I am using a LabJack U12 to measure a DC voltage of approximately 24 V versus ground. I do this by reading ~1/3 of the voltage (~8 V, resistor divider) using single ended analog input. The absolute accuracy for this reading is specified as .2% of full scale, therefore .02 V so that the absolute accuracy of the 24 V measurement is ~.06V (= 24 V +- 0.06V). I would like to measure this 24 V value more accurately and I am imagining subtracting a fixed voltage of about 20 V and measuring the difference of about 4 V within an accuracy of .02 V which would then be the absolute accuracy of the 24 V measurement (= 24 V +- .02 V) which is satisfactory for me.
Can you suggest a circuit would produce a very stable 20 V that I could subtract from my 24 V signal?
Is there a better way for me to get the accuracy I need?