I asked my dentist if he would help with x-raying the Honeywell T110 probe to see where the reed switch was embedded in it.
The probe is an official, global meter supplier and water board approved device. It is shaped correctly to be fitted directly to one of the common water meters widely deployed, the Elster V100, by removing a plug, and screwing it into the receptacle with the captive screw, the reed switch placement being ideal for pulse sensing. The reed switch device has a series 100ohm current limiting resistor, as well as two tamper wires, making a total of four wires.
A reed switch is two flexible seperated wires inside a small glass enclosure that make contact when energised by a magnet with sufficient flux to attract them to touch, with two external wires to connect to. These are widely deployed in the security industry for decades, sensing door and window closures by adding a magnet to the moving part and the reed switch on the fixed part closes when the magnet is near. Simple and very reliable, the glass enclosure excluding the elements while keeping the two wires inside at a specific distance. Did I say robust? The magnet does not have to be very strong if it is very close to the reed switch. Normally magnetic detection is most sensitive at right angles to the switch. The meter company designs their meter so the wheel that rotates the fastest (the smallest digit) has sufficient magnetism to trigger their reed switch reliably. Unfortunately, by their inherent design, a passing magentised wheel will not always cause them to snap shut immediately (often if it is a very slow moving magnet such as a water meter with a slow leak) and the contacts will bounce or have a small spark/arc which a digital pulse sensor will interpret as multiple closures and give inaccurate results. You filter these with a resistor/capacitor combination, or digitally, over a small period with multiple pulses being averaged out. Both have limitations - how to differentiate a fast moving pulse that is valid from multiple bounces that are actually one pulse, and a meter where the wheel is so slow moving that you are not sure if the pulse is yet over or not?
Honeywell have the reed switch enclosed in a black plastic shape which has a captive screw and the wires are looped so a strong yank will not dislodge the internal reed switch. The series resistor is for current limiting and can be useful to detect remotely if there has been a cable defect or break. The other two wires are looped to enable cable tampering detection, if the cable is cut to stop pulses being reported for billing purposes.
Reece Plumbing (a large trade plumbing supplier in Australia) will get it in for you if you ask nicely, and trade price is a smidgin over $50 if you have a cousin in the trade (grins - don’t ask for the walk-in over-the-counter price without sitting down if you have a pacemaker - daylight robbery). Takes about a week according to my cousin.
What you end up with is a simple reed switch with all the resulting contact bounce issues, attached to a length of four core wire. These contact bounces-s-s-s can be filtered out in software, with varying success, as evidenced by this thread and others. If the magnet wheel stops turning just at the threshold of your reed switch contacts closing, the oscillations can run the readings to the moon and back, especially if you live close to a busy road with heavy vehicles passing.
Using a hall sensor is another slightly more expensive but more reliable option - if we knew the magnetic hotspot that is optimum for each meter type. You get a bell shaped response curve on your magnetic sensor for magnetic proximity as the magnet rolls past, and you sense a change as a pulse when it hits a threshold. Some posters seemed to have fluked it, and others haven’t from reading between the lines. What I would like is consistency, reliability, and reproducibility.
What would be nice (and the meter folks and Honeywell would be absolutely livid) would be for somebody to x-ray the reed switch size and placement and post the details on somewhere like ThingVerse where everybody with a 3D printer and a 20 cent reed switch from China could replicate, as my sense of outrage at the cost of the genuine article gnaws at me.
Every dentist I have approached seems less than keen on using their $500,000 xray machine to x-ray a $50 part. ‘What settings to use’ was the closest I ever got to a yes.
I suppose I could melt it down, or microwave it till the guts fell apart. Materials testing lab attached to an university with motivated students and professors with a burning curiosity? I’m surprised nobody has reverse engineered this simple part yet, and by the time it has been done, smart sensors will be widely deployed.
Others that have a different meter type sense the rotating head on the display that is often magnetised, or reflective, specifically for this reason to be able to be digitally sensed.
Other sit a camera over the top and use AI optical recognition smarts to decode it for actual readings. The covering caps over the meter to regulate light levels for reliable camera readings are a dead giveaway to the utility company that something non-approved is happening.
Ideally, a water company fitted monitoring device such as a Cyble (expensive but read the specs and promises) would allow you to import the readings directly into HomeAssistant. For battery consumption purposes (they promise 10 years) the number of times you poll may not suit most people that need to know to the 10 seconds how much water has been consumed, but the water company bills you in THOUSANDS OF LITRES (kL) consumed. And will probably update their database daily or weekly. The granularity requirements are discordant. Giving you access to this over the Internetz while keeping the HomeAssistant DDOS web scrapes every few seconds at bay would give anybody in the company carefully manicured internal billing databases severe heartburn. Witness the ongoing two year old billing integration debacle Greater Western Water are still struggling to solve (and still ongoing losing millions of dollars with unrecoverable bills the regulator prevents them from recovering) - their customer service staff are struggling to access it, let alone customers through a web portal or API. Offering to be a pilot project customer for their R&D labs often involves NDA’s.
Each of these options has cost as well as functionality compromises that have to be worked around.
Similar with your gas meter. A reed switch strategically placed to tuck in underneath the display but still detect the rotating wheel. Where is the optimum hotspot?
Your meter company will gladly remove anything that looks to be a ‘hobbyist’ wiring attachment to their carefully calibrated device, citing tampering regulations. So will any passing drunk that leers at your unfamiliar meter object, as well as bored teenagers with spare time and vandalism objectives. Not so much an issue with water meters, but I cringe at the wiring issues to do with power and gas meters, especially if the attached gas regulators have a small leak (most do), where mains powered devices, especially switching power designs, and thirsty ESP32 designs with solar powered rechargeable lithium batteries are left in the sun all day to roast.
Any dentists out there willing to oblige a scan or two? From the top, and side, and rotated side, so the embedded location of the reed switch can be accurately determined? Is it in the centre of the probe, or close to one edge? What are the dimensions? For gas as well as water meter pulse detectors?
Has somebody already done this and my searches are not detailed enough to find it?