I'm working on a small project at my farm to automate vegetable weighting. The system consists of an STM32 MCU, some additional ICs (matrix decoder, parallel-serial converter for some push-buttons etc.), an LCD screen (HD44780 compatible) and finally, an HX711 ADC to interface a load cell. Everything is supplied by a single 12V PSU (switch mode), followed by a 3.3V and 5V regulator module (switch mode). The system was designed on two PCBs. Everything would be installed in a custom-designed, quite large sheet metal housing.

I'd be really glad to have your opinion about two things:

  • The first question is about noise. When I'm using the system on a test bench, the readings of the HX711 are rock stable. The final measurements are converted using a basic linear regression to get a calibrated values in decagrams. I'm using a 5 measurement long buffer for averaging (when successive measurements do not exceed a threshold value), but even without that I get stable values.

    But when I assemble the system in the housing (same PSU, cables, and other components), the variance increases a lot, which means jumps of roughly 5 decagrams. I tried to introduce some noise when using the system on the bench to reproduce the issue (WiFi enabled mobile, high voltage lines etc.) but nothing seems to disturb the measurements there, only when it's outside in the metal housing.

    Has anyone already faced a similar problem? What can be the issue?

    The only visible difference is that on the bench, the load cell is fixed to a plank, but when I assemble the system, it has direct contact with (screwed on) the metal housing. No other electronics is touching the housing, just the body of the load cell (I couldn't find any connection between the body the cell and the wires of it with a multimeter).

    Is it possible that the housing introduces the noise to the load cell or to the ADC? I'll try to do a test measurement with the load cell "floating" on its wire or with another load cell, just to see if it has any effect, but I can visit only next week, so I thought of asking this in order to get some additional ideas.

  • I'm just a hobbyist, and since I mentioned the housing as a possible cause of the noise, I have a question that most likely to be a basic one. Like I mentioned, all electrical components are separated from the housing, only the load cell and the push-buttons are touching it. Since there are some push-buttons, the PCB is (theoretically) protected by a debouncer IC, which has ESD protection.

    But should the housing itself be connected anywhere (eg.: earth) considering the safety of the user, furthermore the safety and proper function of the system?

Thank you very much for your help in advance!

  • 1
    \$\begingroup\$ I’m only able to travel to the location once in every two weeks. This was just on idea of mine (don’t know if it’s even possible) so I thought of asking other possible reasons which I should check next time. \$\endgroup\$ Commented Oct 16, 2020 at 4:38
  • \$\begingroup\$ I have a couple of questions: (1) Can you give us a link to your load cell? I would like to know if it is 50kg, 10kg, 5kg or else? (2) Can you give us a link to you HX711 module? Is is from SparkFun, Amazon, AliExpress or eBay? (3) Do you have two or more HX711 modules to swap/cross check/calibrate accuracy and noise? (4) Even if you have the same module, did compare and contrast Channel A and Channel B, or both channels at the same time? (5) How solid/vibration free is your home/lab bench top setup comparing to the field/farm setup? \$\endgroup\$
    – tlfong01
    Commented Nov 5, 2020 at 3:50
  • \$\begingroup\$ Mihály Tóth, This is a photo of my HX711 test up: i.imgur.com/ncX2QlB.jpg \$\endgroup\$
    – tlfong01
    Commented Nov 5, 2020 at 5:01
  • \$\begingroup\$ #Mihály Tóth, (1) Please let me know the Vcc of the Wheatstone bridge, and Vcc of HX711, and if they are grounds connected together, (2) The lengths of the Wheatstone wiring. This might be important, because the Wheatstone bridge output is in the range of 0V to 3.3mV (if using 3V3), so if the connecting wires are too long or too thin, the output voltage might not be stable. If in the lab you are using short wires, but in the garden very long wires, then the tiny output voltages might be unreliable. \$\endgroup\$
    – tlfong01
    Commented Nov 5, 2020 at 9:05
  • \$\begingroup\$ #Mihály Tóth, my purple HX711 module has a "yellow" pin which is shorted to the black pin, E-. The sparkfun schematic indicates that this yellow pin is "shield". So have you connected this yellow pin to the sheet metal housing? \$\endgroup\$
    – tlfong01
    Commented Nov 5, 2020 at 13:37


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.