff being closed, here is a place for talking Zigbee that doesn't really require its own topic.

I learnt so much more from just reading generic chat, it would be a shame if this were lost to the enforced forum style where generic chatting is discouraged.

#

First bit of knowledge seeking - I see Zigbee channel 26 utilization is 75.69%! in my logs on startup.... Is this a message that will repeat?

Or is this something that is somewhat expected on HA startup as everything comes alive?

spiral elk
#

Need more info. what cordinator are you using? where are you seeing this message?

paper tide
#

The message is in the HA Logs:

2024-09-16 13:49:01.785 WARNING (MainThread) [zigpy.application] Zigbee channel 26 utilization is 75.69%!

Using a Sonoff_Zigbee_3.0_USB_Dongle_Plus with ZHA... (I forgot there's also the MQTT Zigbee integraion too :D)

spiral elk
#

Yea don't think I ever seen that when I had my cordinator which is not the same as yours connected via usb to HA using ZHA. Hopefully someone who is more in your boat my have better answers but I suspect you are probably right with it having to do with ZHA collecting all those routers/end points during the start

#

I would just ignore log entry if everything is working correctly

paper tide
#

Yeah - I did a lot of tinkering a while back to the RF space specifically - like moving APs to channels 1 & 6, and moving the Zigbee stuff to channel 26... That made a big difference...

spiral elk
#

For me cordinator location was the biggest improvement. moving from ZHA to Z2M seemed to help a lot too. but more on the stable aspect

paper tide
#

Interesting - I moved away from running an MQTT server when I moved my own code for my ESP(8266|32) kit from MQTT to ESPHome....

spiral elk
#

I dunno what it is but mqtt just seems more stable over other forms for me

paper tide
#

I ponder if its more to do with the whole concept of quirks in ZHA stuff? Doesn't MQTT just punt out the raw message in JSON as a published message?

spiral elk
#

yea

paper tide
#

I have to be honest, I only ever tried ZHA and haven't looked too much into the Zigbee2MQTT bits

spiral elk
#

you should. the gui for z2m is so nice lol

#

you get a lot more info and functionality with z2m than zha

#

the last seen in z2m is helpful too

#

I would post a picture of it but I cannot, they don't have pictures enabled in this lol

paper tide
#

Dammit - you're going to make me pull apart my perfectly working network, aren't you ๐Ÿ˜„ ๐Ÿ˜„ ๐Ÿ˜„

spiral elk
#

if what you have is working. then there is no point. but yes, if something were to come up. I would suggest giving it a try

paper tide
#

yeah - but I'm gunna want to try it ๐Ÿ˜„

spiral elk
#

lol

#

make sure your dongle is supported first and foremost

#

I'm pretty sure it is, but make sure

paper tide
#

yeah - its the stock CP1652(?) I think

paper tide
#

I might just break out the HackRF when I get home and see if I can see an increase in noise hahahah

#

the zigbee signal is pretty well defined when you tune to the right channel

iron iris
queen linden
# paper tide First bit of knowledge seeking - I see `Zigbee channel 26 utilization is 75.69%!...

While I understand people are missing the chat feature of regular discussion, this topic very easily could be its own thread - it's specific in its ask. The advantage to this is that then people can focus specifically on what you are talking about and you don't have to re-reference conversations previously when others are talking about different things.

I know people will disagree with this, however in order for forums to work for support, we cannot have general or catchall discussion channels like this. Posting up a question where you are looking to have a discussion is totally fine - you are looking for support with your ask. The point of switching to the forums was to remove this loose chat aspect of the support channels, though.

Regular ole chit chat can happen in #the-water-cooler and showing off projects or content you created has a home in #1284960456880029707. I'm going to change the title of your post here (rather than locking it down since you do have a clear request here) and ask that we respect this - please do not create another general chat thread and stick to creating a thread that is specific to what your ask is no matter the size of it. ๐Ÿ™
-# Also please review the #rules if you haven't already.

#

Zigbee utilization message in logs

paper tide
#

@queen linden with due respect, if I wanted to use forums, I'd use the actual forums.... not discord forums vs discourse forums.... I feel converting a chat medium into yet another forums is a bad move that will stifle discussion - which is what discord is all about... Alas, that's an off topic (but on topic based on your message), and time will tell if it just kills the usefulness of discord.

#

but its also the exact reason I don't use the forums - as I'm not going to spend hours scrolling through the forum to maybe come across some nugget of wisdom - just like I'm not going to trawl through hundreds of discord forums for a random nugget.... But randomly just catching up on chat is a completely different experience.

queen linden
#

No worries, I get where youโ€™re coming from. Since I posted this, things have evolved.

