Excellent PLC
By Kevin Marshall – Industrial Communication Engineer
One of the most misleading things about RS485 is that it can keep working while already being broken.
This case involved a Honeywell 07191/1/1 RS485 communication board installed in a control cabinet that had survived multiple plant expansions, grounding changes, and electrical upgrades. Nothing dramatic happened. No alarms. No sudden loss of communication.
Just slower updates.
Then missed polls.
Then silent data corruption.
What Made This Case Hard to Diagnose
-
No device went offline
-
No checksum errors reported at the DCS level
-
Communication looked “alive”
-
Data quality degraded gradually over months
Operators adjusted. Engineers ignored it. That’s how RS485 failures like this stay hidden.
The Physical Reality Behind the Bus
RS485 is a differential signal, but it is not immune to ground potential differences.
In this installation:
-
Field devices were grounded locally
-
Control cabinet grounded through a separate earth path
-
Shield drain connected at multiple points
-
Ground potential difference measured: 3–6 V during load changes
That voltage doesn’t break RS485 immediately.
It biases it.
What the 07191/1/1 Was Experiencing
From the board’s perspective:
-
Differential signal still present
-
Logic thresholds slowly shifting
-
Noise margin shrinking
-
Valid bits becoming ambiguous
Eventually, bits weren’t wrong enough to be flagged — just wrong enough to slow decoding and force retries.
Observed Communication Behavior
-
Increased response latency
-
Poll cycles stretching unpredictably
-
Occasional incorrect values that self-corrected
-
No clear fault pattern
This is classic common-mode voltage drift, not line noise, not termination error.
The Moment It Became Obvious
We measured common-mode voltage directly on the RS485 lines:
Differential looked fine.
Absolute reference did not.
That’s when the problem stopped being mysterious.
Corrective Strategy
Grounding correction
-
Single-point shield grounding at control cabinet only
-
Removed shield-to-earth connections in the field
-
Verified equipotential bonding between cabinets
Bus stabilization
-
Added isolation barrier between 07191/1/1 and field bus
-
Reduced stub lengths
-
Rebalanced termination placement
Operational adjustment
We didn’t wait for failure anymore — we watched for drift.
Results
-
Poll timing stabilized
-
No further silent data corruption
-
Communication latency returned to baseline
-
Same 07191/1/1 board remained in service
No hardware replacement was required.
What This Teaches (The Hard Way)
-
RS485 can lie convincingly
-
Grounding problems rarely announce themselves
-
“It still works” is not a valid acceptance test
-
Communication quality must be measured, not assumed
Final Note
The Honeywell 07191/1/1 RS485 communication board didn’t fail electrically.
It was slowly pulled out of its reference frame.
And in serial communication, losing your reference is how truth disappears without warning.
— Kevin Marshall