Load cell based fuel gauge

Discussion in 'Burners and their construction' started by Mark's castings, Jun 30, 2020.

  1. This is based around those cheap ebay half load cells and a dedicated load cell digital converter IC with a serial peripheral interface output. Each half cell will measure up to 50Kg so with four of them it'll measure up to 200Kg. The load cells don't have a flat base so I have to machine a steel cradle to hold each half cell and attach to the aluminium base. I started to machine relatively accurate bases before I found the load cells vary in thickness: on assembly I can use epoxy to bed the upper mechanical contact pads at the same time to get uniform contact on all four sensors. So the plan is to hook it up to an ESP8266 or ESP32 wifi microprocessor module which will interrogate the load cell interface and convert the weight readings over time into fuel consumption figures. An old android smartphone can be the wifi interface and display a webpage with the data.

    So once the unit is calibrated, I can zero the unit with an empty fuel container then place a full container with a known volume of fuel and the ESP8266 board can calculate the fuel density. After that, it's a matter of taking weights at set intervals and calculating the amount of fuel used over that interval as well as fuel remaining and consumption rate per hour. The load cell measures 35cm by 40cm or 14" by 16" in 6mm aluminium tread plate.

    load cell assy 1.JPG

    load cell assy 2.JPG

    load cell assy 3.JPG
     
    Last edited: Jun 30, 2020
    Melterskelter likes this.
  2. Melterskelter

    Melterskelter Gold Banner Member

    Dang it, Mark, there you go educating me again. I did not know that those ESP board existed. Very interesting. I watched a couple videos on their use. Incredible the tools at our disposal. I have a feeling that load cell will get a lot of use and not just for fuel metering.

    Denis
     
  3. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    So what is the accuracy and resolution of the load cells, or moreover, what's the smallest fuel flow rate you hope to be able to infer/measure? Any ability to further filter/process the signal? Depending on the answers I may hear footsteps!

    Best,
    Kelly
     
  4. If I can get litres per hour consumption and remaining litres data to 1 to 2 decimal places I'll be very happy. It can run either 10 samples per second or 80 samples per second which would allow for some signal averaging when updating once per second.

    The board and load cells cost AUD $7 so that should give you an idea of quality, the load cells seem to be graded in matched sets with identical hand markings. They are intended for bathroom scales I think. The HX711 interface has 24 bit resolution with 100db common mode noise rejection and 100db power supply noise rejection:

    https://cdn.sparkfun.com/datasheets/Sensors/ForceFlex/hx711_english.pdf
     
    Last edited: Jun 30, 2020
  5. Melterskelter

    Melterskelter Gold Banner Member

    Your question is key, Kelly. In practice, very fine resolution will be needed. I find that adjustments of .01 L/m +/- do make a noticeable performance difference in my furnace. So, to detect that small an increment in a minute sampling time would require .01/25 or .04% precision. Even greater precision to give real-time readings would be nice as tweaking fuel flow by .01 to .02L/m increments and simultaneously observing exhaust characteristics is what I find myself doing. The usual range of my adjustments is between .16 and .20L/m flow rate with lower flows seemingly adequate very early in the melt when the metal is cold and a very high delta between flame temp and metal temp is present almost regardless of optimization of flame temp. Later in the melt hitting the sweet spot seems more important.

    Denis
     
  6. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    Just a matter of how accurate and timely the measurement can be. If you have a heavy tank of fuel and very short sample span you need very high accuracy, resolution, and an acoustically quiet platform. If you only sample once per hour and are satisfied with knowing average consumption rate over that time a bath room scale will do.;)

    Best,
    K
     
  7. So 200 kilograms divided by 2 to the power of 24 gives a theoretical resolution of 0.01192 grams so it would need to be scaled back a fair bit from that, say a factor of 80 to read in gram increments. That stiff old bit of garden hose fuel line I use will transmit a lot of pump vibration back to the tank.
     
  8. Al2O3

    Al2O3 Administrator Staff Member Banner Member

    What is the expected range of fuel mass flow rate to be measured?

