Zigbee Motion Switch | Project Linus (Blue Series)

[Eric_Inovelli]

Hey all, project update time!

I appreciate all the responses, both positive and constructive ones. It’s always nerve wracking for me to post disappointing news as these projects are basically what I pour most of my time into and it sucks when they go south. Especially this one as there’s a lot of pressure on it – but it’s the risk we took and I’m not letting it fail.

That said, I have some good news which I’ve been excited to share.

I’ve had a chance (along with @EricM_Inovelli) to test the new sensor and it is night and day what we experienced and, quite honestly, how I expected mmWave to perform when I first heard of the technology.

Here’s what we’ve tested so far and then I’ll give the update:

Zones
This was the biggest improvement and what we were missing before. The sensor has the ability to set interference zones from an x, y, and z axis. What this means is that the problems with the sensor going through walls and floors detecting people can now be blocked off left to right, top and bottom and via depth.

Here’s an example:

The left side of the screen is a yellowish square, will not detect anything that goes in that area. The red square is the target detection area where I currently was sleeping (or in this case, about to go to sleep). The red circle is me and it can track me moving across the room.

You can set up to 4 different interference zones and customize the detection zone, making this pretty robust in eliminating false positives.

Think of it this way:

• Interference Zones = Use this if you have a something that’s in the middle of the detection zone. In other words, if you want your detection zone to be 6 meters by 6 meters, but you have something in the middle of the room (floor fan) you would set an interference zone.
• Detection Zone = Use this if you want to eliminate any detection across the x, y, or z plane. A use case would be if you have a ceiling fan at the 5 meter mark and do not see the need to detect anything else at or above the 5 meter mark. You would simply set the detection zone to only detect motion up to 5 meters (or slightly below) rather than setting an interference zone just for the ceiling fan

People Tracking
The sensor can detect where you are in the room and it works great and while the sensor can technically detect up to 3 people, this is more of a beta feature right now per the manufacturer of the mmWave sensor. Single person tracking has worked 100% of the time, whereas multiple people tracking has worked about 90% of the time for me, so I don’t want to guarantee this one yet.

Here’s an example of multi-person tracking (single-person tracking was shown in the screenshot above):

In this example, I had Courtney come into my office and it showed me (at the 3m mark) sitting at my desk, whereas Courtney is shown at the 4m mark).

Breathing Detection
The sensor can detect breathing and I’ve tested this by having it track my sleeping for the past 3 nights and it has never not detected me. It even showed in the logs me getting out of bed to go to the bathroom haha.

Eric M tested it in his bedroom with his wife and it detected both of them.

So, this is huge news because our other sensor would fail if we sat too still for a period of time and the manufacturer’s solution was to increase the timeout time (terrible advice).

–

As for the specs, here are the official specs of the mmWave sensor:

• 60 GHz mmWave Sensor
• 2T2R PCB microband antenna
• ARM Cortex -M3 kernel
• FoV is 120 degrees

For the manufacturer of the mmWave sensor, I’m not going to disclose that just yet (I will once they’re in market as I don’t need to give the competition a head start), but just know that it’s one that’s touted all over the Home Assistant forums and used by a lot of the custom mmWave sensors you see in those forums. They were the original manufacturer that we wanted to go with because our testing (we tested a 24 GHz model) was nothing but positive, but the switch manufacturer talked us out of it (looking back, it was one of the worst mistakes I’ve made not listening to our gut). They are constantly innovating and have a ton of sensors developed already. I’m sure those that are following mmWave sensor manufacturers or have seen any of the custom sensor videos will know who this company is.

–

Anyway, I have officially moved on from the original manufacturer and have sent a purchase order to the new manufacturer (our current one that makes all of our switches) and their ETA for production is February 2025.

Full timeline below (which I will also update):

The structure design was sent to me yesterday and I’m having them make a couple adjustments so that the switches will work if installed with a mud-ring as this was an issue I ran into when installing our current switches at a large B2B project. I should receive an updated file tonight or tomorrow night.

