Voltage Divider for Odd ADC?

[GreenBEM]

Just discovered that the ESP-WROOM-32 I'm toying with, has a rather odd ADC.

With other controllers, a simple voltage divider has provided a suitable ADC input voltage for measurement of a full-range voltages.

Now, it appears that I need to both divide a measured voltage, as well as translate it to place it within the 100mV to 950mV ADC input range. Below 100mV down to 0mV seems pretty much unusable, as it doesn't convert at all linearly within this range.

Baard says I need to learn more about non-inverting, summing op-amps, but I have zero faith in anything it comes up with.

Is there a simpler solution?

TIA.

[ESP_Sprite]

You can usually get away with a lot less (as in: a resistive divider) or even nothing if you use one of the attenuation settings. A more important question would be: what are you trying to measure?

[GreenBEM]

Just trying to measure some simple voltages and currents.

I believe that the C3 and S3 variants have an improved ADC. If you extrapolate their linear regions for conversion, they intersect the y-axis at the origin.

For the same extrapolation, the ESP-WROOM-32 I'm using has a significantly non-zero y-axis intercept, hence the need for input voltage translation.

It's not an issue for the ACS712 bi-directional current sensors I hope to use, as their zero-current voltage output may be easily situated central to the linear region for ADC conversion. For unipolar voltage measurements, where the entire span of voltages is of interest, right down to and near zero, it's not so simple.

[GreenBEM]

Maybe this will clear-up my earlier post:

mV_on_ADCreg.jpg (36.75 KiB) Viewed 1448 times