Readout of domestic water meter using a coil and reading the LC loop in the meter

Thread Starter

mar10

Joined Mar 23, 2019
67
At home I have a water meter that has an LC attached to the rotating dial that measures the water.
The official detection circuit is using a coil to pick up the LC and check its resonance. Thus it is able to count the rotations.
I would like to achieve the same, considering the 'easy way' of a proximity detector does not work.

The meter can be found at honeywell V200P and it lists the detection circuit as the falcon PR6
This states:
"
The mode of operation relies on the principle of electrical oscillating circuit. This information displays the number of electrical oscillations that exceeds a fixed treshold value. Once the rotating metallized half disc of the water meter register is located under one of the 3 coils of
the Falcon module, the vibration is dampened. As a result of the dampening effect, only a small number of amplitudes exceeds the fixed treshold value. This change is measured and processed by a processor.
"
This model does not use a metal disk, but the coil and assumed capacitor at the back

OBJECTIVE: to create a circuit and use an ESP8266 (or arduino, 3V3 if possible) to achieve the same effect and report to upstream systems.

Idea so far: connect a coil to an oscillator based on a transistor and trigger the interrupt of the MicroProcessor
Closest circuit found till now: OPAMP LC
I hope you can help me?
I can calculate stuff and know a bit but want to get some direction before spending days in the wrong direction.

Screenshot 2021-07-19 at 18.35.47.png
PS. My first post on the forum, so bear with me
 

Thread Starter

mar10

Joined Mar 23, 2019
67
I have been trying some stuff and bumped into the arduino metal detector I built the colpitts oscillator and used two different coils.
There is a coil without a core and quite wide and is 2mH. It gives a stable sine of 43kHz combined with 15nF Cs.
When I move a metal object inside the frequency drops a little and the amplitude becomes less. When I move a ferrite core inside, the frequency drops to 33kHz and the amplitude stays about the same.
Another coil is much thinner and is 0.23mH. Same story here about the ferrite core dropping the frequency from 200 kHz to 100 kHz.
But to the subject of interest, moving any object near the coils doesn't seem to do anything and that is what I intended to happen.

My guess is that I need to find the resonance frequency of the watermeter coil/condensator combo and make the frequency of the oscillator the same in order to detect anything. I also guess that using a coil without core will be better for this purpose.

If anyone could confirm me in my guesses, that would be great. I'll tinker on anyway.Screenshot 2021-07-24 at 15.10.09.png
 

Thread Starter

mar10

Joined Mar 23, 2019
67
Results at 207kHz are zero. I created an LC set with 1.94mH and 305pF which is supposed to resonate at that frequency.
I used the big coil for that. The big coil can fit around the small one, but there was no noticeable effect to the frequency nor amplitude.

Still interested in advice so I am not pursuing the wrong approach.

PS. I have a component tester, which allows me to tune the components based on real values, not just guesses
 

Thread Starter

mar10

Joined Mar 23, 2019
67
One more update and probably giving up/waiting for others.
In the previous update I mentioned that there was zero effect. This turned out not to be correct. I found that the frequency moved up a tiny little bit, like 1%. My frequency counter had a 4 kHz granularity but in the scope mode I could see the wave creep ever so slightly.

Therefor, I think that what we are facing here looks a lot like RF-ID readout, and from what I read there, we are facing much higher frequencies. The flat printed coil could be estimated at 75nH and using a 1.8nF cap, we could resonated at the 13.56 MHz level used for RF-ID. It is unlikely that the 125kHz mode of RF-ID would be feasible with such a small coil since it would require a 22microF cap. But then, they could have opted for an altogether different frequency although that could be more difficult considering this would not be an ISM band.

These assumptions are all I will put in here for now.
The official falcon reader easily costs 70 EUR so this is a rewarding hack if we can get away with it.
If others pick up on this subject, it would be cool to find a solution.
And maybe an informed statement as what is supposed to be the official resonating frequency of this loop?

Waiting for a reaction,
M10
 

michael8

Joined Jan 11, 2015
233
Assuming there is an LC circuit on the dial in the meter, perhaps you
don't need to know the frequency. If you sent a fast short pulse the
LC should ring (if you don't load it down after your pules). You should
then be able to see the ringing in your coil.

This would take a coil which saw a high impedance when not being
driven.
 

Thread Starter

mar10

Joined Mar 23, 2019
67
Thanks for this suggestion.
Would you have an example article that describes this idea?
The three last sentences are not something I know to interpret.
What is 'to ring' and how to see the 'ringing in the coil'
And what condition is needed such that the 'coil sees a high impedance when not being driven'.
 

