Excellent PLC Co.,Ltd

PLC and DCS professional supplier

When Three Brains Think at Once: Synchronization Jitter in the Triconex 3101

Troubleshooting

When Three Brains Think at Once: Synchronization Jitter in the Triconex 3101

When Three Brains Think at Once: Synchronization Jitter in the Triconex 3101

By Robert Klein – System Integration Lead


If you’ve ever commissioned a Tricon system, you know the comfort of triple modular redundancy.

Three processors.
Three independent opinions.
One decision.

With the Triconex 3101 main processor module, that promise holds — but under extreme load, the way those three “brains” stay in sync becomes visible in subtle ways.


Project Background

This happened during the final phase of a refinery upgrade:

  • New safety logic added to an existing Tricon system

  • Scan cycle tightened to meet faster interlock response

  • Large number of analog inputs sampled at once

  • Redundancy channels fully loaded

Everything passed FAT.
SAT revealed something different.


What We Actually Saw on Site

  • Control outputs remained correct

  • No safety trips occurred

  • System logs showed minor timing jitter between channels

  • In rare cases, one channel lagged a few milliseconds behind

From an operator’s perspective, nothing was “wrong.”
From a systems perspective, timing drift had entered the picture.


Why This Happens

The 3101 MPM performs:

  • Real-time logic execution

  • Continuous cross-channel comparison

  • Input sampling and voting

  • Output commit only after agreement

Under high computational load:

  • Task scheduling becomes denser

  • Channel execution phases can slip slightly

  • Synchronization logic realigns channels before committing outputs

This realignment introduces tiny timing shifts — not failures, but temporal friction.


How We Proved It

We instrumented scan cycles across all three channels:

for each ScanCycle:
log(ChannelA_Time, ChannelB_Time, ChannelC_Time)
calculate(Delta_Max)

Under nominal load, deltas were negligible.
Under peak load, deltas widened — still within tolerance, but noticeable.


Engineering Lessons

  1. Redundancy synchronization has a cost under heavy load

  2. Timing jitter does not equal functional failure

  3. Scan cycle tuning matters more than most teams expect

  4. FAT rarely reflects worst-case site load


What We Changed

  • Slightly relaxed scan cycle timing

  • Redistributed logic execution

  • Reduced burst sampling of non-critical inputs

  • Added performance monitoring during peak operations

The jitter disappeared without touching hardware.


Closing Thought

The Triconex 3101 did exactly what it was designed to do:
maintain safety first, timing second.

When three processors argue, the system pauses just long enough to make sure they agree — and that pause is part of the design, not a defect.

Robert Klein

Prev:

Next:

Leave a message