Edit: Forgot to mention this will use the MG24 chip – our current manufacturer has better connections to chip suppliers and the concerns brought up by the original manufacturer are no longer an issue. I’m working through the Matter strategy here, but this will definitely help speed things up and keep costs down since they will share the same chip.

[Eric_Inovelli]

Here are the adjustments I’m having made to the structure.

Currently, our switches are a tight fit for gang-boxes that have a mud-ring attached. They have to be installed perfectly centered and that’s a tough task, especially if there are a ton of wires in the back.

Here’s an example of some that I installed down in Florida for a company.

I thought I took a better picture, but apparently not. You can kind of tell from the picture that the switches are not flush with each other and these were impossible to push in fully as the mud-ring was scraping the sides of the switch and there was no wiggle room.

This is what I’m referencing when I say mud-ring btw as I actually have never heard of these.

I went to the drawing board yesterday to see what could be causing the issue and found that there was zero free space when two switches were installed next to one another:

Even the corners had zero clearance:

–

So, then I deconstructed the switch and found that there was absolutely zero space on the PCB to move things in a bit to free up space on the sides. My thought was to move in the sides a bit as shown below:

The issue here is that it would cause the screws to have to be moved on the front and there was no room for this.

–

Back to the drawing board and it was not looking pretty because I didn’t want to make them redesign the PCB for a minimal use-case. But I also didn’t want to give up on it because there has to be a way to make it work and it’s a good selling point for B2B (and our customers).

I was about to call it quits for the day, when I noticed the “lip” where the front metal plate overlaps the plastic siding and thought, “what if we made this flush”?

It may not be best case scenario, but it would at least free up the switch a bit to move around a bit left to right as the plates were touching when there are two next to each other.

Should be an easy fix (SHOULD… lol).

–

The next part was trying to round off the corners a bit on the back so it wasn’t so tight.

–

The final part has been something that has bothered me for a while, but we’ve never been able to fix, which is the ground screw being either flush with the top or recessed. Not a huge problem as it’s already pretty low and it doesn’t interfere with most gang-boxes, but if we can improve it, that would be preferred.

In the mud-rings, the ground screw is awfully close to the metal:

This one I spent a lot of time on it and came up with this idea:

But… it was ultimately not possible per the feedback. The engineer said he could potentially get it down 1mm (to make it flush, it’s like 1.7mm), so unfortunately, we’ll have to table this one again.

–

The good news is the first two modifications can be done and I’ll get the new file tonight or tomorrow so we can confirm the structure design (target date to complete is Oct. 20, so ahead of schedule).

[CroVlado]

Absolutely fantastic.

Only sad news here is no announcement for thread/matter for it yet

[bobzer]

honestly, I’m glad it’s zigbee and not matter, the ideal scenario would be to start zigbee then with a firmware update to then go to matter. Currently, many more people have zigbee network than matter, but that will change slowly in the future. Also right now matter is not mature enough to totally switch it to matter and for big network it’s a big investment that don’t bring more value.

[rohan]

@bobzer

the ideal scenario would be to start zigbee then with a firmware update to then go to matter.

FYI - it was discussed (much earlier in this thread and others) that it is not possible to do an update from zigbee to thread on anything with an MG24 chip (which this switch will use). The Matter/thread stack ended up being too big to allow for that.

[PJF]

@Eric_Inovelli

It may not be best case scenario, but it would at least free up the switch a bit to move around a bit left to right as the plates were touching when there are two next to each other.

I’m not understanding the point of this change. this would allow the switches to move closer together so they fit in the box but a cover plate won’t fit onto them properly if they’re closer together than the distance the current design allows.

[bobzer]

thanks I missed that

[fisheromega]

@Eric_Inovelli, how will the interference zones be set within the device’s settings? I use Hubitat, and I can see how you could define a detection zone (sorta), but the interference zone seems like it would be a bit harder to define within the device settings. I’m guessing the interference zone would be what is needed to block off the right or left (to keep it from detecting people on the other side of a wall).

