It was noted that the bias level of all 4 CCDs seem to vary between observing runs, and the bias level of one of the CCDs has shown some variability on timescales of minutes. Both of these things are being investigated in detail.
Some observers also expressed an interest in increasing the amount of over-scan pixels to be able to properly estimate the bias level of the different CCDs during an exposure. However, due to the limited size of the memory of the array controller this will be at the cost of the science pixels, and we have to look carefully how to minimize this.