@SteveSh For ground based telescopes, the noise/errors during the acquisition process will probably overwhelm the ~6 count difference out of 65536 that will be detected. But as you say, with a large number of samples it may be possible to statistically detect it.

I get [0,10,20,10,0,0] as the solution from that system here, which doesn't seem to be correct. Can you elaborate on this?

Looking at the datasheet, it looks like it specified the valve's input resistance

Luckily the column and row pins were sequential and separate on the controller. I created an input matrix and also measured all pins over time, so I knew which ones were outputs with pullups. So I could be quite sure it was that way.

This is exactly what I just did, it works! Setting all column pins to LOW, reading which of the row pins was set to LOW, and then setting the right column pin to HIGH only when the right row was reached, otherwise keep it LOW. I was concerned that the Arduino might be too slow, since the microcontroller seems to run with 12MHz. But the row sampling/strobe interval seems to be large enough! Thank you!