Measuring weights with load cells

Projects General Guidance
1

I have a project where I need to measure weights that ranges from 0 to 200 grams. Specifically, I need to measure weight changes, such as when someone removes items or add items to a scale. Most importantly, the scale need to measure/detect weight changes that happens quickly, such as when someone add items to the scale in quick succession (around 0.5 sec), and I'd like the weight measurement to be stored in memory that can identify when such event happens.

Currently, my peer already has a prototype developed, using a strain gauge load cell like this one:

He also hooked it up to a Nuvoton NAU7802 24-Bit Dual-Channel ADC For Bridge Sensors (2.7 - 5.5 V input)

https://www.nuvoton.com/resource-files/NAU7802%20Data%20Sheet%20V1.7.pdf

I cannot identify the board he used, but it works with the Arduino IDE and the microprocessor seems to be L432KC by STMicroelectronics.

He powered the whole thing with a CR123A battery (3V).

Currently, he has it set to "wake up" at fixed time intervals and record the weight measurement for a duration (say x measurement for y seconds) and then goes to sleep again. I asked him if it's possible to have the system be awake all the time so that when there are quick/sudden weight changes it can be recorded. In the current setup, if someone adds five 2-gram items to the scale in rapid succession during the time that it is asleep, we'd only see a net change of 10 grams the next time the scale wakes up. I'd like to capture the five 2-gram weight changes in the data.

He tells me it's not possible because those load cells can only output a "stable" measurement every 0.8 second or so if the weight changes happen faster than that, it cannot be captured. I wasn't able to find any information on this from the datasheets. But 0.8 sec might be sufficient for my application.

Most importantly, he tells me that if the system stays "on" the whole time, the battery will quickly be drained. But I wonder if it's possible to power the whole thing using a USB like wall plug. For instance, I know that for the Arduino Uno you can power it using a 5V USB cable that can be plugged into the wall outlet. Even if the actual microprocess need something like 3V, it's easy to get 3V from 5V right? I only know of linear voltage regulator, if there are better ways please let me know. Either way, I just don't think we should be limited to using a battery for an indoor fixed application like this.

I cannot think of a reason why 5V USB cannot be used to power something like this. I haven't dived into how the load cell works or how its "4 wires" are connected to the rest of the system.

Please advise on the feasibility of my idea. Because I'd like to develop something on my own or advise him better if possible. Thanks

2

Quite feasible. Look for similar projects.

3

You don't need to bother about any 3 volt. 3.3 volt to be precise. Power the UNO by 5 volt USB!

4

After doing some learning, I understand the current drain of the load cell is given by the excitation voltage divided by the input resistance of the load cell. The excitation voltage is provided by the ADC since most load cells are connected to an ADC. As I understand, the excitation voltage is always "on" and are not in pulses. The frequency of the data capture is based on the sampling rate of the ADC. So if I need 10 readings per second, I can just set the sampling rate (which should be programmable) to be 10 SPS. Is that correct?
thanks

5

I'd guess it's going to be "difficult" to get accurate or precise readings at low weights.

You get tiny voltages from a load cell and whenever you amplify there is noise. It's done "everyday" but usually not by hobbyists. Maybe the amplifier board you have is low-noise and maybe the load cell won't pick-up much noise and maybe it will be OK...

A regular ADC only has an analog input but if your board is designed for a load cell it might have a bridge amplifier and supply the voltage, as well as have an ADC.

6

I have had luck in the past with the use of an Hx711 amplifier to read the thrust of rocket motors to with in a very narrow time window with millisecond resolution.
You need to make sure your stand is very rigid. With your time window being that small you need to make sure you’re not measuring the force of you putting it down.
It wouldn’t be too difficult to have the measurement start after the weight is > a small value that is above the loadcell’s sensitivity then start a small timer. The length of that timer should be long enough to make sure the system isn’t vibrating. Use the max mass you’re expecting for a test and use that time. You can wire an LED for when the time is over.
The 5V output should be plenty for that amplifier as they run 3.7-5V. Your project is perfectly feasible.

7

It depends on the ADC. The sample rate of the commonly-used hobbyist ADC, the HX711, is not programmable (by software), but it can be set to either 10 SPS or 80 SPS, sometimes via a solder jumper or a PCB trace that you cut.

The commonly-used hobbyist HX711 supplies the bridge excitation voltage.

The HX711 excitation voltage can be turned on and off by software (see the datasheet regarding the "power down" mode).
.
.
.
PS: if your microprocessor uses 3.3v, be sure to get an ADC that is compatible. For example, most HX711 breakouts are intended to run on 5v, not 3.3. If you use the HX711, get one from Sparkfun - it's designed for either 3.3v or 5v operation.