Transistors are the diodes that never quite behave like diodes. A transistor datasheet describes a three-terminal device with gain, saturation, and switching behavior, and the multimeter test that works for a plain diode only takes you part of the way. Still, a surprising amount of transistor diagnosis reduces to one model — the transistor seen as two diodes sharing a terminal — plus one extra test that proves the part actually amplifies. This guide maps the base, collector, and emitter across common packages, walks the five diode-mode checkpoints of an NPN, then adds the hFE test that turns a “possibly working” verdict into a confident one.
Why the Two-Diode Model Explains the Pin Test (and Where It Breaks)
An NPN transistor’s three junctions look like two diodes if you squint: the base-collector junction and the base-emitter junction both behave as diodes in the forward direction, sharing the base as a common terminal. In a diode-mode test, the collector and emitter should each read as a diode against the base, and the collector-emitter pair should read nothing at all, because there is no direct junction between them. That model predicts the whole pin test and explains why a transistor that “tests bad” often tests bad in a specific pattern.
The model breaks precisely where the transistor’s real behavior begins. A working transistor conducts between collector and emitter when the base is driven, which the two-diode model cannot represent — the model sees an open circuit there and calls it good. That is why the diode-mode checks are a necessary filter, not a verdict: a transistor that fails them is certainly bad, but one that passes them has only proven it is not obviously destroyed. The gain test in a later section supplies the missing proof.
The limit also means a transistor on a populated board can fool the diode checks in the same way a diode can — parallel paths and the base drive circuit corrupt the readings. The rule from the rectifier testing discipline applies unchanged: test bare, out-of-circuit parts for a verdict; treat in-circuit readings as candidates to be re-measured on the bench.
Mapping EBC Across Common Packages: TO-92, SOT-23, and SOT-223
The base, collector, and emitter are labeled on every datasheet, but the physical pinout varies by package, and the mapping is the first thing to get wrong. Three packages cover most bench work.
| Package | Typical pin order (left to right / top row) | Notes |
|---|---|---|
| TO-92 | E, B, C | The classic through-hole transistor; center pin is the base |
| SOT-23 | B, C, E (left to right) | Small-signal SMD; base on the left pin |
| SOT-223 | B, B?, C, E | Larger SMD; check the datasheet, several conventions exist |
The TO-92 with the center base is the easiest to remember and the most common in repairs; the SOT-23’s left-base convention is consistent enough for a first test but should still be confirmed against the package drawing, because manufacturers differ. The SOT-223 and other larger SMD families vary enough that the datasheet drawing is the only safe reference — the same rule that applies to SMD diode polarity, with the same consequence: a rotated part fails the very test that would have identified it.
The pin map is also the place where the two-diode model earns its keep. Once the three terminals are identified, the diode-mode checks test each junction against the map, so a mislabeled pin produces test results that contradict each other — a useful signal that the pin map is wrong before any part is condemned.
Diode-Mode Sequence: The Five Checkpoints of an NPN
The diode-mode test of an NPN runs five checkpoints, each mapping to a junction in the two-diode model.
First, check base-collector forward: red lead on base, black on collector, expect a forward drop of roughly 0.5–0.7 V for silicon. Second, check base-collector reverse: swap leads, expect OL. Third, check base-emitter forward: red on base, black on emitter, expect the same diode-like reading. Fourth, check base-emitter reverse: expect OL. Fifth, check collector-emitter in both directions: expect OL both ways, because the working junction there is not a plain diode.
Checkpoints one through four prove the two base junctions conduct in the forward direction and block in reverse. Checkpoint five confirms there is no direct collector-emitter short, which is exactly what a failed transistor develops when the junction collapses. A part that passes all five has no open junction and no collector-emitter short, and is a strong candidate for being good — pending the gain test.
