All this would clear up a little if you were to explain in a bit more detail your real requirements. As @jremington indicates with some frustration is that you started with a very low impedance voltage source you'd want to measure, and now it's apparently something involving PWM as well as a requirement for a high impedance measurement. In principle just about anything is possible, but it really helps if it's clear what the actual requirements are.
If you want to have a fairly fast sampling speed and sufficiently low impedance on the ADC side and high impedance on the DUT side, you could simply buffer the signal with an opamp. Today's opamps are quite battery-friendly/low current, so that shouldn't have to be a problem.
Requirements are clear moving forward measuring 1000 times a second and +/-0.2V accuracy and a voltage divider that will require the least amount of current.
So given the recommendation of max 10k source impedance you can compute the resistor divider. If you find that sample current too high, I stand by my recommendation to use an opamp.
I bumped up R1 and R2 almost one order of magnitude to R1 = 75k and R2 = 20k ratio is slightly different (20k/(20k+75k)) = 0.21 vs what I had (2k/(10k+2k)) = 0.166. I didn't notice any performance differences. Recorded these readings. I will bump up one more order of magnitude and perform a similar test and superimpose the two trends.
I am also using a ESP8266 GPIO pin with a 10M/2M divider to detect when I should disconnect a solar panel form the lead acid battery to prevent over-charging. The divider works fine until I connect the centre tap to D5, at which point the voltage collapsed from 1.83Volts (the LOW/HIGH threshold) to about 0.4V. The current flowing into the ESP is unmeasurably small so I really don't understand why this is happening. Any thoughts would be appreciated.
