Ah, so your interpretation of Table A-14's "Total input capacitance" is that's it's the ADC capacitance, not the total input the ADC can handle. Such that the pin has 10pF and the sample and hold cap is 12pF. My initial interpretation was that a high capacitance would cause a large inrush of current to the sample and hold cap, which would couple to other channels. Causing the energy in that channel to be diminished, resulting in the potential being lower than it really should be, and therefor would have a certain level of error. I was thinking if the connected capacitance was more than that spec, the switching inrush currents would exceed the 2.5mA injection current.
If injection current is the DC current, than 5v/1k doesn't match the spec noted. I believe that the injection current references the induced currents from the two adjacent channels. My first interpretation of that injection current was that the internal cap had an ESR that would be buggered if you measured a 5V signal with more than 1k.
From page VERR = K * RS * IINJ What's K? I'm guessing it's a constant of some sort, but I don't know what it is. Perhaps it's A-14's Kp or Kn, I don't know which it might be, as they are 100 times different from each other. I also don't know how to calculate the sum of injected currents.
Basically the sample and hold cap makes accurate measurement a difficult task. There are several sources of error and I would suspect that the 10 bits of accuracy, are really more like 8 bits usable. Unless you can balance several variable just right, and the datasheet doesn't give an excess in detail about how to do those predictions. Something as simply as changing the muxing speed, can change the injected currents, and input noise floor of the ADC.
I'm not crazy horny about that P+H here. It creates many sources error that raise the noise floor for the ADC. I can see how much of this can be calibrated via software by connecting a battery, and adjusting component parts or muxing speeds, until the signal in your log viewer becomes stable. So I guess your hands on experience will produce the best results.
Perhaps a basic schematic with suggested cap values and resistor could be done with Dia or paint, and posted here.
