Custom zigbee multisensor

[JVKran (Jochem)]

Hi there!

As a long-time home-automation fan and embedded systems engineer, I’ve always found it unfortunate that all home-automation devices are closed-source and we’re dependent on the manufacturer’s efforts regarding compatibility and functionality. This has resulted in a user-base effort of implementing quirks and workarounds and a large stack of similarly functioning devices.

This, in my opinion, requires an unnecessary amount of time, effort and resources. A few months ago, I took matters into my own hands and came up with a product idea; a multisensor/actuator with a breakout for extendable functionality. Soon, I started prototyping hardware and software. Since then, a lot of software has been tweaked, hardware designed and soldered (and fried) and multisensors tested. This brings us to this post; we’re ready to take things to the next level, by enabling other people to provide feedback to further improve the product.

While aiming not to make this a complete advert, we do encourage you to check out our website with a complete list of features, blog and manual.

In short, we’ve made a multisensor based on the nRF Connect SDK in combination with a temperature, humidity and pressure sensor, illuminance sensor, motion sensor, piezzo buzzer, RGB-led and over-the air update support. These work as expected; sensors are represented as sensors, the buzzer is controllable as alarm and the led can be controlled as a color-bulb.

Ironically, we’ve chosen for closed-source software for now. This will become open-source when we have a larger production capacity and have incorporated more feedback from user experiences. Keeping the code closed-source allows us to precisely control the speed of development. While we’re aware this is quite controversial, we do believe this is the right choice for the upcoming 3-4 months. Early 2023 we’re going open-source nonetheless; we either accomplished our goals, or we’re halting our efforts.

After having open-sourced our code, we hope to see a community forming and collaborations happening in the open GitLab environment. We also hope people to come up with new ideas, concepts and implementations for expansion boards; the ultimate goal.

I’d love to hear your opinion and get some feedback!

[nathan.tarillon]

Hello,

It is a really nice project. Can’t wait to see the future of it !

What are the ideas of extensions board you have in thoughts that may come from the community ?

I really would like to buy it and test it, but the price is a little bit to high for me right now.
I understand that you’re developping the project and that you maybe dont have better industrials costs. As comparison a cheap zigbee PIR sensor is about 5€ and a temp/hy/pressure sensor about 11€ so 80€ is a bit much for only a led and a bipper in addition and no extension board at this time .
Don’t get me wrong, this is only my thought, but once again very nice project !

Are you planning to put your project on a crowfunding platfrom ?

[ottomatt (Matt)]

I am new to automation and have spent months online searching for something to plug my mix of “industrial” sensors into and communicate via Zigbee. I have purchased some “boutique” maker products on Tindie, but would like an “off the shelf” choice, with a consistant supply of chips.

Everyone makes your basic temp, humidity, PIR and switch but I have had great difficulty trying to find a zigbee device to send tank level, pH, EC, better/more robust consistant temp/humidity, and PAR.
The switches tend to be 10 amps when my mains wiring is 15 or 20 amps. a 20 amp relay would be more usable.
These sort of communicator devices are made for LORA, WIFI, BT, and PLC controllers but zigbee is cheap and works for me.

I am puzzled that there doesnt seem to be a simple way to connect standard sensors. I am not a electrical engineer or programer, but to someone with those skills it seems most sensors are either a voltage or a current, or an on of state.

I decided that the Nordic chip on a development board might work, they are available, (Unlike cc2530)
But with my limited skill set.I am not sure that i would be able to get it to work,
This morning I was finally ready to post a question if so "experts " thought the Nordic development board would be a good choice when I saw your post.
Could your zigbee multi sensor be adapted for my sensors?
Thanks,

[JVKran (Jochem)]

Hi @nathan.tarillon!

Thanks, I’m happy to hear you think it’s a nice project. After all, that’s what we’re doing it for!

Some ideas we currently have are a volume-level monitor (with volume-sensor), tamper-sensor (detect motion, angle, acceleration, etc.), window/door sensor (reed switch), rotary encoder (with magnets) to for example control the brightness of a led.

Regarding the cost, we’re aware the price is steep and we’re not happy with that. Unfortunately, like you said, especially this batch, we’re not able to lower the price as we would then not even break even. While we hope to lower the price in the future, don’t expect miracles; we won’t be able to (commercially) compete with large manufacturers (I don’t need to name names I think ).

We’re not currently planning on crowdfunding, but to be honest, we haven’t put much thought on that. That’s something to look into. Thanks for the suggestion and your thoughts!

