There is a band structure seen in the biases, at the start of each amp's readout. The flux rolls up over ~6 pixels from about -15 ADU, stays about 0.5 ADU high for ~30 pixels, then dips to the nominal amp level.

The structure is essentially identical for all amps.

The structure is not visible in bias subtracted superdarks or superflats.

The attached image is from a superbias for the r1 detectors, taken in the optics lab dewar on 2015-12-22. Frames 7251-7259. The amp geometry has been corrected, so this shows the readout for two amps, with time for both starting in the middle of the image and moving out.