Tutorials
Which Way Does a Diode Conduct? Symbols, Polarity and Meter Clues
Published · Updated
A diode symbol tells you about direction, but it does not promise a universal forward voltage or explain every meter reading. This lesson connects the drawing, the physical identification record and a deliberately limited circuit model. Its practical output is an evidence tree: what can you infer from two low-voltage diode-mode observations, and which questions remain unanswered? No particular reading is presented as proof that an unknown board is safe or functional.
Find the anode and cathode from a real record

A basic diode has an anode and cathode. In the conventional schematic symbol, the cathode is associated with the bar. For ordinary forward conduction, the anode is at a sufficiently positive potential relative to the cathode under the operating conditions. This description establishes a reference direction; it does not supply a complete current-versus-voltage curve or an allowable operating current.
Connect the schematic to the physical component using its actual documentation. Some axial diode bodies use a band to identify the cathode, but a body appearance is not a universal pinout rule for every diode package or array. Multiple-diode packages may have common terminals or other connections. A package drawing, marking definition and exact code provide a stronger identification than a photograph of a vaguely similar component.
Nexperia's archived 1N4148/1N4448 datasheet, dated 10 August 2004 is an example of a document that separates outline, symbol, marking and operating tables. It is used here to teach document reading, not to establish current supply or recommend that historical record for every new design. Keep the revision and applicable type visible when using any such drawing.
A reference designator such as D4 identifies a position, not a diode technology. Trace its two nets and find the matching BOM line. If the symbol, orientation mark and BOM disagree, stop the identification rather than choosing whichever record seems convenient. A reversed assembly, an incorrect library symbol or an outdated drawing calls for a different corrective action, and a meter alone may not distinguish those causes.
Replace the perfect one-way switch with a stated model
The ideal one-way-switch model is helpful for learning which paths can conduct, but it omits the voltage and current behavior of a real junction. Another simplified model assigns a fixed forward drop. That can support a first arithmetic example if the assumed drop is stated clearly. It should not become a claim that every silicon diode always drops one familiar voltage at all currents and temperatures.
Real forward voltage is associated with current, temperature and the particular device. Datasheet entries may be specified maxima at a test point, while curves may describe typical behavior. Those are different kinds of evidence. A typical curve can help understand a trend without guaranteeing every unit at every condition. The relevant question is which source information supports the actual operating point you intend to use.
Reverse bias is also more than an open switch. Leakage, capacitance and breakdown conditions can matter. An ordinary rectifier, a Zener reference and a TVS protection device are not interchangeable simply because all are called diodes. Their intended operating regions and rating conventions differ. An introductory symbol lesson should point you toward those distinctions instead of presenting a single generic diode specification.
Use the simplest model that answers the current question, then name what it leaves out. For tracing a possible path, an ideal model may suffice. For estimating resistor current, a declared forward-drop assumption may be appropriate. For thermal loss, switching or protection behavior, more evidence is needed. Changing the model is a normal part of engineering reasoning, not evidence that the earlier drawing was dishonest.
Follow one source-limited example without claiming a reference design
Consider an illustrative 5-volt source, a 1000-ohm series resistor and one forward-oriented diode returning to the source. Assume a 0.7-volt diode drop solely for this model. The remaining 4.3 volts appears across the resistor. The calculated current is 4.3 milliamperes. The resistor dissipates approximately 18.49 milliwatts, and the diode's simplified conduction dissipation is approximately 3.01 milliwatts. These numbers come from the assumptions, not from a bench test.
Reverse the diode in the ideal blocking model and the intended forward path stops. The real circuit still needs a reverse-voltage and leakage check. The example does not authorize applying an arbitrary supply to an unknown diode. In particular, the series resistor and source conditions are part of the explanation. Describing the diode alone hides the part of the circuit that limits current.
Now suppose the resistor were accidentally replaced by a much smaller value. The simplified current estimate would rise, but the assumed forward drop might no longer fit. Source limits, heating and component ratings could become important. A calculator that keeps the same drop throughout would continue producing numbers even after its assumptions became poor. The lesson is to inspect the model boundary before trusting an apparently precise output.
For a board with programmable logic, a protection diode cannot be selected from this arithmetic alone. Pin absolute limits, allowed injection current, power sequencing and the protection event need exact evidence. Consult the XC7A100T-1CSG324C record only if it matches the controller in the project. It is an entry to the device evidence, not an endorsement of an external clamp or this circuit.
A meter reading includes the meter and the rest of the board
Diode mode applies a limited test stimulus and reports a response according to the meter's design. Test current and available voltage differ between instruments. A reading therefore describes the junction under that stimulus, not its forward voltage under every application load. The actual meter manual should be part of the measurement plan, particularly for LEDs, stacks or parts that require more compliance voltage.
Use only a de-energized, appropriately discharged low-voltage educational circuit for this introductory exercise. Confirm absence of voltage according to the instrument instructions. Fluke's diode-test guidance (source address: https://www.fluke.com/en-sg/learn/blog/digital-multimeters/how-to-test-diodes) explains that board connections can affect interpretation and that isolating a device end may be necessary. This lesson does not replace procedures for high-energy or hazardous equipment.