[JVKran (Jochem)]

Hi @ottomatt,

That’s quite an interesting question! To be extremely short; yes, our multisensor could be used for that. But, two things to consider that immediately cross my mind;

1. Operating conditions and requirements. A device operating in an industrial environment needs to be rated and tested on a whole other level than an IoT-device like our multisensor.
2. While the software would most likely be usable as basis (extensions needed for new measurements of course), the hardware would most likely need to be modified to support connections/connectors for ‘industrial’ sensors. Do you have an example of such an ‘industrial’ sensor? If it’s just I2C, GPIO, etc. then you could come very far, even with our current approach. It would then not be a weird thought to implement these as extension board.

So, in general, it’s most likely that you could use our software but it’s desired or required (based on the type of the ‘industrial’ sensor) to create a new hardware design.

Thanks for your interesting thoughts!

[ottomatt (Matt)]

Nathans point of the cost 80 Euros is high for those “common” residential uses but I would gladly pay that for a Zigbee board that could sense and control 2 to 5 sensors and have 1 or more outputs that the LARGE manufacturers are ignoring.
I use the term industrial loosely. I believe that most of these chips and boards are rated to 105 degrees C in order to be used in light fixtures. Beyond that I think that your product would be reasonably moisture and dust resistant or could be placed in an additional plastic case if needed.
That would suffice most use cases. I use my HA system in a commercial building with 12" (30 cm) thick concrete walls with steel fire door and zigbee works reasonably well in the four rooms that are connected. I do have to place a few extra plug in routers in the hall way to reach every thing.

[JVKran (Jochem)]

Thanks for the elaboration and description of your use-case. In that case, indeed, I think our multisensor would be very suitable for that! We could always provide custom cases/provide technical drawings so that people can design their own cases for specific use-cases and environments.

[nathan.tarillon]

I totally agree with the idea of a board that can do such a thing. I also thinking to use some of “industrial” sensors that I own for some application that, as you say, are missing from almost every supplier.
As an hardware electronic engineer, the hardware part wouldn’t be a problem, but the software part is an other matter.

[ottomatt (Matt)]

I think that most sensors have some opensource software. It is my understanding that it basically multiplies the voltage or current by a constant, which I think is called “SCALING” to give you a output value that is understandable
That said, most output a 0-5 volts, 0-10 volts, or a current of 4mA-20mA

Below, is a chart of the theoretical out put of a pH Probe. For accurate measurements in the real world
you callibrate it against a known solutions.( 4.0, 7.0, and 10.0 pH calibration solutions are common in the US.)

pH Voltage
pH Alkalinity Millivolts
0 pH +400 mV
7 pH 0 mV
14 pH -400 mV
pH equals Voltage
breakout number values
0 pH +400mV
1 pH +342.86 mV
2 pH +285.71 mV
3 pH +228.57 mV
4 pH +171.43 mV
5 pH +114.29 mV
6 pH +57.14 mV
7 pH zero mV
8 pH -57.14 mV
9 pH -114.29 mV
10 pH -171.43 mV
11 pH -228.57 mV
12 pH -285.71 mV
13 pH -342.86 mV
14 pH -400 mV

[JVKran (Jochem)]

Thanks for the clear example. This is certainly doable, but not without its caveats of course;

1. The scale-factor and range have to be set by the user. This is possible through the ZCL configuration cluster so one can do this from Home-Assistant’s UI.
2. Both, voltage and current outputting devices, have to be interfaced to the microcontroller’s ADC. As a result, we’d have to convert the output current to voltage and convert/shift the negative voltage to a positive one.
3. Another power-supply will be needed; 4- 20mA is a lot in terms of a battery.
4. The device/breakout needs to be equipped with connectors for the sensors. I assume these don’t come with JST-connectors?

All interesting and doable stuff!

Oh, by the way, I’m rethinking our base-pricepoint. Making a loss, at least for now, is not a big deal. While it’s not ideal either, it’s better than this not lifting off the ground at all. Furthermore, as we’re still figuring things out, we understand that the cost of shipment and additional expansion-boards has to be added to the base-price. That becomes an even sturdier price-tag. What would you, @ottomatt and @nathan.tarillon, think a tempting base price-point is? We have held a reddit poll a while ago, but while the results are to be expected, they aren’t realistic in our current state.

We are planning on a new hardware version with different and cheaper sensors to reduce the cost-price while maintaining the same functionality.

Personally I’m thinking about 60 euro (58 usd). This is a big difference and will result in no profit until we have the new hardware version ready for use (and a small one afterwards).

