Every power supply, every charger, every board has one — and most engineers pick it from habit, not from the spec sheet. This guide covers every diode family we stock and support: standard rectifiers, fast and ultrafast recovery, Schottky, zener, ESD protection, and optocouplers — 27 part numbers across 6 families.
From what we see across Shenzhen distribution (2025–2026), the diode selection error rate in field returns is higher than any other component class we handle.
The failures cluster into five patterns, all covered below — and almost all trace back to one question answered wrong: how fast does the circuit switch?
The good news: diode selection is a five-question decision, not a datasheet hunt. This guide walks it end to end.
Answer these three questions before touching a datasheet — they eliminate most of the field.
| Question | If Yes | If No |
|---|---|---|
| Does the circuit rectify mains or 50/60Hz power? | Standard rectifier (1N4007-class) — recovery speed is irrelevant at line frequency | Switching circuit → recovery speed is now the top spec |
| Does the load need the lowest possible forward drop? | Schottky (1N5819, SS34, MBRS-class) — but check reverse leakage | Ultrafast or standard, depending on frequency |
| Is the function signal protection or reference, not power rectification? | Zener / ESD / optocoupler family — different job entirely | Keep going down the rectifier path |
Three answers, three directions. The rest of this guide makes each branch precise.
| Family | Model | Key Specs | Package | Best For |
|---|---|---|---|---|
| Standard Rectifier | 1N4007 | 1A / 1000V | DO-41 | Mains rectification, line frequency |
| 1N4007GP | 1A / 1000V, glass passivated | DO-41 | Same, higher reliability tier | |
| Fast / Ultrafast | FR107W | 1A / 1000V fast recovery | SOD-123 | SMD high-voltage input rectification |
| UF4007 | 1A / 1000V, 75ns | DO-41 | SMPS input bridge | |
| MURA110 | 1A / 100V, 30ns | SMA | Low-voltage high-frequency output | |
| MURA260 | 2A / 600V ultrafast | SMA | Flyback output rectification | |
| MURS360B | 3A / 600V, 75ns | SMB | Higher-current output rectification | |
| Schottky | 1N5819 | 1A / 40V, low VF | DO-41 | Through-hole low-voltage rectification |
| SS34 | 3A / 40V | SMD | Low-voltage, higher current | |
| MBRS130L | 1A / 30V, 395mV | SMB | Minimum-loss low-voltage rails | |
| MBRS3200 | 3A / 200V, 840mV | SMB | High-voltage Schottky territory | |
| BAT54A | Dual, common anode | SOT-23 | Signal clamping and logic protection | |
| Switching | 1N4148W | 100V switching diode | SOD-123 | Logic and signal switching |
| Zener | BZX84-C2V4 | 2.4V, 300mW | SOT-23 | Low-voltage SMD reference |
| BZT52H-C16 | 16V, 375mW | SOD-123F | SMD reference, automotive-qualified | |
| BZX79-B4V7 | 4.7V, 500mW, ±2% | DO-35 | Through-hole reference and repair | |
| 1SMA5925B | 10V, 1.5W | SMA | Power zener clamping | |
| 1SMA5942B | 51V, 1.5W | SMA | High-voltage zener protection | |
| ESD | ESD200-B1 | ±5.5V, 6.5pF | CSP0201 | Signal-line ESD protection |
| Optocoupler | PC817 | 5000Vrms isolation | DIP-4 | Feedback and logic isolation |
| SFH615A | 5300Vrms, graded CTR | DIP-4 | Higher isolation, graded transfer ratio |
Read the table top to bottom as a speed ladder: standard rectifiers handle line frequency; fast and ultrafast parts handle switching supplies; Schottky handles low-voltage efficiency; the last three families are different jobs — references, protection, and isolation.
1N4007-class parts exist for one job: rectifying line-frequency AC. At 50/60Hz, recovery speed is irrelevant — the reverse voltage changes slowly enough that any diode turns off in time. What matters is peak reverse voltage (1000V covers 240VAC mains with margin) and surge capability.
The classic error is taking this part into a switching supply.
