Oscilloscope how many times of averaging they can remove noise sufficiently

[Kiriakos GR]

Not all oscilloscope are build equal in hardware level, even so all include the averaging function.

In theory:
Best use of averaging this is due mathematical removal of acquisition points with low repeatability.
Sine wave this is signal of highest repeatability, and therefore anything on top of that this is noise.


Oscilloscope front end noise:
Key marketing point at medium and high quality oscilloscopes this is input (channel) front end noise.
The lesser = the better

Inspection of per channel front end noise this is performed at 100us / Time division and at the lowest voltage division of 1mV.
No probes connected ...
 
Oscilloscope math averaging
Due trial and test, I did discover that averaging this it can also remove portion of noise this generated by the ADC within the oscilloscope.
At my GDS-2102A  ADC noise (front end noise), this were measured at 460~480uV PP.

After the activation of averaging at GDS-2102A, front end noise output this become half.
Averaging available choices:
2x
4x
8x
16x
32x
64x
128x
256x


By testing now different averaging values, I did discover one sweet spot at 64X (times of averaging), which this helped front end noise to go down at 120 -160uv PP.

In summary, my finding this is limited to the very specific oscilloscope and its a result of combined performance.
Hardware parts + acquisition memory length + CPU performance at math calculations.

Either away by you executing identical steps of testing at your own oscilloscope, you might be able to discover an identical sweet spot of oscilloscope performance about highest noise rejection setting.
That is the best that your own and specific oscilloscope can do. 

In case that I did motivated you enough, so you to run similar tests?
Then please add your oscilloscope best score at this topic.
In time we might be able to have a database of oscilloscope models VS their performance at averaging.