LesJones

Joined Jan 8, 2017
3,571
I think what is required is something similar to a GDO (Grid dip oscillator now just called a dip oscillator.) Looking at the inducter on the dial I suspect the resonant frequency will be quite high. (I would guess between 30 and 100 Mhz) I think you need to find someone with a dip oscillator that covers that frequency range. Once you know the frequency build a dip oscillator for that frequency with a level detector on the output.

Les.
 

michael8

Joined Jan 11, 2015
233
I've done some internet searching but am having problems not knowning
what water meter you have exactly. I thinking that the "PR6" is a
pointer that the PR6 reader goes there not that the meter type is PR6.

I'm also wondering where the knowledge that the coil on the meter dial
is part of an LC circuit. Perhaps there is just a resistor across it?
This resistor would represent a load to a magnetic field in that space
and would be absent as the dial rotated around. It wouldn't have any
frequency response other than that required by the small coil size and
magnetic coupling to the external to the meter probing coil.

If it's not an LC it won't ring at all (or not very well). However an
external LC would ring less (or damp down faster) when coupled with the
resistor loaded coil.

For a probe/sensor I'd try a coil on the same or 1/2 the diameter of
the coil in the meter with similar number of turns (or parhaps more?).

A high impedance sensor like a 10 M ohm oscilloscope input or a FET
follower input might be good. For the driver you need a high output
impedance pulse, possibly as simple as a microprocessor output fed
through a resistor of say 10k to 100k.

Perhaps something like this:
water1.png
 

Attachments

Thread Starter

mar10

Joined Mar 23, 2019
67
Thanks for the replies. I am in the process of building a GDO so this can take a while.
I do have all the components in home (squirrelling for the last 50 years ;-)
Main issue is that my scope/freq-counter is not going to help for these frequencies
so the learning curve is that much steeper.
Stay Tuned (pun intended)
Screenshot 2021-07-27 at 18.06.59.png
 

michael8

Joined Jan 11, 2015
233
A small amount more information on the V200 water meter. It appears to come with two different inductive coupling
options. Either a stainless steel target or a "resonant pointer". The PR6 may target the steel target.

https://assets.publishing.service.g...data/file/783856/UK_0126_0012_Revision_15.pdf

see page 5 section 4.1 for a description and see page 23 for the "Resonant Pointer" notation with picture

I modified my simulation to let the target resonate. The diode must be small signal schottky at these frequencys,
a simulation with a 1n4148 shows current going both ways through the diode.

Each plot has two traces, one for each value of coupling constant. The upper plot is the current in the resonant
coil (in the meter). The lower plot is the voltage at the "scope" point. The common grounds between the
sections is an artifact of the simulation (it's happier without floating circuit elements).

water2.png
 

Attachments

Thread Starter

mar10

Joined Mar 23, 2019
67
Thanks Michael8 for that reference and your idea, but for the moment I will try to make the GDO work.
I might still come back to your idea so don't feel I ignore you, OK ?

As for my GDO, I have made this setup with a slightly different transistor and I have simulated it in EasyEDA. The strange thing is that the simulation shows that different values for C1/C2 keep the Ammeter at the same current, BUT my real life setup does not.
There, when lowering C1/C2, the current goes up in what looks like a linear fashion.
I also deviated from the original design in removing the source resistor MOD IN but this should be OK,right?
The reason I used these components is that this is a PCB from the december 1985 Elektuur !
Does anyone have an idea here? Do I need a ferrite bead on the ammeter e.g.?
Screenshot 2021-07-27 at 21.29.22.pngThe C1/C2 can range between 20-550pF

As an alternative: I am in the Netherlands in the Utrecht region. Does anyone have a real GDO I could use?
Or even more easy, you might have a water meter like this in your house already. Vitens is installing about 2 million of these.
 

michael8

Joined Jan 11, 2015
233
What test equipment do you have? DMM? Oscilloscope? Frequency counter?

On the GDO: At RF, layout and what's near the circuit is really a part of the circuit.

I think the source resistor is at least part of the transistor biasing and might be part
of the RF circuit too (rather an RF block, so an RF choke and bias resistor in one).

I thought that dual-gate mosfets like the BF990 were usually biased with 0 volts on G1 and 4 volts on G2
(or something like that). They are similar to two FETs in series. Perhaps the biasing could be
different for an oscillator.

