A dropout is an event in time, not a state. Proving one means catching it while it happens — and your tool is only as honest as the window it can see through.
An intermittent dropout is a brief electrical event: a voltage leaves its band, a return lifts, a signal vanishes — then everything goes back to normal before you can look at it. That makes it fundamentally different from a steady fault. A steady fault is a state you can measure at leisure; a dropout is an event that exists only inside its own time window. Proving an event requires a witness that was watching at event speed. The tool you choose is that witness — and every witness has a speed limit. A dropout faster than your tool's window is invisible to that tool. Invisible is not absent. Confusing the two is how intermittents get "cleared" by tests that never had a chance of seeing them.
Every capture tool answers a different question about time. A digital multimeter on MIN/MAX tells you the extremes it saw within its response time — it does not tell you what happened between its samples. An oscilloscope shows you the actual waveform across a chosen timebase — it sees everything inside that window and nothing outside it. A scan tool shows you what the module reported, when the module got around to reporting it — filtered, sampled, and delayed by the module's own update rate. Three tools, three different relationships to the event. The diagnostic question is never "which tool is best" — it is "does this tool's window cover the complaint's time scale?" If you don't know the complaint's time scale, say so on the record: an unknown time scale caps what any capture can prove.
A DMM with MIN/MAX capture is the honest workhorse for dropouts inside its window. It proves extremes: the lowest the reference fell, the highest the return lifted, during the capture period. But its window is only as fast as its stated MIN/MAX response time — and that number comes from the meter's specifications, not from optimism. Record the meter model and its stated response time with every capture. If the response time is unknown, the honest bar is conservative: only excursions lasting a full second or longer are credited. A shorter blip on an unknown meter might be real, or it might be the meter's own sampling artifact — and "might be" is not evidence. The DMM also proves nothing about what happened between its extremes: MIN/MAX gives you the floor and the ceiling, not the shape of the event.
A scope is the only tool that shows you the event itself — the shape, the duration, the recovery, the exact moment relative to the provocation. That makes it the right witness for dropouts faster than a DMM's proven window. But a scope proves only what its timebase covers: set the timebase from the complaint's time scale, name it on the record, and use triggering so the event captures itself instead of relying on your reflexes. A scope also demands more of the operator — probe placement, grounding, and timebase selection are all part of the evidence. An untriggered scope trace that "didn't show anything" proves the trace didn't show anything, not that the event didn't happen. The scope is the sharpest witness and the easiest to misuse; both facts belong on the record.
A scan tool shows you the module's report of the signal — not the signal. Between the wire and the PID sit the module's sampling rate, its filtering, its plausibility checks, and its update cadence, any of which can hide, smooth, or delay a brief dropout. So PID behavior is supporting evidence only: a PID freeze or glitch that lines up with a direct measurement strengthens the case, but a PID anomaly without direct-measurement corroboration is suspicion at most. And the reverse discipline matters more: absence of a PID dip is not absence of a dropout. A dropout shorter than the module's update window can come and go without the PID ever moving. Clearing a circuit because "the PID looked fine" is trusting the module to report an event it was never fast enough to see.
The discipline runs in this order: first, name the complaint's time scale — how long does the event last, as best you can tell from the symptom? Second, choose the tool whose proven window covers it. Third, if no tool you own covers it, say so: the test is inadequate, not the circuit is good. A DMM-only shop is not disqualified from intermittent work — most heat-soak and wiggle dropouts live comfortably inside a DMM's window — but the match has to be stated, not assumed. "The meter didn't catch it" is a finding about the meter until the window match is on the record.
A capture nobody wrote down is a story, not evidence. The record for every capture names: the tool (meter model or scope), the window (stated MIN/MAX response time, or scope timebase), the measurement points, the provocation applied, and what was seen — including "nothing seen," which is a real result when the window match is stated. This record is what lets the second pass confirm the first, what lets another mechanic repeat the test, and what keeps an honest "I don't know the time scale" from silently becoming "it must be fine."
When an intermittent is in question, SpeedNeeds matches the witness to the event before any capture counts: the complaint's time scale is named (or admitted as unknown), the tool's proven window is stated on the record, and captures that don't cover the complaint are treated as inadequate tests rather than clean bills of health. Direct measurement leads; scan data supports but never overrules it. Every capture is recorded with its tool, its window, and its provocation — and a single sighting is treated as suspicion until a second independent capture confirms it.
Think of it like photographing lightning: the camera doesn't make the strike happen, and a camera with too slow a shutter doesn't prove the sky was quiet — it proves the shutter was slow. Name the event's speed, pick the witness that can see it, write down what the witness saw. That is what a capture proves — and knowing what it can't prove is the other half of the discipline.
Stay in the loop
New QTS builds, tools, and garage stories — straight to your inbox. No spam. Unsubscribe anytime.
Read the Privacy Policy.