Parallel resistors, another junction, a charged node or a damaged connection can change the observation. Record both lead orientations and the relevant nodes. If the part remains in circuit, label that condition explicitly. A two-direction reading on a populated board is not automatically a failed diode, and two open readings are not automatically an open junction. The observations must be interpreted with the circuit and instrument.
Before removing or lifting any part, consider whether the operation is authorized and whether it could damage the board or lose evidence. A repeatable unpowered observation is often a better next step than immediate rework. Keep the original orientation and reference designator in the record. The board bring-up checklist supports this broader habit of separating observations, hypotheses and corrective actions.
Try the evidence tree: inconclusive is a useful answer
The interactive tree accepts a forward-direction observation and the opposite-direction observation. Enter a numeric reading in volts or OL for an over-range indication. State whether parallel paths have actually been excluded. The tool intentionally avoids a universal pass voltage because it has no exact device, temperature or meter specification. It organizes the evidence rather than pretending to diagnose an unknown component.
Static reference: If parallel paths are not excluded,the result remains inconclusive. Interactive fields appear when JavaScript is available. The explanation and example remain readable without it.
If the device is not isolated from other paths, the result remains inconclusive regardless of how familiar the numbers look. If both directions show OL, the tree lists possible open connection, device behavior or inadequate meter stimulus without selecting one cause. If both directions give readings, it prompts a check of connections, polarity identification and possible parallel or internal paths. It does not convert two numbers into a guaranteed failure report.
One direction giving a numeric reading while the other shows OL is consistent with asymmetric behavior at that meter setting. It does not establish reverse breakdown voltage, current capability, switching speed or successful operation when hot. That distinction is particularly important when a board fault appears only under load. An unpowered junction observation and a functional stress test answer different questions.
Use the copyable result as a measurement note, then add the actual meter model, range, device identification, board revision and conditions. If you cannot supply one of those fields, keep it unresolved. Good troubleshooting narrows the next question. It does not require every preliminary check to produce an immediate pass or fail. The tree's most useful output may be the reason another piece of evidence is needed.
Choose the next diode lesson by the circuit job
Rectification, switching, voltage reference and transient suppression are separate next topics. A rectifier lesson follows load current and reverse stress. A switching-diode lesson considers charge and timing. A Zener lesson needs a current-dependent reference model. A protection lesson must define the event and the protected node. Grouping all four under one generic diode recommendation would conceal the conditions that make the selection useful.
A circuit drawing is also not an assembly footprint. Confirm the actual package view, pad numbering and orientation convention before releasing files. The PCBA assembly guidance is relevant when those records must become a build package. The engineering workbench and tutorial directory provide other verified paths as the task moves beyond junction identification.
Do not infer that one diode is a direct replacement for another from its forward voltage or package alone. The schematic role, reverse conditions, current waveform, thermal environment and switching behavior may all change the decision. A proposed replacement should carry an explicit comparison and unresolved checks. If those records are absent, the correct status is a candidate for review, not a qualified alternative.
The practical output of this lesson is a marked drawing, an identified source record, a declared circuit model and a two-orientation observation sheet. Those four items support a more informed next step. They also expose the gaps that a supplier or project team needs to resolve before procurement or assembly. A neat symbol or plausible meter number on its own is a much weaker handoff.
Ask for the right quotation after the evidence is organized
Try a hypothetical evidence exercise before requesting a replacement. A board observation gives 0.62 volts in one lead orientation and 0.59 volts in the other. Those two numbers alone do not identify an exact diode or prove a short. First ask whether parallel paths were excluded, where the probes touched, which meter range was used and what the design record says about the package. The evidence tree deliberately keeps this result unresolved rather than inventing a diagnosis.
Now consider a different hypothetical observation on an isolated identified device: 0.65 volts in one orientation and OL in the other. This supports a narrower statement about asymmetry under that test stimulus. It still does not establish the device's leakage at a higher reverse voltage or its behavior after heating. A parts request can include the observations, but the selection and acceptance criteria need the actual application requirements and source data.
When reporting an intermittent problem, preserve when it occurs and what changed between observations. A thermal or load-dependent symptom may not appear in a room-temperature diode-mode check. A connection disturbed during probing can also change the result. The goal is to retain enough context for another person to reproduce the question, rather than to make the report sound certain. That discipline reduces unnecessary rework and makes an eventual comparison or replacement review more meaningful.
For components, request a diode quotation through Contact with the exact manufacturer ordering code, quantity, destination and required date. Preserve packaging and traceability requirements when relevant. If the identity is unresolved, describe the board position and missing evidence rather than guessing a suffix from an appearance. Stock, price and lead time require a current response and are not implied by an educational example.
For a board-level issue or new assembly, submit a project quotation request with board quantity, schematic and BOM revision, the circuit job, observed behavior and requested test scope. Keep confidential files out of URLs. The requester should be able to distinguish a parts purchase from component identification, design review or assembly troubleshooting, since those scopes require different work and deliverables.
The decision tree is intentionally conservative about missing evidence. Use its boundaries as part of the lesson: direction is a starting point, a meter is a limited observer, and an application circuit must still be assessed in its own operating conditions.
Show the original diode concept animation

Concept animation, not a hardware recording or a validated simulation. Close this disclosure to hide it. Static steps and numerical assumptions are explained in the article.
Back to Tutorials