F
Member
【Editorial Discussion】
A program change on a PLC is often still performed from a shared laptop, a vendor tool, and a cable that anyone in the shop can pick up. Network architecture matters, but so does the practical question: after the change, can you tell who wrote what, whether the file matches the approved project, and whether the controller was left in a mode it should not be in.
1) Identity and approval should sit in front of the cable, not only in a procedure document. A shared engineering login, or a project file copied from a USB stick whose origin is unclear, makes later incident review almost impossible. If you claim change control, the station, the user, and the project version need to be attributable.
2) Compare the running image to the approved project after the write, not only before. A download that was interrupted, a subset of blocks written, or a colleague’s unofficial tweak will not be visible if you only archive the file that was supposed to go in.
3) Controller mode, forced values, and leftover test code are part of the change. Leaving the CPU in a programming mode, or leaving forces on, is a safety and security issue as much as a process issue. The close-out checklist should include these, and a password or access control that is not taped to the panel.
4) Offline copies age. The “master” on a shared drive that is older than the running controller is how unofficial changes become the unofficial truth. Reconciliation should be periodic, not only after a failure.
Open question: what actually prevents an unapproved download in your shop — technical controls, or discipline? Have you found a running PLC that did not match the archived project? Practical controls, short of naming sites, are welcome.
A program change on a PLC is often still performed from a shared laptop, a vendor tool, and a cable that anyone in the shop can pick up. Network architecture matters, but so does the practical question: after the change, can you tell who wrote what, whether the file matches the approved project, and whether the controller was left in a mode it should not be in.
1) Identity and approval should sit in front of the cable, not only in a procedure document. A shared engineering login, or a project file copied from a USB stick whose origin is unclear, makes later incident review almost impossible. If you claim change control, the station, the user, and the project version need to be attributable.
2) Compare the running image to the approved project after the write, not only before. A download that was interrupted, a subset of blocks written, or a colleague’s unofficial tweak will not be visible if you only archive the file that was supposed to go in.
3) Controller mode, forced values, and leftover test code are part of the change. Leaving the CPU in a programming mode, or leaving forces on, is a safety and security issue as much as a process issue. The close-out checklist should include these, and a password or access control that is not taped to the panel.
4) Offline copies age. The “master” on a shared drive that is older than the running controller is how unofficial changes become the unofficial truth. Reconciliation should be periodic, not only after a failure.
Open question: what actually prevents an unapproved download in your shop — technical controls, or discipline? Have you found a running PLC that did not match the archived project? Practical controls, short of naming sites, are welcome.