[nathan.tarillon]

Hello Jochem,

I was thinking more about 50-55€ but 60 seems correct too, as I understand the process, and I don’t want to stole you ! And if you lower the price tag I will very certainly order one (to begin)

Are you able and will you OK to maybe share the bill of material with the cost of each part, in order to, if you want, help for sourcing better price for current components or cheaper components ?

Also, are you planning in the future a wire powered version ?

[ottomatt (Matt)]

I have a very basic question.
It seems like off the shelf ESP Home and maybe LORA can do everything that I would like to do with Zigbee. Is there a reason that they have chosen to not use Zigbee.
I have the belief that WIFI is less reliable and that battery units are impractical.

Should I just give up and switch these sensors to ESP or LORA?
What would i gain and loose?
Why did HA choose to acquire ESP over Zigbee?

[JVKran (Jochem)]

Hi @nathan.tarillon,

Well, price is, as my companions and I found out today, a difficult and ethical aspect of starting a business. We invest time and money into developing a product, make mistakes that cost us money, and afterwards make the product available for the larger public. While we don’t want to make people pay for our mistakes, we still are trying to start an open-source company, not a non-profit. As a result, we try to work for a market-conform value aiming towards a market-conform price-tag. As of today we decided and believe €60,- to be that market-conform price-tag.

From V2.0 onwards, we hope to also be able to work conform a market-conform value, thanks to cheaper components. As a partly omitted bill of material (complete except for resistors and capacitors), we’re looking into changing the following components;

1. We’re swapping Minew’s MS88SF2 for Ebyte’s E73-2G4M08S1C.
2. Same goes for ROHM Semiconductor’s BH1750 for Vishay’s VEML3235.
3. And for Bosch’s BME280 for Sensirion’s SHTC3. This unfortunately removes the air-pressure sensor, but we believe this not to be of great essence for in-home appliances.
4. Osram’s KRTBAELPS1A6747 remains.
5. The AM312 PIR remains as we haven’t found a likewise performing and power consuming passive infrared sensor.
6. We’re replacing Harwin’s pricey M22-6540642R with a pricier but we think more suitable board-to-board connector from KYOCERA 209159006101916. This is part of the reason why future expansion boards won’t be connectable to the current batch of multisensors. As stated on the website, we do aim to provide an adapter board to make expansion boards available for the current batch as well, only requiring a new case to be printed (supplied with the adapter).

I would like to name prices, but the prices change with the day. Hence, I provided links to the components in the hope they’re also available for other locations than The Netherlands.

If you have any suggestions for alternatives, albeit cheaper/better fitting/other, please let me know. Thank you very much for your participation and help.

Regarding the wire powered version; yes, this is on our to-do list. Whether we’re going to develop this as another product or breakout, has not yet been decided. What would you like to see?

[JVKran (Jochem)]

Hi @ottomatt,

That’s a very broad and situation dependent question, but I can list some considerations;

1. ESPHome uses chips from Espressif. These are very cheap, like €1,- for an ESP32, in China. While cheap, they only support WiFi and Bluetooth and use extraordinary amounts of power. Truly power houses. Dual-core processor. Current-draw of up to 400mA. As a result, they require a beefier power-supply than a battery; more like a lipo, powerbank or wall-adapter. ZigBee is low-power, drawing less than 40uA on average while maintaining a low latency.
2. The more devices with a WiFi-connection on a network, the more congested the network becomes. This results in worse network performance, not just for the ESP’s, but also for your phone and other devices with WiFi. ZigBee performs on the same channel as WiFi, but traffic is not routed by the router.
3. Security. Chips from Espressif, but chips with WiFi in general, are known to have vulnerabilities. ZigBee and other local networks don’t face issues regarding vulnerabilities.

I’m sure there are others, but I’m by no means a networking expert.

Then there’s another key point to understand; HA didn’t choose ESP over ZigBee. They’re just different solutions to partly the same and partly different solutions. There’s more than WiFi, LORA, Bluetooth and ZigBee. Why not LTE, Bluetooth Mesh, Matter, Zwave, Thread, 433MHz, etc? They all have their strengths and weaknesses.

I’m not an ESPHome user, but from what I understand, they aim to provide people with the opportunity to easily create devices and interface them with Home-Automation platforms. That’s great! But not standardized. ZigBee for example, doesn’t work with just Home-Assistant, it also works with SmartThings, proprietary networks, alarm systems, etc; it’s a standard defined by the ZigBee alliance.