[jamesdunham]

Total disagree here I’m afraid. Thread is very quickly (as far as core networking technologies go) emerging as the default next-gen wireless networking standard - with backing from many big players. Not shilling for thread, just observing real-world support out there. I have a substantial zigbee network, but I’m now starting to move things to thread (actually just installed my first white dimmer switch and it’s working great) as to be future proof. If I’m spending a lot of money and time to outfit my house with great Inovelli switches I’d rather go straight to the best possible networking solution.

[nasa01]

Great news! Appreciate the update. One question - will the z-wave version be coming at the same time as the zigbee?

[Eric_Inovelli]

@CroVlado

Only sad news here is no announcement for thread/matter for it yet

Yeah I’d like to actually talk about that with you guys and get your opinion on it honestly.

I’ll write up more once I get a chance to talk with @EricM_Inovelli about it, but I have some concerns based on feedback we’ve been getting lately regarding our White Series.

The biggest issue I see is that most Matter users are Apple users and the sticking point is that there’s no way to change any of the parameters except for doing it locally, which, quite honestly, is a pain.

Now imagine trying to set zones locally at your switch. I feel like that’s going to be a trainwreck lol. And if the zones are set improperly, it would lead to false positives (or negatives) and frustration.

I realize there are Home Assistant, Hubitat and SmartThings users that can use Matter and change parameters easily, but I have a hunch that even if we put the disclaimer on the site that zones need to be set locally on Apple, those disclaimers would be ignored or not seen and lead to a lot of returns out of frustration.

However, we do have a contact now that is willing to help us with Apple, and he mentioned that if we can sell them on an innovative product, they usually listen and it’s possible they would help integrate it. But, at the end of the day, it’s out of our hands, and it’s a big, “if” on if he can help us.

One idea is to release it in Zigbee as I think we’ve found a way to convert it to Matter via OTA (I want to confirm this…). If this is true, it would at least weed out the Apple people initially as Zigbee will not work with Apple (and Google, Amazon, etc - not just trying to single out Apple, they’re probably the best of the three listed honestly in terms of Matter implementation) and people with hubs that support Zigbee and Matter would just update to Matter.

Idk, just throwing out some thoughts. Would love an open discussion on it!

[Eric_Inovelli]

@rohan

that it is not possible to do an update from zigbee to thread on anything with an MG24 chip (which this switch will use). The Matter/thread stack ended up being too big to allow for that

Yeah, this has gone back and forth and you’re right the current stance was that it was not possible. However, I think we may have found a way. Let me confirm. I know we’ve been able to convert Matter to Zigbee and I believe we can do it Zigbee to Matter at the very least via the wire harness.

@PJF

I’m not understanding the point of this change. this would allow the switches to move closer together so they fit in the box but a cover plate won’t fit onto them properly if they’re closer together than the distance the current design allows.

Hmmm… You have me thinking a bit more now. I was looking at this from the point of using screwless plates vs traditional plates and yes, this may cause an issue with traditional plates.

With screwless plates, the easiest way to install them is to screw the switches in just enough so it’s just about fully tightened, but leave enough slack so you can move the switch around and position the under plate accordingly. Then once the under plate is screwed in, you tighten the screws down and put the top plate on. Removing the sides, would allow the switches to easily get into the gang box.

I suppose this concept would work on traditional plates, but I will have to do some testing.

@fisheromega

@Eric_Inovelli, how will the interference zones be set within the device’s settings? I use Hubitat, and I can see how you could define a detection zone (sorta), but the interference zone seems like it would be a bit harder to define within the device settings. I’m guessing the interference zone would be what is needed to block off the right or left (to keep it from detecting people on the other side of a wall).

I’ll defer to @EricM_Inovelli on this one, but I’m guessing it will be a parameter that you can set.

