Missing bit values in GW Instek GDS-1054B

[maxwell3e10]

I just purchased a new Instek GDS-1054B and could immediately see that ADC is missing some bit values. Here are screen shots with a triangular input wave, one can see gaps in the data lines. But its not just a graphical problem. If I save the data and then do a histogram of the bit values, one can clearly see that certain values are missing. The period is roughly every 4th bin if acquiring from one channel and every 8th bin if acquiring from 2 channels. It does not depend on the sample rate or which channel is being used. When acquiring from all 4 channels, the gaps are less obvious, but still seem to appear under certain conditions.

I have contacted GW Instek support but have not heard from them yet. Unless its a peculiar hardware defect in my scope (which is hard to imagine since the output of the ADC is serial), it seems to be a pretty glaring flaw in their firmware.

[rf-loop]

(which is hard to imagine since the output of the ADC is serial),

Really(?), what is this 1GSa/s ADC what have serial mode output? This interface must be fun..

Back to ground surface
As far as I know there is just one ADC chip and it is Hittite Had1511.

[maxwell3e10]

(which is hard to imagine since the output of the ADC is serial),

Really(?), what is this 1GSa/s ADC what have serial mode output? This interface must be fun..

Duh! That's what I get for reading in the datasheet "Serial LVDS/RSDS Output" and not thinking any further.

Anyway, I think I figured out the problem with the firmware. They must be doing some data scaling using  8-bit numbers. Here is a histogram if one takes an integer, multiplies it by 1.3 and makes it integer again.  Looks exactly the same as the scope data.

I haven't seen this problem in other scopes I checked. They must apply an overall vertical calibration correction without forcing the result to be an 8-bit number.