The Marginal vs Failed Component Decision Tree
Why this matters
Not every bad reading is a dead part, and not every in-spec part is a good one. A component can be failed (clearly out of spec, done), marginal (technically passing but weak, on its way out), or genuinely healthy. Confusing these three costs both ways: condemn a healthy part and you waste the customer's money; pass a marginal part and you are back next month for the callback. The skill is reading where a component sits on that spectrum and matching the action to it. This tree separates failed from marginal from fine, and tells you what to do with each.
Safety first
A marginal component in a safety-critical role is not a "watch it" item. If a part that is drifting or weak protects against a hazard - a relief valve, a safety switch, a high-limit, a ground, a gas safety - replace it now, do not watch it. A marginal safety device is a failed safety device with a delay. Lead with the replacement when the part's job is to prevent a gas, electrical, pressure, or fire hazard, regardless of where it sits on the spectrum.
Start here: where does the reading sit
Measure the component against its spec or a known-good baseline and place the reading in one of three zones.
- Well outside spec - failed. Clear-cut replace. No judgment needed.
- Inside spec but near the edge - marginal. This is where the real decision lives.
- Comfortably inside spec - healthy. Do not replace it; look elsewhere for your fault.
The hard cases are all in the middle zone. A reading that is "passing, but barely" is the one that demands judgment, not a parts swap reflex and not a blind pass.
If the reading is well outside spec
Failed component. Replace it. Confirm it is the cause and not a victim of an upstream fault (see the related fault-behind-the-fault method), then swap it. There is no watch-and-wait on a part that is clearly done; the only question is whether something killed it that will kill the replacement too.
If the reading is marginal
This is the heart of the tree. A part inside spec but near the limit is not automatically a replace and not automatically a pass. Weigh these factors:
Branch A: is it trending toward fail or just naturally near the edge. A part that has drifted from healthy toward the limit is going somewhere. A part that has always read near that value and held may simply live there. Your own prior readings or a known-good twin tell you which. Drift toward the edge argues replace; stable-at-the-edge argues watch.
Branch B: does the marginal reading explain the symptom. If the weak component accounts for the complaint (a weak start, a slow response, an underperformance), it is your cause even though it technically passes. Replace it. If the marginal reading does not explain the symptom, do not let it distract you from the real fault.
Branch C: what does it cost to fail later. A marginal part that is easy and cheap to reach, with low consequence if it dies, is a reasonable watch. A marginal part buried deep, expensive to get back to, or one whose failure strands the customer or cascades into other damage, argues for replacing it now while you are already in there. Access and consequence drive this, not the reading alone.
Branch D: is it cheaper to do it now. If you are already inside the unit with it exposed, the labor to replace a marginal part is mostly spent. Coming back later doubles the trip. Proximity makes "replace the marginal part now" the economical call even when it would be a watch on its own.
If the reading is comfortably inside spec
Healthy component. Leave it. The most disciplined move in diagnosis is to not replace a good part because you ran out of other ideas. If the reading is solidly in spec, the fault is elsewhere. Reopen the symptom tree and keep looking.
Confirming the diagnosis
Confirm a marginal call by comparing against the strongest baseline you have and, where possible, by testing under load. A part that reads fine at rest but sags under load is marginal-trending-failed, and load testing exposes it.
| Reading | Under load | Action |
|---|---|---|
| Outside spec | Worse | Replace, find what stressed it |
| Near edge | Holds steady | Watch, or replace if access is open |
| Near edge | Sags or drifts | Replace, it is on its way out |
| In spec | Holds | Leave it, look elsewhere |
Document the marginal reading and your call. "Capacitor reads near low limit, sagged under load, replaced" or "reads near limit but stable and easily accessible, advised customer to monitor" both protect you and inform the next visit.
Next steps
Replace failed parts and safety-critical marginal parts without debate. For non-safety marginal parts, let access, trend, consequence, and proximity decide, and tell the customer plainly what you found and why you chose replace or watch. Never replace a comfortably-healthy part to feel busy. When you choose to watch a marginal part, note the reading so the next tech can see whether it drifted further.
References
- Manufacturer nameplate and tolerance specifications (the basis for in-spec vs marginal).
- ISO 13379-1 (Condition monitoring and diagnostics of machines; degradation and trend assessment).
- Trade-standard practice for load-testing components near tolerance limits.
- See related: Baseline vs Faulty Comparison Diagnosis Method; The Tolerance Stack-Up That Causes the Fault.