In #the-water-cooler, there was a discussion about the โ€œhardware groupsโ€, as Iโ€™ll call yโ€™all. Iโ€™ve heard people and what has been heard is that my communication of what the new community channels mean was unintentionally alienating groups like this. #the-water-cooler message

Hereโ€™s where I am on this: #1285154411114532935 message I have the birthday stream to run tomorrow, though, so I canโ€™t promise I can create this before Wednesday. Iโ€™ll happily chat more with you about this specifically over there if youโ€™d like. ๐Ÿ˜Š

paper tide
#

All good - any / all changes will piss someone off - its the nature of human life... Just wanted to add in my 2c... I'm just old enough that I still use IRC on a daily basis - and its some of the best content I ever look at hahahaha

paper tide
queen linden
paper tide
#

Sounds like my younger days ๐Ÿ˜„ ๐Ÿ˜„ ๐Ÿ˜„

queen linden
#

Iโ€™m at the young end of users. ๐Ÿคฃ

paper tide
#

the early 90s were wild online.

#

I used to moderate the biggest IRC room in Melbourne, Australia - and it was an experience. but it turned toxic a.f. but I remember the era well

#

what's funny is that from that one thing - I can make a rough guess that your parents are probably about their 50s, and you around 30s?

#

peoples knowledge of IRC is like carbon dating.

queen linden
#

HAHAHA yes ish. Youโ€™re about a decade under, Iโ€™ll be 40 next year.

paper tide
#

the other good carbon dating thing is if people know what NNTP is

#

before that, its fidonet.

queen linden
#

Youโ€™ve broke my brain with those ones. ๐Ÿคฃ

#

I bet my dad and step dad would know though. lul

paper tide
#

NNTP is now known as newsgroups..... FidoNet was before email existed as we know it today... The first real wide-scale interconnected message handling with message routing...

paper tide
#

Back on topic though - Zigbee2MQTT does seem to be much more informative than ZHA.....

#

I'm guessing its just extra time & evolution.

paper tide
#

Bah - no posting of pictures here lol

oblique steppe
paper tide
#

Yeah - Z2M really seems nice.... I just used ZHA when I first started cos I already removed MQTT

#

but its kinda worthwhile for the functionality

queen linden
queen linden
#

You should be good to post images now!

oblique steppe
paper tide
#

oh - nice ๐Ÿ™‚

#

to both ๐Ÿ˜„

oblique steppe
#

Mostly Iโ€™m excited for ZHA to be decoupled from HA. That is pretty much the main reason I use Z2M.

paper tide
queen linden
#

There was an issue with Discord just now, but also I didn't uh sync the category.

#

One moment, I am just gonna give the role it.

paper tide
#

The more failings I had for ZHA was:

  1. The amount of missing quirks.... So far, every device in Z2M has 'just worked' - in ZHA, I had to make custom quirks...
  2. The 'network maps' that Z2M generate are far superior to the ZHA layout
  3. It's a lot of information presented in Z2M which is largely hidden in ZHA.... Being able to do custom bindings is just win... Even though my devices don't really support it ๐Ÿ˜ฆ
queen linden
#

Alright....one more try?

paper tide
#

Yarp ๐Ÿ˜„

#

That's a Zigbee data network on Channel 26

queen linden
paper tide
#

the sweep I interestingly tracked down to a wireless headset doing seeks for the headset when its powered off

#

but that's a community behavour thing - not a technical problem ๐Ÿ˜„

queen linden
#

omg wth is this. lul

#

I mean, I created #memes for a reason. lul

paper tide
#

didn't even see that ๐Ÿ˜„

queen linden
paper tide
#

I don't suppose anyone has done groups with Z2M?

#

I'm trying to figure out how to bind two light switch buttons together from different devices.... but I seem to be having trouble - and I'm not sure if its device or me.

#

I tried sending this:

#

but get: [2024-09-18 02:50:47] error: z2m: No converter available for 'off_state' ("last_member_state")

#

It also mentions stuff in configuration.yaml - but I don't have any options at all in there for Z2M at all

#

facepalms - There's a z2m configuration.yaml too..... That was my confusion.

