We have encountered this before on troublesome networks. The way it has been fixed is to run the update on a consolidated network. One MFD and the SCX20, removing everything else from the equation. This ensures a clean data flow and will increase the speed of the update as well.