The sequence hangs on one step and nobody can say why

Because the step has no timeout, so there is nothing to report. A step that waits for a transition condition and never gets it will wait forever, and the only symptom is that the machine stopped. The operator says "it just stopped", which is completely accurate and no help at all. The fix is a rule rather than a repair: every step gets a timeout, and the timeout raises an alarm that names the step and the condition it was waiting for. Then the same failure reports itself as "step 4 timed out waiting for XV-101 open limit", which is a fault somebody can act on. Two more things that pay for themselves immediately: - Expose the current step number to the HMI. Diagnosing a stuck sequence without it means going online with the programming software, which not everyone on site can do. - Log step transitions with timestamps. A step that usually takes 2 seconds and today took 30 is a bearing starting to go, and you can only see that if the transitions are recorded. If you are retrofitting this to an existing sequence, start with the steps that wait on field devices. That is where nearly all of them hang.