#

but the group still doesn't work ๐Ÿ˜

iron iris
paper tide
#

in Z2M - I can see the group events:

[2024-09-18 03:02:30] info:     z2m:mqtt: MQTT publish: topic 'zigbee2mqtt/living_room_light_group', payload '{"state":"OFF"}'
[2024-09-18 03:02:30] info:     z2m:mqtt: MQTT publish: topic 'zigbee2mqtt/living_room_light_switch', payload {"backlight_mode":"inverted","last_seen":"2024-09-18T03:02:30+10:00","linkquality":65,"power_on_behavior":"off","power_on_behavior_left":"off","state_left":"OFF","state_right":"OFF"}'
[2024-09-18 03:02:28] info:     z2m:mqtt: MQTT publish: topic 'zigbee2mqtt/living_room_light_group', payload '{"state":"ON"}'
[2024-09-18 03:02:28] info:     z2m:mqtt: MQTT publish: topic 'zigbee2mqtt/living_room_light_switch', payload {"backlight_mode":"inverted","last_seen":"2024-09-18T03:02:28+10:00","linkquality":65,"power_on_behavior":"off","power_on_behavior_left":"off","state_left":"OFF","state_right":"ON"}'
#

but it just doesn't actually trigger anything changing

oblique steppe
#

You don't see anything changing in HA? Or anything changing in Z2M?

paper tide
#

I'm not exactly sure......

#

I see the MQTT messages go out.

#

so the HA -> Zigbee2MQTT must be working

#

but the actual light doens't change.

#

I have a feeling that the binding on the devices isn't working

#

if I understand it, the groups happen at the zigbee level - so the message gets broadcast out, and the devices have to figure out if its for them or not

#

iirc, the Coordinator doesn't get involved in that message handling - so even though it looks like I have the bindings for an endpoint to a group, if the firmware doesn't work properly, then its just SOL...

paper tide
#

in other news, this is what my mesh has settled on at the moment:

#

seems pretty reasonable.

lost pine
#

wait until you hit 100's of devices

paper tide
#

I can't see myself ever needing that lol

#

Although - I'm pondering if I can redo the firmware for these lightswitches - as they really should be able to act as routers as well

oblique steppe
oblique steppe
paper tide
#

They do have neutral wires - but they're marked as "optional"

#

tracing out the PCB, I can see it does look to be connected to some components.

#

but I also did have to put some caps on some circuits to stop LED downlights from flickering

paper tide
#

Does the ZHA integration allow the coordinator TX power level to be changed? Or is it firmware default only?

ancient yarrow
#

efr32, I think similarly, but can only go lower than what is set in fw.

paper tide
#

I forget this one every time ๐Ÿ˜„

#

usb-ITead_Sonoff_Zigbee_3.0_USB_Dongle_Plus

#

Which I think is a cc2652p?

#

but between -22 and 19? wtf?

#
#

it has the specs of the -P model as: RF Transmit Output Power 9dBm (firmware hardcoded), Max: 20dBm

ancient yarrow
#

the -P is cc2652, the z-stack fw it runs default is 9 it can be bumped to 20 but it's not a great idea

paper tide
#

I did notice in Z2M, you can just set a number between 0 and 20

ancient yarrow
#

the exact config esacpes me right now, there maybe an example in the issues on that repo, or in the old zigbee channel

paper tide
#

that seems to make sense - but ZHA doesn't have any of that stuff easily exposed, so I'm kinda curious on the different approaches.

#

its nothing critical - but more me just trying to get knowledge ๐Ÿ˜„

ancient yarrow
#

puddly and dmulcahey are the main zha devs so they would know for sure.

paper tide
#

but its not even clear to me what power level that uses by default

ancient yarrow
#

9

paper tide
#

See, the hardware page on ZStack says:

ancient yarrow
#

that's the max power. I'm fairly confident 9 is the default.

paper tide
#

it could well be - I'm just struggling to find an authorative souce hahahaha

#

within Z2M, it says: Transmit power of adapter, only available for Z-Stack (CC253*/CC2652/CC1352) adapters, CC2652 = 5dbm, CC1352 max is = 20dbm (5dbm default)

ancient yarrow
#

okay then I'm thinking of the router, FW as I just saw KoenKK reference it as 9.

paper tide
#

I'm going to guess that a CC2652 is the same as a CC2652P - which would mean a 5dbm default? but not sure if that's firmware, or a default setting in Z2M

#

that's the issue when looking for something concrete - there's the firmware default, then whatever framework default, and then any override setting as well ๐Ÿ˜„

ancient yarrow
#

2652P uses same fw as the 1352 launchpad which also has the Power Amplifier. the P is short for PA - Power Amplifier

paper tide
#

I mean - putting in a value of 10 doesn't even tell you what that means..... A register? Or a dBm level, or what? lol

ancient yarrow
#

its dB

paper tide
#

ie my cisco APs have power levels of between 0-7 - and what they mean differs per channel ๐Ÿ˜„

ancient yarrow
#

raising it doesn't make a lot of sense, an end devices from far away could hear the strong. signal and join to coordinator but a nearby router is a wiser choice usually, as the coordinator may not be able to receive the signal from the end device well.

paper tide
#

true - but knowing what setting is good depends on you knowing what the variables actually are....

