Calibrating 500 kg load cell with an ESP32 and HX711 without putting actual 500 kg on it. 3 V excitation

Question

Edit and update !

Considering the following data sheet ( calibration data from factory) what is the correct math to calculate the weight?

500 Kg = Excitation * rated output. ?

Or

Weight = 41.01 * loadcell reading

It’s the most linear approach and I don’t think it will be accurate.

How I can use all these data in a formula so I don’t do the point by point calibration by myself?

Thanks

<<<<_____ the Old question ___>>>> I am making a 500 kg scale using an ESP32 and HX711. It is battery powered and uses 3 V excitation.

In the calibration process, I increase the weight by 50 kg steps and read the HX711 output and then use regression to find the fit formula.

It does work and there is no problem beside normal tolerance and issues with hysteresis.

My question is: Is it possible to calibrate the scale without adding actual 500 kg on it? Is there any way to just use the load cell datasheet from factory and use that in a formula to find the fit formula parameters? Does anyone know if load cell manufacturers do this step and provide the fit formula parameters for the customer?

---

Thank you all for your responses, they are all useful.

It sounds like I do need to put 500 kg to know the load cell characteristics. So, no cheating.

I reach to this solution:

1. I will use @lorenzo suggested method to avoid actual 500 kg
2. I will use @Tony and @Drew suggestion but with a slight variation. I will use an electric scissor jack.

I will use 10 kg weight as a reference and record the load cell output, then I will use the jack to increase the pressure until I read the 20 kg output until I reach 500 kg. It will get interesting now. Then I will decrease the weight step by step and record the reading.

In this way I will have two graphs and can compensate the hysteresis better.

What do you think?

Answer 1

Actually… well, no you can't since for a full test you need to see if it's linear at all the points in the scale. And 500kg are not too much, consider that load cells are used to weight whole silos of product for several tens of tons.

However you can do it without a calibrated 500kg reference using the weight sostitution method (it's slightly less effective but for periodical performance testing is fine). You can find the exact procedure online (NIST has the official say on it) but it works more or less like this:

• You put your calibrated standard (say 50kg) and weigh it. Record the number (of course, you are doing it for tracing the response)

• Remove the standard and put bulk material (water, sand, whatever is practical) on the cell until you have the same exact reading so you are sure you have 50kg on the cell

• Put the calibrated standard and note down the 100kg value

• Repeat until end of range

Due to hysteresis and the accumulated error in the substitution is slighly less accurate but a lot handier than hauling a 500kg standard around.

Answer 2

When I was working for a load cell manufacturing company (decades ago), we used to take 4 readings with calibrated weights and then fit a curve through them.

Not all of them were fully linear. Since we were manufacturing them for ourselves, we would put the parameters on the EEPROM on the weighing arms.