Gas Meter Reading Using a Magnetometer

[ThomasDr]

Hello,

i have the same Project, i use a HMC5983 magnetometer.
i'm testing now, but it looks good.

regards
ThomasD

[ThomasDr]

Hello,

here my sketch to form a puls:

/***************************************************************************
  This is a library example for the HMC5883 magnentometer/compass

  Designed specifically to work with the Adafruit HMC5883 Breakout
  http://www.adafruit.com/products/1746
 
  *** You will also need to install the Adafruit_Sensor library! ***

  These displays use I2C to communicate, 2 pins are required to interface.

  Adafruit invests time and resources providing this open source code,
  please support Adafruit andopen-source hardware by purchasing products
  from Adafruit!

  Written by Kevin Townsend for Adafruit Industries with some heading example from
  Love Electronics (loveelectronics.co.uk)
 
 This program is free software: you can redistribute it and/or modify
 it under the terms of the version 3 GNU General Public License as
 published by the Free Software Foundation.
 
 This program is distributed in the hope that it will be useful,
 but WITHOUT ANY WARRANTY; without even the implied warranty of
 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 GNU General Public License for more details.

 You should have received a copy of the GNU General Public License
 along with this program.  If not, see <http://www.gnu.org/licenses/>.

 ***************************************************************************/

#include <Wire.h>
#include <Adafruit_Sensor.h>
#include <Adafruit_HMC5883_U.h>

/* Assign a unique ID to this sensor at the same time */
Adafruit_HMC5883_Unified mag = Adafruit_HMC5883_Unified(12345);

unsigned long t = 0;
unsigned long counter = 0;
bool trigger = 0;
float out = 0.0;

void setup(void) 
{
  Serial.begin(115200);
  //Serial.println("HMC5883 Magnetometer Test"); Serial.println("");
  
  /* Initialise the sensor */
  if(!mag.begin())
  {
    /* There was a problem detecting the HMC5883 ... check your connections */
    Serial.println("Ooops, no HMC5883 detected ... Check your wiring!");
    while(1);
  }
  
  /* Display some basic information on this sensor */
  // displaySensorDetails();
}

void loop(void) 
{
  /* Get a new sensor event */ 
  sensors_event_t event; 
  mag.getEvent(&event);
 
  /* Display the results (magnetic vector values are in micro-Tesla (uT)) */
  out = event.magnetic.y;
  if (out > -0.7 && trigger == 0)
  {
  trigger = 1;
  Serial.println(trigger);
  }
  if (out < -110.5 && trigger == 1)
  {
  trigger = 0;
  counter++;
  Serial.println(trigger);
  }
  
  if (millis() > (t+5000))
   {
   Serial.println(trigger);
   t = millis();
   }
 
  
  delay(250);
}

And here the Output, 1 Puls is 0.001 m3, first at slow flow and then with my max.

regards
ThomasD

[Yveaux]

@ThomasDr do I understand it correctly that you get a full sine swing for each 0.001 m3 of gas usage?

[dpcr]

@Yveaux ! The picture shows the sine wave produced by the HMC5883L. The orange line is the Y axis, which is the one I am using. The large visible sine wave is when the gas is flowing for the furnace. When the gas stops the line is creeping upwards, this is the pilot light on the water heater.

[dpcr]

@Yveaux Not quite. What I did was look at the actual gas meter and determine how many full "cycles" of the sign wave I received for 2 CuFt. What I got was 16 full cycles per 2 CuFt. I then divided that by 2 to get how many cycles per CuFt (8). Then I divided that by the number of divisions per one full cycle (20), 10 from TOP to BOTTOM and 10 from BOTTOM to TOP. This gave me the the CuFt of gas used per division or pulse. I then converted that to litre. This gave me the 0.17698 per litre. Hope this help and I hope my math is correct.

[ThomasDr]

@Yveaux Sorry, it is 1 puls at 0.01 m3
That depends on the gas meter you use, the BK4 gives a magnetic pulse per 0.01 m3.
For this puls you can use a reed contact or a magnetometer.

regards
ThomasD

[Yveaux]

@dpcr Thanks for the nice chart!
It is indeed a clear sine wave. Having the sensor run for a while and checking with the actual meter counts will prove if your math was correct ;-)
You could also use inverse sine to calculate the angle of the sine pulse, to have linear readout and probably even increase resolution.

@ThomasDr I'm currently using a pulse counter (reed contact) to count the pulses of my meter (also 100/m3).
Works well, but the magnetometer has the potential to increase the measurement resolution.

[gohan]

Sorry for the question, but why would you need a magnetometer to have better resolution in the measurements when you only need to count pulses? Given of course that the reed switch provides accurate readings, because if it doesn't then I understand why to use the magnetometer

[Yveaux]

@gohan well, it seems like a magnetometer is able to also give detailed information inbetween the pulses.
For a pulse counter you only see a complete revolution as a single pules; it doesn't tell you how far the revolution is. The magnetometer creates a sine signal, which indicates where you are on this single revolution.

[gohan]

I can see that, but looking at the graph if that climbing line is just the consumption of the pilot light, how would you benefit for this added resolution? I am just referring to the fact that this kind of sensor is used to keep track of gas usage during the day and not a live view of the actual flow even for the small amounts: we are talking 0.01 cubic meter for each pulse and a pilot light usually uses 3 cubic meters per hour, so we end up with a pulse every the 10-15 seconds and to me it seems a good resolution, or maybe I just don't see what are other possible benefits.

[dpcr]

@gohan I had to use a magnetometer because my meter did not have a magnet in it. I tried a reed switch but it never actuated and my meter is outside. So I was stuck using a magnetometer which I mounted in a water proof plastic box. The issue I had was getting the sine wave to create a pulse that was usable for MySensors. I did look at using an inverse sine, or as someone else mentioned, use an exponential moving average of the data but they both failed. When the gas stops flowing, a pilot light as an example as seen in the attached picture, there is little to no gas flowing so the angle doesn't change much. It looks like no gas is flowing when it actually is. I decided to go with dividing the wave or meter movement into separate divisions to create a more accurate way of measuring the gas flow in all flow situations.

I haven't had a chance to test the accuracy of my math against the gas meter but I will at some point. The issue I'm currently having is that the bellows inside of the meter that the magnetometer is reading do not move at the same rate at all times. This causes the FLOW output to change when I know the gas is flowing at a steady rate. The VOLUME doesn't seem to be effected too much. So what I did was to take a moving average of the last three FLOW readings. This helps a little.

The hardware and ideas were done by me but the coding was done by my 24 year old son. I always enjoy a challenge. Please feel free to comment.

[gohan]

Ok, now it makes sense. If you plot the data with the pilot using the gas, don't you still see a sine wave but with, of course, a longer period?

[Yveaux]

@gohan I assume you do. By interpreting the sine and thereby increasing the resolution you might be able to see there's gas flowing, even if the rate is very low.

[gohan]

good luck with the math then, I am calling myself out! :sweat_smile:
Back in high school I was quite good at math, but now... :D

[dpcr]

@gohan Yes you do, but what about the times when there is gas flowing. If I remember correctly there was about 20 or 25 minutes for full sine wave when only the pilot was running. That SEND_FREQUENCY (1500000) was too large when the gas was actually flowing to get an accurate measurement. I just wanted to get an accurate reading of the gas flow at all flow rates. Am I missing something?

Here are some screenshots from Domoticz for the past week:
)

[ThomasDr]

@gohan Hello,
My thought is simpler.
The original sensor is too expensive.
With a Reedcontact I must find the right position.
The magnetometer I could simply attach somewhere in the proximity.

regards
ThomasD

[gohan]

@ThomasDr
I see your point, but you pay the simplicity with a more complex coding ;-)

[dpcr]

@ThomasDr Sorry not to get back to you when I saw your post, your project looks interesting. Does the HMC5983 magnetometer output something similar to the HMC5883L? I just used the HMC5883L magnetometer because I found it useful, I didn't run across it during my research.

[dpcr]

Ran into a problem. The magnetism value changes when the outside air temperature changes which causes many incorrect readings. So my initial sketch that finds the TOP and BOTTOM of the wave is having problems so I'm in the process of trying to find another way to get some useful data out of the magnetometer. Any ideas?

[gohan]

If you find a correlation between temperature and change of magnetic field, maybe you could add a thermistore and use it to compute a correction coefficient.
Or maybe even a second magnetometer with a fixed magnet to use as reference 🤔