ADC resolution - voltage dividing/multiplying stage
General
When you are using ranges (20mV/div - 100mV/div) it uses a different path than when using (200mV/div or higher), this is to reduce noise and to get more resolution. You'll always hear the relay switching in the transition between those ranges.
On each voltage range (vertical resolution in volts/div) you will have different minimum and maximum value that the scope can measure.
This depends on your current vertical offset and there is some extra room out of the screen (that is different on each range) but as general rule of thumb if a signal goes outside of the screen vertically you are probably exceeding the limits of that range and the scope is clipping the signal.
So you need to adjust the resolution and vertical offset so you can see the whole signal in the screen (vertically) if you want to measure it.
Example for a 2V signal
Since the main grid on the SmartScope software has 8 division and your signal is 2V that means that 200mV/div x 8 divisions = 1.6V is not enough, and that is why your voltage got clamped to 1.4V, probably your were using an offset of -3 division giving you only 7 div. x 200mV = 1.4V! so your signal was clamped to 1.4V
And in lower ranges it will be even worse, but that is the normal operation of any scope, you cannot measure larger signals than what you can fit in the screen in your current range.
Voltage resolution
The resolution of the scope also varies between ranges, so don't expect to see exactly the same value in each range, but they should be really close.
The resolution of the ranges is displayed in the next table:
| Range Max. | Resolution Max. | Resolution Typ. |
|---|---|---|
| 20 mV/div | 1.6 mV | 2 mV |
| 50 mV/div | 1.6 mV | 2 mV |
| 100 mV/div | 3.2 mV | 5 mV |
| 200 mV/div | 6.3 mV | 10 mV |
| 500 mV/div | 17 mV | 20 mV |
| 1 V/div | 32 mV | 50 mV |
| 2 V/div | 63 mV | 100 mV |
| 5 V/div | 157 mV | 200 mV |
In addition to that, different ranges use different amplification so noise, temperature and small calibration errors could make also the values differ even more.
And after all, this is an 8-bits scope, it is not intended for precise voltage measurement, so you can be confident that it will give you an accuracy of arround 10% of your current volts/div but do not ask for more wink. So in 1V/div the figures will be accurate to 0.1V, and in 500mV/div to 0.05V.