I wish to measure weights up to 80 grams with resolution of around 0.1g or less (preferably less).
I had once used the HX711 with a 1kg load cell and the baseline was not stable. Perhaps it was vibration in the table or something.
I need to know:
As far as I know it should be possible in theory.
0.1 gr is 1/1000th of full scale.
Better would become serious harder.
In practice you need a very stable setup, no vibrations no wind, no heat onvection etc.
Make multiple measurements and take the average or better the median.
Other problems to tackle are temperature compensation and calibration of the oad cell used.
Please share the datasheet of the loadcell, it should contain all details.
I'm sure you can get 0.1g resolution from 100g cell with hx711. But stable readings depend on mechanical construction, wiring and shielding as well. Also digital filtering/averaging has huge effect (if speed is not critical). If you want inexpensive better alternative, consider this as well:
The most important thing that you have not considered is the load cell. You will not get 0.1g accuracy with a 1000g load cell.
You need one with a max load rating just a little higher than the max load you want to measure.
Sorry, didn't mention, I am using now with a 100g load cell.
What is the mV/V specification?
Can you please, post the picture of the load cell?
That resolution is possible but accurate to that is another story altogether. Post the data sheet.
Here is the datasheet
TAL221.pdf (282.1 KB)
@jim-p from the datasheet I think mv/V spec is 0.6 ± 0.15
If I make a conservative estimation of he HX711 noise free bits to be 16, then your resolution will be ~0.026grams. The load cell has a combined error of 0.05%FS = 0.05grams, so resolution and accuracy of 0.1grams is achievable.
You do need to be careful with the wiring so it not susceptible to noise pickup. The wires between the loadcell and amplifier should be as short as possible.
Do you know what is: mV/V?
It is actually: mV/V of excitation.
Your posted datasheet has not said the value of the excitation voltage. So, I cannot calculate how much voltage your load cell will produce under full load?
If the excitation voltage is: 5V, your load cell will produce only 3 mV (0.6 mV x 5 = 3.0 mV) when full load (100 g) is applied. Will your 5 V operated signed 24-bit HX711 ADC Module handle this signal?
If your answer is yes, then tell me how many counts there will be out of the ADC?
What are those 16 bits?
<= 6 VDC
I have found it now. Thank yoou very much.
That are the 16 most significant bits of the 24 the HX711 provides.
Excitation voltage is less than 5V, maybe 4.3V. HX has gain 128, so even at effective 16 bit resolution it would be ~0.02g.
More importantly they are noise free.
How, it is: ~0.02 g?
If lower 7-bits (b6 - b0) are considered noise, then the resolution of the system will be the positional weight (in gm) of bit-7.
At 4.3V excitation, gain 128 and 100 gm full load applied, the ADC will produce:
==> ((8388607)/4.3) x (0.0006 x 4.3 x 128) counts
==> 644245 counts
Resolution: (100/644245) x (27) = 0.0198 ~= 0.02 gm (derived and verified)
Congratulation to @jim-p!
Now, question is: how accurate (quantitatvely) would be the response of this measuring system?
Now to answer your questions.
1 Is the HX711 module is up to the job of measuring sub gram weights with precision or if i need a custom pcb with a good instrumentation amplifier and layout.
See post #4
2 How important is shielding in this.
In a normal laboratory environment shielding should not be necessary if wires between the load cell and HX711 are kept short and the overall system design is neat and clean with properly soldered connections.
3 How much needs to be worried about vibration.
Why do you ask?
It's a sum of dozen of factors like quality of the cell, mechanical assembly, calibration approach (output is rarely perfectly linear), wiring, creep, temperature, how you use it etc.