For example, something like the following:

• Parameter 1 = Detection Zone - X-Axis Max
• Parameter 2 = Detection Zone - X-Axis Min
• Parameter 3 = Detection Zone - Y-Axis Max
• Parameter 4 = Detection Zone - Y-Axis Min
• Parameter 5 = Detection Zone - Z-Axis Max
• Parameter 6 = Detection Zone - Z-Axis Min
• Parameters 7-12 = Interference Zone 1 X/Y/Z Max/Min
• Parameters 13-18 = Interference Zone 2 X/Y/Z Max/Min
• Parameters 19-24 = Interference Zone 3 X/Y/Z Max/Min
• Parameters 25-30 = Interference Zone 4 X/Y/Z Max/Min

Let’s use a room as an example where you only want to detect people in that room. The room is 5m Long x 4m wide x 3m high and your switch is at 1.5m above the floor.

Let’s assume your switch is positioned in a way that there’s 1m to the left and 3m to the right).

Let’s also assume you have a ceiling fan.

You would set the Z Values at (this is the height above and below the switch):

• Max = 2 (if you want to exclude the fan - assuming the fan is 1m below the ceiling) or 3 (if you want to use an interference zone)
• Min = 3

Set the X values at (this is left to right)

• Max = 1
• Min = 3

Set the Y values (this is front to back)

• Max = 5
• Min = 0 (as the sensor can’t detect anything behind it)

Now, let’s assume you have something interfering (stationary fan) in the middle of the room.

You would follow the same procedure above to set a “barrier” around the device that is interfering with the sensor.

Hope that helps better clarify?

[ibrewster]

@Eric_Inovelli

@CroVlado

Only sad news here is no announcement for thread/matter for it yet

Yeah I’d like to actually talk about that with you guys and get your opinion on it honestly.

Honestly, my opinion is that I don’t see the draw of matter/thread. I started with z-wave, added ziggby to be able to use the fan canopy modules, and just recently added a matter/thread door lock. They all work, but I honestly don’t see any reason to go with matter/thread over the others.

Sure matter may be the future at some point, but from what I’ve seen it isn’t here yet, so why would I switch? I have zigbee capability, I have z-wave capability, and those aren’t going anywhere anytime soon, so I’m perfectly happy to continue using those - including with new devices - until such a time as I can only get matter devices. Perhaps at some point I’ll have more matter devices than zigbee/zwave, but I don’t imagine it will be any time soon.

[marsara9]

Just my own personal experience. But I actually started migrating everything to thread/matter specifically because my zigbee network was getting too big and I was starting to experience delays or non-responsive devices. My hope was was that thread would, in theory, support many times as more devices per network than zigbee could. Plus it’s a lot cheaper to install a single light switch than it is to install 2-6 Hue bulbs, per switch. That and I’m already at the limit for Hue devices without getting a 2nd hub (eww).

When I saw this project I got excited because the first question I got from my wife was if there was a way to tie the new switches to a motion sensor. I ended up getting a couple Apollo MSR-2s but they still clutter up my WiFi network AND they require a nearby outlet that’s hopefully in a central-ish location.

So I’m more than willing to wait, as the MSRs are satisfactory but personally I’d only be interested in this switch if it could support thread/matter.

[sdholden]

Matter offers nothing right now, and it MAY, MAY offer marginal improvements in the future, over existing protocols. The only reason to go with matter right now is if you’re tied to the apple home ecosystem, and even then, it’s an awful experience (I’ve done it).

[ibrewster]

@marsara9

Plus it’s a lot cheaper to install a single light switch than it is to install 2-6 Hue bulbs, per switch

Yeah - personally, I think “smart bulbs” and the like are “smart home done wrong” I have a grand total of three smart bulbs in my house, and that’s only because I got them free. Everything else is on zwave switches.

Not quite sure how that has anything to do with Matter or no Matter though…

[Eric_Inovelli]

I understand the pro/cons of Matter and can see both sides for sure.

