I am looking into the possibility to log the water usage in Home Assistant. I currently have an Elster water meter which unfortunately does not have a magnetic/metal component spinning or a diode indicating the water use. Apart from image recognition, I suppose there are no other options to integrate its readings in Home Assistant.
I reached out to my water company (Vasyd, I live in Sweden) and they informed me that in the near future they will be replacing my water meter with an Axioma Qalcosonic W1 ultrasonic water meter.
According to a very kind person from Vasyd:
"The meter is measuring the water using ultrasonic and outputs a LoRa-signal with end-to-end encryption.
IĀ“ve heard about a solution where one can get a m-bus dongle and we can activate m-bus signal on the meter remotely "
I also read in the user manual that the meter can have an IR diode indicating the use (I assume).
Does anyone have any ideas on how to proceed with this? Which option is more viable and easier to implement: mbus dongle or IR readout (I have to check if my model would have this option.
No update from me unfortunately since I am at a stage where I do not even know when the meter might me installed. I will of course let you know if I learn more and when the meter is installed.
Iāve got one of these meters, and been thinking of ways to read it in a useful way for HA.
I found a manual online (canāt figure out how to attach anything other than images though), which says itās only possible to communicate with the meter for a limited amount of time to save battery.
In order to save the battery, communication credit
system is implemented into the meter. Time of communication through additional interfaces is automatically limited to save the battery (up to 20 minutes per month). Unused communication limit is summed up. If the limit is expired, the interface is blocked and the new time limit of communications will start only after the change of the hour (16 seconds for each next hour).
As Iām looking for realtime, constant updates, it kind of kills it for me.
I have been doing some tests with pointing a camera at the display and ocr-ing the digits using ssocr, but itās really hard to get a good enough image in my environment (pitch black, and light source added reflects in the display ruining the images)
Interesting! I am also looking for real time data. I have considered grohe sense because it is both a water meter and a water valve in one device. I am still following the discussion related to integrating it into HA. However, since Grohe Sense measures the waterās flow through it, I think that the easiest and cheapest solution would be to install a second water meter after the cityās water meter instead. I would then choose one with a reflective or magnetic patch that can be used to read out the real time data. This solution would be many times cheaper than Grohe Sense and would be independent of when I get the new meter and how tricky it would be to read out the data from it. I already installed an electro valve to be able to cut the water supply whenever there is a spill somewhere. To have a complete system, I only miss the water use data.
I will also get a new meter soon, so will follow this thread.
Today I use AI-on-the-edge for my old dumb meter and it works fine. So I will probably use this for the new meter as well, but it would be cool to communicate with the meter directly. I read in the spec that it has NFC so this could be a way to get the values as well.
I will have my meter installed next week. I got a letter with it with some very general information that it will allow for a remote readout of the water consumption (by the water supply company) but also some āsmarterā functions like a water leak notification with time. I will try to followup with the person whom I mentioned in the first post but I am afraid that the things will move very slowly through the official channels. It would be a lot easier if someone already had a tested solution different than the image recognition.
I have tried the NFC approach.
There is an Android app called Qalcosonic W1 available from the usual apk download sites which successfully reads data from the meter.
However, reading using a regular NFC reader app does not produce anything I have been able to make any sense of, presumably itās encoded somehow.
Someone with reverse-engineering skills would probably solve this pretty fastā¦
I got the meter installed but the smart capabilities are not active yet. I did not get any date regarding when my company might start using the remote readout. I will check with them again to see if there is any chance that they could activate mbus signal remotely as they mentioned but likely not until they have the whole infrastructure for remote readout in place. I suppose this will never work with the live updates given that the meter needs to save the battery by sending the updates only couple of times a day.
I have the same flow meter and utility operator takes remote readings once per month. As it says on the meter, the battery (optimistic) lifetime is approx. 15 years (see expiry date on the device). If they would allow me to read the data even once per week, the life time will be 4 time shorter, maybe?
Didnāt find any interface for the battery change on the device (neither nothing about that in manual) - I would guess that the whole unit is replaced when the battery dies.
Having in mind all above, I donāt see why the service provider would allow someone else to access the readings and shorten the lifetime of the device.
Personally, I went with the ESP cam and ai on the edge solution and I am very happy. It would be cool to read the data directly from the meter but I wanted more frequent readouts which I think would never be possible. AI on the edge
Today Iāve received a new water-meter - after short check here on this forum page and internet googeling - Iāve decided to re-activate my AI on the edge setup.
Seems to work well.
Mostly the light setup was tricky, not to get a ābadā reflection.
Update-Rate was first every 2 minutes - now I try even each 1 Minute (since the log says) done after 39 seconds.