I'd try making an RF sensor circuit , see: http://n5ese.com/rfprobe1.htm

Then, perhaps you would be able to tell if your GDO was oscillating

Circuit simulation programs are interesting, but it's necessary to know which parasitic components need to be
included to match reality. Every component has R, C and L and sometimes all have an effect. At different
frequencies a capacitor might be more of an inductor.

Many GDO schematics (but layout is bound to be a large part of the design): many GDO schematics

A GDO using a dual-gate MOSFET one GDO design
 

Thread Starter

mar10

Joined Mar 23, 2019
67
michael8, that is really a great find!

I will reconstruct my GDO with the right JFET BF245A and use the schema below. The coils have been made this evening and the values are the measured values. And at these frequencies, my scope/freq-counter (Gabotronics XprotoLAB) will still be of use.
The GDO should swipe 135kHz-650kHz and I suspect then it will be a matter of building your ringing proposal using the concept of the patent text. But let me first build this GDO. This will likely take till the weekend.
Stay Tuned
Screenshot 2021-07-28 at 22.15.33.png
 
Last edited:

Thread Starter

mar10

Joined Mar 23, 2019
67
I've built the schema of the previous post to the letter, but while it behaves better, it is not detecting watermeter LC.
The meter can be set at zero (= middle scale) well and when moving from 550 to 25 pF it moves like 1/4 of a scale gradually to the left. This is not ideal, but is OK for this experiment.

The test LC is 80uH/15nF and it is clearly detected because when it is inside the GDO coil and tuning, at the resonance point, the meter goes way out the other side of the gradual move, to the right.

The scope/frequency counter indicates it is at 136kHz and this meets the expected 145kHz from calculation.

With this coil and the big variable condensator bank, I seem to tune from 90kHz to 180kHz (so much for the simulator prediction).
So, as expected, I did not detect anything on the water meter yet, which is supposed to be around 300kHz.
BUT at least I have got my GDO to work! Yeah!
Let me change to the smaller condensator bank...
More tomorrow,
Stay Tuned
Screenshot 2021-07-29 at 23.07.31.png
 

Thread Starter

mar10

Joined Mar 23, 2019
67
OK, some advice on these coils
I thought to be smart in order to wind two identical coils by taking the 100m of AWG31 (0.25mm with enamel)and double it to 50m parallel wires and put it on a spool for dispension. Then I took the two wires that are in parallel and started winding them together on a coil former. Like this the two inductances are the same value, 550uF. But my oscillator now doesn't get any higher than 165kHz!
I suspect that the mutual capacitance between the two coils which is 1n6F (!) spoils the setup.
This in hindsight would also apply to the spool I made first where the first coil is on the inside and the second on the outside of the same former forming one body.

Considering the creation of coils is a bit of a hassle, can someone confirm that joined bodies of coils are no good for the oscillator at hand?

I guess I will start over again and form two separate bodies meanwhile.
Screenshot 2021-07-31 at 10.36.14.png
 

michael8

Joined Jan 11, 2015
233
Many things matter...

I'd try to duplicate the meter situation as much as possible.

For the "pointer" I'd make a flat "coil". Just now I made
a very rough calculation and the coil might be around 114mm
total length -- I'd try to make a 3 to 4 turn flat coil
with something like #34 wire (close to same cross section area)
and a 1uF capacitor (likely needs a SMD capacitor, only needs
to be 3 to 6 volts..).

The driving/sensing coil can be thicker wire (for higher Q) and more
turns to allow a smaller capacitor. But keep it small, say 2X the turns
(about 4 times the inductance) and 1/4 the capacitor (.25uF).

What test equipment do you have?
 

michael8

Joined Jan 11, 2015
233
wild calculation of pointer coil resistance (very approx):
tracks from patent page 5:
tracks are preferably between 0.2mm and 0.4mm, more preferably 0.3 mm wide
coil 50um to 100um preferably 70um thick

looking at rough measurments of "coil": approx 3 turns
outside 13 mm x 10? mm -> length 2*13+2*10 -> 46mm
inside 8 mm x 7 mm -> length 2*8+2*7 -> 30mm
guess at middle turn as 1/2 between:
(46+30)/2. -> 38mm

total length 46+30+38 -> 114mm

cross section area 0.3mm x 70um -> 0.3e-3*70e-6 -> 2.1e-8m**2
radius of wire with same area:
a=pi*(r**2) --> r = sqrt(a/pi)
sqrt(2.1e-8/pi) -> 8.175883811466259e-05 m
diameter = 2*r -> 0.000163m -> 0.163mm

copper wire table #34 is .160mm diameter & 855.752 ohms/km
.855752 ohms/meter
.855752 * .114m -> .097 ohms for 114mm trace
 
Top