HX710 High Precision Voltage Meter

I drew it casually, mainly to verify the HX710, a 24-bit ADC costing 9 cents (5V 10Hz, 17 noise-free bits), to see if it smells good.
I did not use an analog switch to switch the voltage dividing resistor. I have tried CD4066 before, and it works. The impression of leakage is not good. The impedance is relatively low when using the pin directly
at low range, because the stock does not have a large stock resistor. [s]It is like a pointer meter[/s] and it is not unusable.
Then the input part is single-ended input, through TL072 Convert to differential, and then provide low-pass filtering to the ADC (HX710)
voltage reference. It supports three modes. According to the official diagram of HX710, it directly uses the 5V after the voltage regulator tube as the reference. This should be feasible.
In addition, AD586 is provided as the reference (power supply Requires 6-12V) and an external reference, which can be switched through the pin. If not switched, it can be soldered.

There was a problem with the differential input of the first version, and then it was redesigned. The single-ended input was removed and converted to differential. The direct differential input was
then used as a voltage by TL072. There was a deviation of 10.89uV (close to 0V input to the op amp, and the voltage increased during output). 10uV, Tina simulation result),
which will cause the final differential waveform to float upward by 10uV when it is sent to the ADC. I am not sure whether it will cause problems with the final result. I have removed the voltage for the time being.
For the ADC, I did not see the input impedance in the document, but after all 24-bit ADC, the impedance will not be low (guess
the AC part uses single-ended input full-wave rectification. The third version of the PCB does not add DC blocking capacitors. The schematic has been added. Let’s test it again after the board is printed.
The peak value of the AC voltage will be It is relatively high, so half the voltage is used.

The design consideration is mainly that it can be used as a component for computer-side interconnection
and involves common ground issues. I did a lot of simulations and finally determined that this third version
is suitable for computer-side capture and recording of voltage values ​​and waveforms. , if the training is good, it should be able to have an accuracy of 5 to 5 and a half digits. This depends on the efforts of the big guys.