Ah, ok, I see where you are going with this now.
I'm not familiar with one wire products at all but I did glance over the application note #125 for monitoring voltages above ten volts, and the specs on the DS2438 analog to digital converter.
Again, I did this very quickly
.
What you are really doing is changing the ratio of the input voltage to the voltage measured by the A-D converter with R5 and R6 in order to get a higher measurable range.
The app note suggests calculating R5 (R6 needs to be 10K) as follows:
R5 = R6 * (Vmax - Vvad)/Vvad
So if I am correct in the assumptions for voltages (your Max is 24 and the A-D Max is 10):
R5 = 10,000 (24 - 10)/10 = 14,000 or 14 Kohms (may have to series available resistors to get this value)
The actual voltage can then be calculated from the measured via:
Vactual = Vmeasured * (R5 + R6)/R6
Vactual = Vmeasured * (14,000 + 10,000)/10,000
Vactual = Vmeasured * 2.4
To determine accuracy the DS2438 is a 10-Bit Analog to Digital converter. If you looked at the Analog to Digital Converter Guide you would see ten bits is "2 to the 10th" power or 1024. This means the "measured" voltage (ten volts max) has a bit accuracy of 10/1024 or .01 volts.
Remember that this is the accuracy is for the voltage range into the A-D input, not the measured range so you have to multiply the .01 by the ratio setup with your resistors above, which was 2.4.
So I'm calculating the accuracy for 24 volt measurement to be .01 * 2.4 = .024 volts
I'd really appreciate someone else looking over these numbers since, as I stated above, I'm not familiar with one-wire products.