HX711 with four load sensors returns strange graph

Projects General Guidance
1

Hi,

I have four load sensors (Link) hooked up to a combinator on a HX711 wheatstone bridge that is connected to a NodeMCU (ESP-12E).

I am using this library for accessing HX711 values.
It should be used for a permanent scale, so I had to figure out the offset and scale factors initially.
Recently I looked at the values history and noticed a strange pattern among the values:

1301×583 44.6 KB

The pattern looks like the value is increasing 5 times and then drops to a lower value.
I could low-pass the value with no trouble, but the pattern makes it look like there is something strange.

I thought already of a wiring problem, but to me the cables look OK.
Can anyone give me a hint of what could cause the strange pattern?

My code:

#include "HX711.h"

// HX711
HX711 hx711(D6, D5);
float weight = 0.0;

void setup()
{
  Serial.begin(9600);
  delay(200);

  // HX711
  hx711.set_scale(-12140.0);
  hx711.set_offset(55000.0);
}

void loop()
{
  hx711.power_up();
  delay(500);
  float w = hx711.get_units(3);
  
  // if value is not ready to be read, retry until succeed
  while (isnan(w)) {
    delay(100);
    w = hx711.get_units(3);
  }
  hx711.power_down();

  // low pass filter the value
  weight = w;

  Serial.print("Weight: ");
  Serial.println(weight);
  
  delay(60000);
}
2

I have a similar setup and haven't seen results like this. I have found that those load cells are quite susceptible to drift with quite small changes in temperature. Have you got a thermostat in the room that is cycling the temperature every 5 - 6 minutes? On mine, as the temperature goes down the weight reading goes up.

I have had more luck with stability by reading the scale until I get a number of consecutive values that are within a given tolerance. This eliminates any spikes that may throw your average out, especially in an average of only three samples.

3

I habe a DHT22 sensor close to the load sensors that I am reading as well (not shown in code above).
Contrary to what I read on load sensors I found out that mine are sensitive to the inverse air humidity. Plotting the graph shows a clear similarity between air humidity and weight. I am trying to work out a formula to correct the value, but it's not good enough yet so I had a deeper look at the raw values.
Interestingly I couldn't find a clear correlation between temperature and weight reading. There are some similarities between the two but the humidity seems to be a bigger player in the equation.

Looking at the load sensors I can see that they have some white glue coating above the three cables that are used to measure the resistance. Maybe they let humidity flow through. Could adding a better (glue) coating maybe help isolating the sensors from humidity?

I think that after having a good base I can continue worling on my formula...

4

Hi,
Welcome to the forum.

Can you please post a copy of your circuit, in CAD or a picture of a hand drawn circuit in jpg, png?
Can you post a picture of your project so we can see you component layout.

Thanks.. Tom... :slight_smile: