thanks for your answer.
I compare my data with the specification in the datasheet:
"Noise (on 50 mV range) 8-bit mode 120 μV RMS • 12-bit mode 110 μV RMS • 14-bit mode 100 μV RMS •15-bit mode 85 μV RMS • 16-bit mode 70 μV RMS".
This is valid also for the scopes with 200MHz bandwidth.
As my scope is a 100MHz bandwidth the noise should even be lower.
The peak to peak noise is about 10 times larger.
Yet with the noise data in the datasheet one can ask whether 14Bit till 16Bit makes sense.
In my experience 14Bit and more brings not much accord to 12Bit.
I test the noise with open inputs and with 50Ohm. It is no much difference
Only with lowpass filter the data with 50Ohm becoming substantial better than with open inputs. Presumably a good part of the noise comes from the higher frequency.
In 8Bit mode the cannel A has much higher noise than the cannel B in RMS. (250 uV against 100uV)
Peak to peak is cannel A not worse than channel B.
I will this mail with the requested file to firstname.lastname@example.org