I have a project where I use 4 strain gauges in full bridge configuration with a hx711 module and an Arduino mini pro.
How can I estimate the smallest voltage I can measure with a given adc from its datasheet, in this case the hx711 module?
6
-
\$\begingroup\$ Full scale voltage range divided by 2^number of bits \$\endgroup\$– DKNguyenCommented Dec 24, 2019 at 3:23
-
\$\begingroup\$ Is this also true if I want to measure small voltages, like 0,05 mV or below? It might lead to another question but is hx711 capable of measuring such low signals? According to your formula the answer is yes. \$\endgroup\$– Hegedus.CsCommented Dec 24, 2019 at 3:25
-
\$\begingroup\$ Or noise, whichever is higher. \$\endgroup\$– DKNguyenCommented Dec 24, 2019 at 3:27
-
\$\begingroup\$ So basically I can only be certain with measurements with different loads? \$\endgroup\$– Hegedus.CsCommented Dec 24, 2019 at 3:35
-
\$\begingroup\$ Honestly, it's difficult to say. But if multiple readings that should read the same are close together you have higher confidence. Filtering is a thing. Getting a 24 bit ADC to actually work at its full potential is an art. \$\endgroup\$– DKNguyenCommented Dec 24, 2019 at 3:38
|
Show 1 more comment
3 Answers
\$\begingroup\$
\$\endgroup\$
$$ \frac {V_{in-max} - V_{in-min}}{ 2^{bits}} $$
or your noise voltage, whichever is higher.
If multiple readings that should read the same are close together you have higher confidence. You will likely need filtering of some sort.
Actually getting a 24-bit ADC to work to its full potential and output with a resolution of 24-bits is as an art.
\$\begingroup\$
\$\endgroup\$
Some ADCs are nice enough to define the number of effective bits they have, which is the number of bits that are above their noise floor. You'll often find for instance that a 16 bit ADC only has 14.5 or 15 effective bits.
Yours doesn't, but it does give an input noise of 50nV RMS (or about 18 effective bits), so values smaller than that will not be measureable without at least averaging. 50uV should be measureable, at least assuming you obtain the ideal performance listed in the spreadsheet.
answered Dec 24, 2019 at 5:36
\$\begingroup\$
\$\endgroup\$
With over a 70 dB dynamic range , it doesn’t matter what the theoretical resolution is, it will be limited by your ability to manage EMI to achieve a 70dB SNR full scale. That is not possible with twisted pairs due to the imbalance with stray noise.