In the end, it boils down to situational factors; do you already have such a network, how fast do you need it operational, what are the requirements (power-source, latency, security, etc.), how reliable does it need to be, does installation allow user intervention, etc. There’s no such thing as the perfect suitable network. Otherwise, there would only be one. ZigBee and Zwave just turn out to be popular amongst Home-Automation.

[colinb (Colin)]

Liking the look of this JVKran! Will be following with interest. Currently in the midst of a home renovation so not in a position to test unfortunately.

How is the sensor mounted? Is it possible to e.g. glue/screw the rear door to a wall, and remove the whole sensor when battery change needed?

Can you comment on feasibility of using rechargeable CR123a batteries?

One thing that would make me happy to pay even the EUR 80 price range would be presence detection and object discrimination as opposed to motion detection. It’s annoying when, because you’re so engrossed in the movie, the lights go off because you haven’t moved. Or when the cat sets off the motion detector and turns on the lights. Being able to use, or already having installed!, something like the AMG8833 8x8 IR sensor would be tremendous, and a real USP Vs the rest of the market

[nathan.tarillon]

Jochem,

You didn’t make mistakes, you learned a way how to not do it !

Otherwise, thank you for sharing some parts of the BOM, I will have a look.

About the wire powered version, I see 2 options :
- an USB connector with a 18650 battery to replace the CR123A
- as you are in the modularity, maybe 2 different power boards, one with the CR123A, and another with only a power supply (12 or 5V)

Also, for the case, I worked with a company that create box in plastic, that may be an alternative for 3D printing : https://www.ltp.fr/ .
It is a french company but i’m pretty sure that they have someone who speak english, otherwise, and if you are interested I can do the translate !
For example we created a black box about 450x120x120mm, with a sliding opening, a translucent visualization window, a printed logo, an 4 captive screws (that cost a lot!). For 120 units per year we were at 40€ each.

[JVKran (Jochem)]

Hi @colinb,

I’m happy you’re interested and like the look of it! That’s what we’re doing it for, after all.

The sensor is a unibody with backplate to keep the battery in position and keep dust out. The backplate is a snap-fit lid with little clamps that can be removed with quite some force. When the backplate is mounted (screwed/glued) to a wall or other object, I assume it would be easier to just take the multisensor of. No issues there; possible!

Rechargeable CR123A batteries are no different to the multisensor than one-time use ‘normal’ CR123A batteries. As long as it is able to provide ample current for the leds and buzzer, using these should be no problem. @nathan.tarillon’s comment underneath also makes us think about using a modular approach for the power-provision.

The presence-detection comment is very useful; we looked into this briefly before but didn’t really find such sensors with I2C bus. The one you mentioned, might be a perfect fit! One challenge/consequence to beware of is that these sensors draw quite (in comparison with currently present sensors and modules) a lot of current, drastically reducing lifetime. This however, won’t be a problem when using a wire-powered version.

Thanks for your thoughts and good luck with the renovation!

[JVKran (Jochem)]

Hi @nathan.tarillon,

That’s a kind and, to my opinion, true way of putting it. Thanks.

The 2 options regarding the wire-powered version seem great. It would then even be possible for people to make custom power-providing boards with e.g. a lipo and solar panel, AC-mains connection (don’t know why you would want that, but still) or for other niche situations. Then we could simply have two base modules (one battery and one with USB-C) and let people choose what variant they would like it to ship with.

The company you mentioned seems like a great contender. For the time being though, I think we’re better of 3D-printing. This is, at least for now, cheaper and enables us (and soon maybe/hopefully others) to keep tweaking the design. Nevertheless, when we do go for this option, we’ll keep your offer of translating in mind. Thanks again, for your cooperation.

[nathan.tarillon]

Hello @JVKran !

First, thank you for your fast treatment of my order ! Really can’t wait to receive it !

Concerning the wire-powered version, the AC-mains connection may be useful in certains integrations cases, in-wall for example… but not my expectations.
I said earlier an USB connection would be great as it is the most common, but I really believe that an old jack connection is, in my opinion, less fragile as the connector can rotate, and also won’t broke if it fall.
I have 2 broadlink mini powered by a micro USB, and I’m planning to change the connection.
I’ll follow you either way you going, and the custom power-providing board is a great idea

I totally understand that you keep going on the 3D printed way, that’s the better choice at the moment. Just keep in mind for the future there is a lot of possibility

[JVKran (Jochem)]

Hi all,

I wanted to let you all know that we just went open-source! Our source-code can be found over at GitLab.

Have a nice weekend!