A sensor signal can be healthy on every axis — clean amplitude, correct frequency, solid wiring — and still mark the physical event at the wrong moment. "Late" is an observation, not a diagnosis. Somebody owns the lateness; the discipline is naming which one.
Engine position is measured two ways at once: crankshaft degrees and milliseconds. They are not interchangeable, and the difference between them is where most timing conclusions quietly break. A processing delay inside the sensor or the module is constant in TIME — the same number of milliseconds at idle and at redline. But degrees get smaller as RPM climbs (each revolution takes less time), so a constant-time delay occupies MORE degrees at higher RPM. A mechanical position shift — a stretched chain, a jumped tooth, a slipped trigger wheel — is the opposite: constant in DEGREES, and occupies LESS time as RPM climbs. Every offset reading has to be written down in both units or the classifier cannot run: milliseconds alone can't tell you the chain stretched, and degrees alone can't tell you the module filtered. The single most useful move in timing correlation is measuring the offset at two different RPMs and watching which unit stays put.
Here is the trap: the sensor under test cannot prove it is late, because by construction it marks the event wherever it marks it. Comparing the sensor's edge to the sensor's own edge is circular proof. You need a witness that is independent of the sensor — something that sees the physical event through a different path entirely. The witness ladder, in order of directness: an in-cylinder pressure transducer (you are watching the combustion event itself), ignition primary (the spark's own electrical moment), injector current (the commanded delivery), a strobe on the timing marks (the mechanical position, visually), or a known-good reference sensor on the same shaft (trusted, but still secondhand). Pick the highest rung you can afford for the job. Scan-tool data sits at the bottom of this ladder, and arguably below it: a PID is the module's report of the signal, and the module sits downstream of every owner on the list. A PID cannot witness a timing event for the same reason a suspect cannot be the only witness to their own alibi.
Variable cam timing means the physical event is supposed to move. A phaser commanded to advance moves the edge on purpose — that is intended movement, owner number five, not a stretched chain. The discipline is mechanical: command the phaser to a known state (full park, or a commanded position the data confirms), record the state on the capture, and only then read the offset. An offset read against an unknown commanded state is an offset measured against a moving ruler. The commanded-state question gets asked and answered before the correlation runs, never after it surprises you.
Before any offset is decision-grade, the jitter gets a number. Edge-to-edge jitter — how much the edge wanders event to event — is the noise floor of the measurement. If the jitter band is wider than the measured offset, the offset is not a finding; it is the noise wearing a costume. Jitter is the entry gate the offset has to clear, and both numbers go on the record. This is also why a single capture is never enough: one clean edge is a sighting, not an offset.
A miscounted signal cannot be correlated. Before the offset is even read, the signal has to prove it is what you think it is: correct count, no dropouts, no noise riding the edge. A noisy or dropout-ridden edge correlated against a pressure trace produces a phase number, but the number is fiction — garbage in, garbage out. The signal-quality gate comes first. Correlating a dirty signal is the fastest way to diagnose a measurement problem as an engine problem.
The separation order is the owners' order, and it is deliberately elimination-style: (1) nail down the measurement — channel assignment, skew bound, trigger edge, and a channel swap or repeat pass so the measurement's own error is quantified; (2) verify signal content before any correlation runs; (3) record RPM, load, temperature, and commanded phaser state with the reading; (4) read the offset in BOTH degrees and milliseconds against an independent witness; (5) repeat at a second RPM. If milliseconds stay constant while degrees move with RPM, the owner is module-side time. If degrees stay constant while milliseconds shrink with RPM, the owner is mechanical position — the timing drive or the trigger wheel. If the offset moves with commanded state, the owner was intended movement all along. Nothing on this list requires trusting the module about its own timing; that is the whole point of the ladder.
Before any timing offset counts, SpeedNeeds names the witness and proves the witness is independent of the sensor under test; the signal's content is verified before correlation runs; every offset is recorded in degrees AND milliseconds at the captured RPM, load, temperature, and commanded state; the reading is repeated at a second RPM; and the five owners are eliminated in order so that "late" becomes a named finding instead of an observation. The unit system that stays constant names the owner — and the record shows which unit did it.
Think of a relay team handing off a baton: if the baton arrives late, the finish-line clock doesn't tell you who dropped it — the runner, the handoff, the judge's stopwatch, or a headwind. Timing works the same way: the sensor, the module, the drive, your own measurement, and the conditions all touch the moment before it reaches the record. Measure it in both unit systems, witness it from an independent seat, and name the owner before you name the fault. "Late" is where the diagnosis starts, not where it ends.
Stay in the loop
New QTS builds, tools, and garage stories — straight to your inbox. No spam. Unsubscribe anytime.
Read the Privacy Policy.