    Best,
    Kelly
     
  9. I've been burning up to seven US gallons per hour of kerosene which has roughly 30% less energy than diesel, so I'd expect consumption in the range of 3-7 gallons per hour. That would equate to very roughly 12-28 kilograms of fuel per hour. So far the whole project has cost AUD $30 for all the electronics/sensors and aluminium sheet so even if there's some technical issue such as excessive electrical noise or mechanical vibration or poor quality load cells it'll still be worth evaluating that approach.
     
  10. I missed your post late last night: they are quite a useful little device, I've been using the ESP8266 nodemcu boards and programming with the free Arduino environment so far. The more powerful dual processor ESP32 look pretty good: I bought some with 16Mb flash memory instead of the usual 4Mb units and hope to data log thermocouple temp readings with them. Getting the thermocouple to work was relatively easy, the hard part is dynamically updating a web page with a graph of readings and I'm still trying to work that out.
     
    Last edited: Jun 30, 2020
  11. I spent an hour or so soldering 38 wires to the fine pitch contacts of the ESP32 WROOM32S module last night with it's 50 thou contact spacing. Some common 0.1" header pins easily solders to every second contact and some old Tandy wire wrap wire from the 1980's solders the remaining contacts to the board in staggered form. The ESP32 modules also support Bluetooth comms and I think I'll go down that route first as I can then use a generic Bluetooth serial terminal program to talk to the module with text command strings like "TARE" and "ZERO" etc. while receiving weights and fuel consumption in return. I found this Bluetooth specific serial terminal program for an android device and it runs fine on my old mobile phone: https://apkpure.com/serial-bluetooth-terminal/de.kai_morich.serial_bluetooth_terminal

    The ESP32 will run from 2.2 volts to 3.6 volts so I plan to run it off three 4500 mAH C size Nimh cells as I have a whole stack of those lying around unused, it should give a quiet noise free power supply for the 280 mA current spikes the module uses in wifi mode.

    Wroom 32 1.JPG Wroom 32 4.JPG
     
  12. Barrybooth

    Barrybooth Silver

    Hello Mark, please excuse my naivety but I'm trying to understand what the purpose is for your building the digital scale. It's nicely done for sure but it seems to me that you could do the same thing by weighting the propane tank before and after the burn also keeping track of the total burn time. With the installation of a cheap flow meter in the gas line you could precisely track the fuel consumption. Again please excuse me for not fully understanding what you're trying to achieve. Barry.
     
  13. Jason

    Jason Gold

    WTF does thing do? Is it a scale?:confused:
     
  14. rocco

    rocco Silver

    The items in his first post are load cells, which are basically the sensor components of a weigh scale, the item in the most recent post is a wifi capable microcontroller,it can take the data from the load cells, do calculations on it and output the results in a format of your choosing and because it's wifi capable it can send the output to any other wifi capable device or the web and if you want, you can create a complete user interface on a webpage.
     
  15. Barrybooth

    Barrybooth Silver

    I'm also confused Jason, seems a bit, just a bit of overkill. But hey, to each their own, I'm sure he has a purpose for it tucked away in the back of his mind, I wish him good luck with it...
     
    Last edited: Jul 1, 2020
  16. Hi Guys, it's simple: I run an oil fired furnace and I want to measure fuel consumption in litres per hour and display the remaining fuel in litres....If I have measurements of weight loss over time and the density of the fuel, I can calculate and display the required information. Second to second fuel consumption figures will let you easily hit a consistent mixture setting each time you run the furnace.

    It's an alternative approach to using a small flow sensor to measure fuel use.
     
    Last edited: Jul 1, 2020
  17. Barrybooth

    Barrybooth Silver

    Thank you Mark for taking the time to clarify what your trying to achieve! Good luck with your project...Barry
     
    Last edited: Jul 1, 2020
  18. OMM

    OMM Silver

    I think this is a very cool project! I would love to have a weight scale that displays weight rate of change per minute.
     
  19. rocco

    rocco Silver

    I agree. I've played around with microcontrollers on and off for the last 15 or so years and I'm really looking forward to seeing where this project goes.
     
  20. Jason

    Jason Gold

Share This Page