The same five-point sequence, with the red-lead assignment reversed, tests a PNP, which is one reason the NPN/PNP distinction matters before you start; the NPN vs PNP guide covers the polarity differences that this test relies on. For an NPN, base-to-collector and base-to-emitter are the two diodes, and the order above simply walks both junctions in both directions.
hFE Test: Using Transistor Gain Reading as a Second Opinion
The diode checks cannot see whether a transistor amplifies, and that is where the hFE (beta) feature on many multimeters earns its place. In the hFE/diode-gain socket, the meter drives the base with a controlled current, measures the resulting collector current, and displays the ratio — the DC current gain of the part.
The interpretation rule: a healthy small-signal NPN shows a substantial gain reading, often in the tens to hundreds, while a degraded or counterfeit part shows a much lower number or an unstable one. The absolute value matters less than the comparison against the part’s typical range and against a known-good sample, because beta depends on the test current and the meter’s socket conditions. The gain reading is the “second opinion” that the diode checks cannot supply — a part that passes all five diode checkpoints but shows collapsed gain is damaged in a way the junction tests miss.
The hFE measurement carries one caveat that keeps it honest. The meter’s gain socket drives the base at its own fixed test current, so the displayed beta is a sample at that specific operating point — not the gain at your circuit’s real base current. A transistor’s beta varies with current, temperature, and age, so a reading of 150 at the meter’s test point says nothing precise about gain at 50 mA of base drive. What the test does say, reliably, is whether the amplification path is present and healthy; a collapsed reading is conclusive damage, while a healthy reading earns a trial in the circuit. Treat the hFE socket as a go/no-go amplifier probe, not as a datasheet measurement, and the interpretation never misleads.
Real bench work can stop at the hFE test for most parts. The combination — five diode checkpoints plus a gain reading — distinguishes a good part, a dead part, and a degraded part with confidence, which is exactly the verdict a repair bench needs before swapping transistors. The transistor selection guides on the site cover the family-level questions this test informs, and the same two-step protocol applies across the small-signal transistor range.
The part-level knowledge behind the tests is worth a read alongside: the digital switching transistor explainer covers the family’s switching role and the general purpose PNP page handles the polarity mirror. The same protocol scales to other NPN families with one modification: the pin map changes with the package, not the test. A SOT-23 NPN uses the SOT-23 pin order from the table above, and the five checkpoints run identically once the map is correct; a larger power NPN may show a different forward-drop band on the base junctions at the meter’s test current, so the healthy-window comparison should use a known-good sample of the same family rather than the general 0.5–0.7 V silicon figure. The method is package-independent exactly because it is anchored to the two-diode model rather than to a parts list.
Bench Case: Distinguishing a Good 2N2222 from a Dead One
A worked case ties the protocol together. Two 2N2222 transistors arrive from the same drawer: one from a known-good supplier, one suspected dead.
On the good part, the five checkpoints read 0.6 V forward on base-collector, OL reverse, 0.6 V forward on base-emitter, OL reverse, and OL collector-emitter both ways; the hFE socket reads about 150. Every checkpoint passed and the gain is in the healthy range — the part is good, and the verdict matches its history.
On the suspected part, checkpoints one through four read normally — the base junctions conduct — but checkpoint five shows a small forward reading collector-to-emitter in one direction. The two-diode model flags it immediately: there should be no direct collector-emitter conduction, and the presence of a reading there indicates an internal short between collector and emitter. The hFE test would settle it, but the checkpoint five failure already convicts the part. A transistor that shorts collector-to-emitter in one direction has a damaged internal structure and belongs in the bin, regardless of the healthy-looking base junctions.
The case illustrates the real value of the protocol: the diode checks localized the failure to the collector-emitter path, and the two-diode model explained why that reading is conclusive. The same sequence works for any NPN in a drawer full of unknowns, and it is the first step before any replacement part is ordered from the transistor categories. The SOT-23 bipolar transistor explainer covers the package whose pin map this guide tables, and the general purpose NPN article provides the family-level context behind the test.