I work with Mary West, so this question is related to the same system. We use the QEP in quadrature mode to read the encoder inputs. For the most part, this works flawlessly, but on two occasions during long multi hour runs, we've observed on one axis that the position indicated by the encoder is many counts off (> 1000) - both times, the magnitude of the error was virtually the same. Is there any way that the A and/or B inputs from the encoder could cause this? We've observed that disconnecting either A or B input causes the encoder output to hold its most recent value - but that's not what happened in our two observations. We read a totally different value.
Any insights would be appreciated