Morning all, maintenance ran new conduit for the DeviceNet trunk on the oven line last week, the old one was in a tray that got hit by a forklift. 1756-DNB, 14 nodes, 125k baud, trunk is maybe 45m now.
Since the rework the scanner display goes to E91, every node drops and the line stops. Power cycle the scanner and it runs two or three hours, then E91 again. All the node LEDs stay green right up until it happens. We swapped in our spare scanner and it did the same, and I unplugged nodes one at a time overnight and it still went with four left.
Is this a bad node that only shows up hot? The line runs 24/7 and I'm out of ideas. Or has the new trunk itself got something wrong with it that I should be checking before I start pulling nodes again?
E91 is bus-off, the CAN controller gave up after too many errors. Not a node. Was the terminating resistor at the far end refitted after the new trunk? 121 ohm, one at each physical end. Segment length and baud are fine, 45m at 125k is nothing.
Meter it before you guess. Network power off, ohms between CAN_H and CAN_L at the scanner tap.
Metered it with network power off at the scanner tap. 118 ohm. The old trunk had a terminator on the end, I remember the little plug. The new trunk just ends at the last tee on the oven exit. So one terminator, at the scanner end. Is that enough to cause this or am I chasing the wrong thing?
About 60 is what you want. 118 is one resistor. Been there.
So the far end is open. Without the second terminator every frame reflects off the open end, the CAN controllers see error frames, the error counter in the scanner climbs, and when it hits the limit it goes bus-off, E91. It takes hours because the reflections only corrupt a fraction of frames, and more traffic or a warmer cable makes it worse, which is why the oven line and not the others. Put a 121 ohm 1 percent resistor across CAN_H and CAN_L at the last tee on the oven exit, or the proper terminating plug. Network power off, meter again, you want about 60 ohm. Then a power cycle and it stays up.
Put a terminating plug on the last tee last night, metered 61 ohm at the scanner with power off. Power cycled and it has run 19 hours, longest since the conduit job. Display cycles the nodes like normal.
So the far end terminator went in the scrap bin with the old trunk and nobody noticed. One resistor on a 14 node trunk, bus-off after a few hours every time, and a 121 ohm plug at the last tee is the whole repair.
Still don't know why it was worse in the afternoons. Oven heat on the cable maybe. Or just more traffic on day shift.
Thanks dieter_k, edj, stevek.
Warmer copper, more loss. Or just day shift. Either way it's fixed.