Hampo In der Elektrotechnik ist der Offset der Wertebetrag um den eine Spannung in seinem "Grundpotential" abweicht
A camera sensor's offset is used for a different concept.
You are probably aware that the output from CMOS sensors are actually analog voltages. The voltage is converted to digital numbers (ADU == analog digital units) that represents the voltage, by an Analog-to-Digital converter (ADC).
Here is where it is different from standard Elektrotechnik: unlike signals from a microphone, etc, the CMOS sensors can only detect discrete photons that corresponds to the intensity of the light (amount of photons). So, it is always a positive value. It is no accident that ADU are also only positive numbers.
Sensor output noise and the ADC input noise can however be negative. Clipping it when the noise goes below zero will not allow the noise variance to be reduced when you sum sub-images ("stack"). I.e., if the noise causes the ADC to clip at zero, that noise will not be reduced by stacking.
To avoid this, a small voltage (called "offset") is added to the sensor voltage before it is given to the ADC.
This will make sure that none (in practice, it is "almost none" instead of "none") of the voltage that is given to the ADC will cause the ADC to "bottom out." Averaging sub-images now works properly to reduce the noise variance.
Strictly speaking, since noise is statistical in nature, the sensor noise can still cause the voltage to go below zero efor a given "offset." In practice, you just set the "offset" so that the voltage only goes below the offset with some small probabilty (for example only 0.1% of the time).
You can avoid clipping by using a larger offset, but that will reduce the dynamic range of the readings. That being said, newer 16-bit CMOS cameras have lots of dynamic range, so it is less critical when deciding what offset to use.
Wait for ZWO - Technik to give you a more complete answer.
Chen
