TIA V17, CPU 1516F-3 PN/DP, ET 200SP with an F-DI 8x24VDC HF. Machine we shipped in spring, a rotary indexer with one guard door, mechanical interlock with two NC contacts on one actuator, into channels 0 and 4 as a 1oo2 pair, equivalent.
Customer rings, the door passivates a few times a shift. Their buffer says "Discrepancy error, channel 0" and then "Channel fault resolved" a couple of seconds later, always while the door is being closed, never when it's sitting shut.
So I went to site and put a new switch of the same type on, same fault by the afternoon. Checked the wiring at the BaseUnit and at the switch while I was in there, both channels tight.
Their electrician's pretty sure it's a bad batch. I'm not. What am I missing on the F-DI side?
What's the discrepancy time in the channel parameters for that pair, and what exactly is the switch, part number? I'm not guessing at a time without knowing the switch.
Whatever TIA put in when I set the pair to 1oo2. I never touched it, that one's on me. 10 ms I think, I'll confirm tomorrow from the project. Switch is a Siemens 3SE5 with the separate actuator, two NC.
We run 500 on the DCS blocks for mushroom heads on our GuardLogix side. Different box, same idea. Is a slammed door really meant to get two contacts agreeing inside 10 ms, or is that number there for something else?
For something else, and not for a door like that. And nobody drop that pair to single channel to dodge the fault, the diagnostics are the safety engineer's call.
The time is the problem. Two contacts on one mechanical actuator don't change within 10 ms when a door gets slammed. There's bounce, and the actuator goes in at an angle. The F-DI compares the channels and if they disagree longer than the discrepancy time it passivates the pair. Correct behaviour, wrong parameter.
Device view, the F-DI, Properties, F-parameters, the channel pair. Set the discrepancy time to what the switch supports. A few hundred ms is typical for a mechanical two-contact switch, and the 3SE5 datasheet is where I'd start. As short as the switch allows. It goes into the response time calculation too.
Confirmed 10 ms in the project. Set the pair to 300 ms after watching the two channels online for a while, on a hard close they were further apart than I expected, tens of ms. Downloaded with the customer's safety person present, new collective signature, door tested ten times. No passivation since yesterday afternoon.
That default was mine to look at and I never did. 10 ms is nowhere near what two contacts on one actuator actually do on a hard close, so the pair passivated every time an operator slammed the door.
The switch I swapped out is almost certainly fine and it's still sitting in my bag. Thanks larsdk and sparkymike.
Same reason we're at 500 on the mushroom heads.