#

I mean, good luck finding out what your devices are using as a TX power too

ancient yarrow
#

5 is pretty standard I think.

paper tide
#

5mW? dBm? 5dBm is only 3mW - and I'd expect them to be higher than that

ancient yarrow
#

dBm

paper tide
#

I'd expect 5mW would be sane

ancient yarrow
#

some run off coin cell batteries for years

paper tide
#

well, I've yet to have my doorbell for a year and already had to replace the battery once lol

#

it lasted ~3 months

ancient yarrow
#

I changed it to some

paper tide
#

thankfully, I have a ton of CR2032's around ๐Ÿ˜„

#

anyway - reason I'm on this rabbithole is that I'm looking at the map in Z2M:

#

and some of what it shows, I wouldn't expect

#

like, the device office_light_switch seems to prefer direct to the cooridinator, when there's a router less than a metre away in the same room.

#

the kitchen_fridge_power_switch is on the opposide side of a wall as the coordinator, yet its signal is about as good as devices on the other side of the house

#

so I'm just digging around being a bit curious about some of the mesh decisions that its making

#

like: kitchen_light_switch and kitchen_power_switch are literally 1 metre apart. Yet rear_bedroom_light_switch is on the opposite side of the house

#

strange mesh decisions ๐Ÿ˜„

#

or at least non-obvious

#

maybe I did it wrong by adding in the routers afterwards.... maybe I should have added all the routers, then the devices when I built the network again

#

but in theory, it should figure that stuff out itself

ancient yarrow
#

most zigbee 3.0 devices do. as do a lot of earlier ones.. original aqara nope

paper tide
#

if only there was a way to rebuild the network without having to rename and do all the devices again hahahah

oblique steppe
#

You can. Just re-pair the devices in place.

paper tide
#

oh? that works?

#

hmmmmm - if the map is a live-look at things, then not much changed lol

#

oh - actually, looking closer, there are a lot of reassociations.....

#

I wish there was just a command that'd tell everything to rebuild the mesh lol

ancient yarrow
#

turn off your power for a few hours

paper tide
#

even if that was like a 'disassociate everything' - without deleting the HA / Z2M entities and then allowing them all to rejoin again.

#

I tried looking for a factory reset command via Zigbee but came up empty

ancient yarrow
#

I don't know how many devices will actually support it:

paper tide
#

huh

#

interestingly, sending that in Z2M like this seemed to reset the switch:

iron iris
#

One thing about TX power to keep in mind is that Zigbee is somewhat symmetrical. There's really no benefit when you crank it up to 20, only downsides

spice cradle
#

doesn't help if the device can hear the coordinator, if the coordinator can't hear the device. same is true of wifi access points too :)

paper tide
#

Same is true with RF in any sense..... I've done my stint of running radios at 120W output ๐Ÿ˜„

#

back in the day, used to be able to hear my radio across most of the state

#

but hearing someone back was always hit and miss.

#

although amatuer radio from an aircraft was always fun ๐Ÿ˜„

#

5W at altitude was amazing compared to 120w on the ground....

spiral elk
#

@paper tideHow you liking Z2M so far?

paper tide
#

I like the extra info.

spiral elk
#

yes, it is nice

#

the last seen i find useful

paper tide
#

being able to do bindings in the UI is nice - even though my hardware is buggy and they don't seem to work

spiral elk
#

your coordinator?

paper tide
#

nah - the light switches

spiral elk
#

oh

paper tide
#

essentially, trying to make a switch group on the zigbee side

#

the bindings seem to get set, but they just don't work

#

the way the map is drawn is MUCH better in Z2M

#

the process of permitting a join imho is much clearer as to what you're doing in Z2M over ZHA

#

ie it gives you a countdown, and the ability to only start the joining process on a single device - say a router.....

#

the actual operation of ZHA vs Z2M for the HA side of things - not much difference

spiral elk
#

zigbee is far from flawless.. pretty sure most devices are only working by means of a bandaid lol

paper tide
#

Yeah - I still have at least one device that 'just works' in Z2M that is weird in ZHA

#

that's the little contact strip / door detector thing that I use on my garage door

#

In ZHA, the device always shows as Closed - but you'll get events to say something has happened, but the status never moves from Closed

#

in Z2M it actually does the Open / Closed bit properly

spiral elk
#

yea, if you are planning to expand your network i'm sure you'll run across other quirks. aqara is one of the weirder ones for me being that the ones I have will not connect to anything other than the coordinator and the map always shows it floating off into space yet the thing continues to work as it should

paper tide
#

So when I created an automation to turn the light in there off when the garage door went to the 'closed' status, then that would trigger when the device went to Opened status - but the sensor never changed to Open

spiral elk
#

some of that could be network stability too

paper tide
#

so the light would turn off when you opened the door.

