Every parts drawer has a handful of diodes with no labels, no data sheets, and a reputation for causing problems the moment they go back into a circuit. The question “is this diode bad?” is really a workflow question: how do you decide whether an unmarked part from a drawer is usable, without knowing its part number or its history? The answer is a three-step bench autopsy — read the physical evidence, run the meter checks, then verify under load — and the discipline of knowing when a “passing” part is still a risk. This article turns that workflow into a repeatable procedure you can run in minutes on any salvaged or suspect diode.
Reading the Physical Evidence: Cracks, Burns, and Bulged Bodies
The first look at any suspect diode is visual, and it goes faster than a meter test. A cracked glass body, a burned or discolored junction, a bulged or melted case, or a broken lead are all irreversible damage — no electrical test will bring those parts back, so the visual pass is the fastest way to sort out obvious failures.
The visual check has limits worth stating. A diode can look perfect and be electrically dead, and a slightly discolored part can still be perfectly functional. The visual pass is a filter that catches the obvious, not a verdict on the rest. What the look does do is establish the baseline: a part with no visible damage earns the meter stage, and a part with a crack is done regardless of what any reading might show.
Use a loupe or the macro setting on a phone for SMD parts, where the body is small and the damage is proportionally tiny. Look for a dull or frosted glass surface (the classic sign of a leaked vacuum on a glass-bodied diode), a dark burn ring around the junction, or a bulge in the body that suggests the part ran hot enough to soften its case. The axial rectifier guide covers the package-level expectations that make this inspection faster.
Step 1-2-3 Workflow: Diode Mode, Reverse Check, Leakage Test
The meter stage of the autopsy runs as a strict sequence because the order tells you which failure mode you are looking at. Each step uses the multimeter’s diode mode, and each step has an expected reading and a meaning.
Step one is the forward check. Place the red lead on the anode (guessed for an unmarked part) and the black on the cathode, and read the forward drop. A healthy silicon diode reads roughly 0.5–0.8 V; a Schottky reads lower. If the forward check shows OL — no conduction — the part is open, and the autopsy can end with “bad.” If it shows a plausible drop, move on.
Step two is the reverse check. Swap the leads and read again. A healthy diode shows OL (blocking); a shorted diode shows the same small drop in both directions; a leaky diode shows a small but nonzero reverse reading. The symmetry — the same reading both ways — is the shorted-diode signature, and it is the fastest single verdict the meter can deliver.
Step three is the leakage test, which belongs to the precise currents the diode mode cannot see. Set the meter to the current mode’s microamp range, apply a reverse bias, and read the leakage. A PA-style leakage in the nanoamp-to-microamp range at room temperature is normal for a silicon PN diode; a milliamp leakage at low reverse voltage is a problem. The leakage and thermal-runaway explainer covers the temperature dependence that makes a warm bench leak more than a cold one.
Testing a Salvaged Diode Whose Part Number Is Gone
The unmarked diode is the case the workflow was built for, and the judgment basis changes when the part number is unknown. Without a data sheet, “healthy” cannot mean “matches the datasheet”; it has to mean “behaves like a healthy junction of its apparent type.”
The first step for an unmarked part is to determine its type from the meter’s forward reading. A drop near 0.5–0.8 V at the meter’s test current suggests a silicon PN or rectifier diode; a drop well under 0.4 V suggests a Schottky; a very low drop with instability points at an old germanium part. That type estimate sets the expectations for the reverse and leakage checks.
The loaded verification is what separates a usable unmarked part from a gamble. Wire the diode into a simple test rig — a battery, a resistor sized for the expected current, and the meter across the diode — and confirm the forward drop at a real current is plausible for its apparent type. If the part is meant to carry amps, pass a fraction of an amp through it and watch the drop; a part that reads healthy in diode mode but collapses under load has a degraded junction, and the autopsy should end with “not trustworthy.”
The general rectifier category lists the families this verification is measuring against, and the rectifier failure modes guide places the verdict in the broader failure taxonomy.
The quick-reference table condenses the workflow so it can be run without re-reading the steps:
| Check | Healthy silicon | Open | Shorted | Leaky |
|---|---|---|---|---|
| Forward (red on anode) | 0.5–0.8 V | OL | Low/no drop | 0.5–0.8 V |
| Reverse (leads swapped) | OL | OL | Same as forward | Small nonzero |
| Leakage at reverse bias | nA–µA | none | none | mA |
| Loaded VF at rated current | ~datasheet | open | — | — |
| Verdict | usable with type check | replace | replace | replace if standby/battery duty |
The table makes the two fastest verdicts visible at a glance: an OL forward check means open, and a symmetric forward/reverse reading means shorted. Leakage and loaded-drop are the checks that catch the slower failures — the parts that look fine on the meter and die in service.
Borderline Cases: When Leakage Is “Within Spec” but Still Fails
The hardest cases are the borderline ones — a part whose readings are technically within a plausible range but whose behavior in the circuit was the original reason it landed in the drawer. Leakage is the classic example: a diode leaking 500 µA may be “within spec” for a part rated for microamps, yet in a standby circuit that leakage is the entire problem.
The judgment rule for a borderline part is to test it against the circuit that failed, not against the datasheet. If the diode came out of a battery-powered product, its leakage matters far more than its forward drop, and a leakage reading that is merely acceptable becomes disqualifying. If it came out of a high-current rectifier stage, the loaded forward-drop test carries the weight and a slightly leaky part may be perfectly usable.
The second borderline signature is thermal instability: a part that shifts its readings as it warms. A diode whose forward drop drifts rapidly with case temperature, or whose leakage climbs steeply as the bench warms it, is telling you its junction is degraded even when each individual reading looks reasonable. The low-leakage small-signal diode discussion explains why the family choice sets the leakage expectation, and the Schottky leakage and thermal runaway explainer shows how a warm leaky part can be a real fire risk.
Logging Failures: A Repair Notebook That Builds Expertise
The long-term value of an autopsy workflow is the log it produces. Recording each part’s visual condition, meter readings, loaded result, and final verdict builds a personal failure-pattern database that makes the next autopsy faster and the judgment sharper.
Keep the log simple and structured: part identifier (or “unmarked”), apparent type, forward drop, reverse reading, leakage at reverse voltage, loaded-VF test result, and the verdict with a reason. Note the temperature at test time, because a hot afternoon changes the readings more than most beginners expect. Over a few months, patterns appear — the same diode family failing open, one package always shorting, a marginal leak rate in standbys — and those patterns are exactly what make a repair technician valuable.
The log also feeds back into the buying decision. When the drawer of salvaged parts keeps producing the same failing family, the verdict is not “bad luck” but “this family is not right for this duty,” and the next purchase should look at the general rectifier and Schottky categories with that observation in hand. The rectifier reliability checklist turns the same logged experience into a pre-assembly audit, so the parts that pass the bench autopsy are the ones that survive the field.
A worked bench case ties the whole workflow together. A salvaged DO-41 diode with no marking shows a clean body on the visual pass. Diode mode reads 0.62 V forward and OL in reverse — healthy signs — but the leakage check shows 2 mA at 24 V reverse bias, hundreds of times what a silicon PN part should leak at room temperature. The log records it as an unmarked, leaky, unusable-for-standby part. In the drawer it would have gone back into a battery-powered product and drained it; the autopsy caught the hidden failure mode that neither the visual pass nor the basic meter checks ever would have. That is the case the workflow exists for, and it is why the leakage step is not optional when the part history is unknown.