I guess my question is, for those that cannot update the zones nor any of the other parameters via an app, are you willing to have a semi-painful setup experience by using the local method (config button).

In other words, let’s focus on user experience and if the limitations of having to change parameters manually outside the app is too much of a deterrent to go with Matter.

[CroVlado]

Continuing the discussion from Zigbee / Matter Motion Switch | Project Linus (Blue Series):

@Eric_Inovelli

@CroVlado

Only sad news here is no announcement for thread/matter for it yet

Yeah I’d like to actually talk about that with you guys and get your opinion on it honestly.

I understand fully what you mean. I’ve already seen the hundreds of complaints about the white switch by people who just want the “non-technical” “no options” blah blah basic dimmer that’s matter over thread.
I am fully convinced that this should be advertised with huge print at the top of the page that this will be painstaking to configure if you choose to not use home assistant. Then again even with home assistant not every parameter is addressable in the config list on the whites (8+9 are bound together, 10-13 aren’t listed at all)

One main reason I do like apple is the convenience of iCloud access to the home. While sure I can set up Tailscale and set that up on my phone etc it’s just not as family friendly. I export back to apple home from home assistant because personally I think their app looks great and works well.
Another big selling point for apple border routers is if I need to extend my network to my shop behind my house, I just buy an Apple TV and it becomes a relay over the network for additional matter/thread as well as Bluetooth devices, with Zigbee (I could be misinformed on this) I’d have to run another hub that can do Zigbee and somehow link it back to home assistant or run a second home assistant instance with Zigbee stick? While that’s alright the Apple TV/homepod serve dual purpose being a smart tv set top box/music player while also doing smart home stuff in the background.

I also think with thread 1.4 standard when implemented having every device on 1 large thread network vs every controller making their own thread fabric should start making interoperability a lot easier. While matter and thread protocols are still fairly new, manufactures not using it will not help its development either. Prime example Nanoleaf leaving matter over thread to go to matter over WiFi on a lot of new products which is a huge disappointment.

The idea of potentially updating the MG24 chip to thread if the possibility exists eventually, maybe, it leaves to question how the chip is tied in to the mmWave sensor. Is the current chip in the white series switches not able to handle it?

Like I said in previous posts, if the cost is higher, I’m all for another kickstarter or whatever for a white series mmWave switch at a higher cost.
I am one of the many users who have tried Aqara fp2 mmWave detectors that after constant false positives and “sticky” detections, it’s just short of ending up in a trash can because trying to automate with it is impossible.

Lastly, what you said about having a contact to help with apple. I’m trying to clarify the context, as perhaps I’m not catching on. Is this contact supposed to help sway apple to implement something, or are they to make something to ease the pain with dealing with apple. Based on the pleading people have done over HomeKit all over Apple support and feedback sites for years now, I doubt a company a stubborn as Apple is willing to listen to anyone but their own egos, it’s actually pathetic how these large companies just don’t care regardless of how much it could improve usability.

Regardless of Apple and apple home, I think matter over thread will outgrow the other standards because the “big 3” are half-ass teaming up to support this protocol and as the standard grows to support more devices, more and more manufacturers will likely go to it to standardize and reduce cost on production across all their lines while increasing compatibility across platforms, and therefore increasing profits for increased sales. No more of this “only works with (alexa/google/apple) bs.

[fisheromega]

Thank you. As always, you have delivered clear and detailed feedback.

Looks like I’ll need to break out my old architecture books from college and draw up floor plans to map out my measurements. Then get a tape measure for the dimensions, then a slide-rule to convert from English to metric.

[vreihen]

@Eric_Inovelli

I guess my question is, for those that cannot update the zones nor any of the other parameters via an app, are you willing to have a semi-painful setup experience by using the local method (config button).

Is there a technical reason preventing the development of an Inovelli iOS app that talks directly to the White Series switches via Thread/Matter for configuring them directly, without Home Assistant or any other HA system besides Apple Home???