#

nah - it 100% works for events - just it doesn't know the difference between opened and closed - probably fixable with a quirk..... but I got lazy

spiral elk
#

when I placed my cordinator in the middle of my home (middle of my devices) things worked so much better than relying on scenarios such as bouncing between routers before hitting the cordinator

paper tide
#

I submitted a quirk to ZHA to make it work

#

now that I see how it works out of the box in Z2M - I'm not sure if my submitted quirk was good enough hahahaha

#

I'm pretty sure my quirk only covered one of the two modes that the devices functions in.....

#

ie you can triple press the button and it'll change mode..... one mode gives you on / off, the other mode gives you single / double / hold status messages

spiral elk
#

works pretty good

paper tide
#

given it works out of the box, there should be a way in ZHA to import quirks from Z2M - whatever form that would take

#

I use it as a doorbell - so I have to set it up in NodeRED as:

#

cos the 'state' in an action could be on, off, single, double, hold

#

I've never seen the brightness states

#

oh - and to be annoying - you'll get two events per press..... ON, then None

#

but I don't know if that's a HA or Z2M thing - or the device itself

spiral elk
#

Even though I have significatnly improved the reliablity of my zigbee network there are some things I still don't trust with it. I fall back to good ole 433MHz

paper tide
#

well, I moved in to a new house, so I just replaced every light switch with a zigbee one

#

annoyingly, like I mentioned, switch groups aren't really a thing hardware wise - and it adds latency having to do that via HA in an automation

#

some of the mesh decisions are just weird - like I have a lightswitch talking to a router at the opposite end of the house - even though the signal has to go past two other routers to get there.....

#

ie almost in a straight line

spiral elk
#

the map is not "accurate" in that regard.. it's most just an idea

#

like i said, i have one device that floats off into space connected to nothing and it works fine

paper tide
#

oh - I thought it was like a route table lol

#

ie I thought it was actually legit

spiral elk
#

it is and isn't

paper tide
#

๐Ÿ˜„

spiral elk
#

it's the best guess representation of the flow of data

paper tide
#

you know the one thing in IT that upset me more than just finding that out?

#

When mainboards are wired up to RAM sockets - there is no reason for them to lay out the data lines in the same way you'd expect - so a binary 10101010 might be stored in the RAM chip as 11110000 if it makes it easier to lay out the data lines on the mainboard

#

just because the order of the bits in memory don't matter compared to how the data is presented on the data lines.

#

that shattered my world view.

spiral elk
#

yea I don't deep dive like that lol

paper tide
#

I found this out when trying to trace some data lines to a RAM chip on a small PCB...... and it had me completely fucked as to how it was working..... cos the address lines were fine, but the data lines weren't D1:D1, D2:D2 etc etc

#

but something like D1:D2 D2:D8 D3 :D7 - like all over the place

#

but a RAM bit is a RAM bit - it doesn't matter where it is on the chip

#

I died a little inside as everything I'd known before was a lie.

spiral elk
#

lol

paper tide
#

I thought the MAP was fact ๐Ÿ™‚

#

now I know its an approximation on reality ๐Ÿ˜›

paper tide
#

damn - I feel that post in my soul

spiral elk
#

the maps are generally right.. but nothing is flawless

#

the map they put on that posts shows devices floating into space too lol

#

i don't have a big network here. at my mothers house is bigger but I can't complain with how reliable it's been regardless what the map shows

paper tide
#

yeah - mine has been pretty stable too

#

but I'm just a tinkerer by nature ๐Ÿ˜„

#

Oh, I did see an improvement when I moved my wifi to channels 1 & 6, and set the zigbee channel to 26

spiral elk
#

yea

paper tide
#

iirc, my APs are throwing out 200mW each lol

spiral elk
#

many ap's regardless of what channel they are on can also cause issues

paper tide
#

so that has the potential to completely drown out stuff

spiral elk
#

just saw a video on the new(ish) esp mesh network those things can do. curious if they will be a competitor in the future

paper tide
#

I just enabled pair mode on one of the routers - then rejoined a light switch - and the map did show that it all came back as expected, with that switch to that router.

#

will be curious to watch and see if it continues to do that