Put a 1N4007 in a flyback output and it heats up or fails — the recovery time is in the microsecond range against a 100kHz switching period. That single mistake is the most common rectifier failure in field returns we see.
Recovery time (trr) is the spec that matters, and it ranges from 75ns down to 30ns in this family. UF4007 and MURS360B sit at 75ns; MURA110 at 30ns.
The faster the switching frequency, the more the recovery charge hurts efficiency and heats the diode — at 100kHz+ you want the 30ns class.
Voltage matters too: MURA110 covers 100V output rails, MURA260 and MURS360B cover 400–600V flyback outputs. The same circuit at 5V and 48V needs different recovery parts, not just different packages.
Recovery time (trr) — longer bar = slower turn-off:
What does a microsecond of recovery cost at 100kHz? Ten switching periods of heating — the standard rectifier simply never turns off in time. Schottky turns off instantly because there's no stored charge to remove.
Schottky diodes win on forward drop and lose on leakage and voltage. The MBRS130L's 395mV drop against a standard rectifier's ~1V is a real efficiency gain at low voltage — but Schottky reverse leakage climbs hard with temperature, and the voltage ceiling sits well below silicon's.
Use Schottky where the output voltage is low and the current is high — 5V, 12V, and battery rails. Past roughly 150–200V, the leakage and the weak avalanche behavior make ultrafast silicon the better choice — MBRS3200 at 200V is about the practical ceiling of this family.
A zener makes a voltage reference, not a power regulator. The 500mW BZX79-B4V7 clamps fine at 5mA of test current; it cannot hold 4.7V while delivering 100mA to a load.
For real regulation, use an LDO (AMS1117-class) — for a reference or a clamp, the zener family is cheaper and simpler than anything else.
The package ladder mirrors the other families: BZX79 (DO-35) for through-hole and repair, BZX84 (SOT-23) for dense SMD, BZT52H (SOD-123F) for the SMD default, and the 1.5W 1SMA59xx parts when the clamp actually absorbs power.
Three different jobs, three different specs. The ESD200-B1 protects signal lines with 6.5pF of load; the 1N4148W switches logic signals at speed; PC817 and SFH615A cross an isolation barrier.
None of them rectifies power — and reaching for a rectifier when you need any of these is the other common selection error.
Five questions, one part: rectification speed decides standard vs fast; voltage decides silicon vs Schottky; signal work goes to the special families. What's the fastest path through this tree? Knowing the switching frequency — it rules out more families than any other number.
1. Standard rectifiers in switching circuits. A 1N4007 in a flyback output rectifies on paper and cooks in practice — its recovery time is measured in microseconds against a 100kHz switching period. This is the most common rectifier failure in field returns, and the fix is a 30–75ns ultrafast part.
2. Schottky voltage margin ignored. Schottky diodes have no avalanche tolerance worth discussing — voltage spikes kill them dead. A 40V Schottky on a 24V rail with a 50V spike is a dead diode; derate to 2× the nominal rail voltage and check the spike.
3. Zener asked to regulate power. A zener is a reference, not a regulator. At 100mA of load current the zener voltage drifts with its own heating and impedance. If the load is real, the LDO is the part — the zener family stays for references and clamps.
4. Recovery speed overkill. A 30ns ultrafast part costs more than a 75ns part and gains nothing at 60kHz. Match the recovery time to the switching frequency — 75ns covers most flybacks, 30ns is for the fast edge cases.
5. Package power assumed, not calculated. A 1.5W zener in an SMA package dissipates 1.5W only with the right pad. The same die in the same package family derates differently — read the thermal derating curve, not the headline number.
Diodes are the most counterfeited component class, and the fakes are cheap to make. The usual trick: a relabeled or undersized die inside a standard package — the marking says 1N4007, the die inside is a 0.5A part.
From what we see across Shenzhen lots (2025–2026), diode counterfeits cluster in the volume SKUs: 1N4007, 1N5819, SS34, and the zener references.
Test the measurable, not the label. Forward voltage at rated current catches undersized dies — a fake 1N4007 reads high on VF or dies on surge. Zener voltage at the 5mA test current catches relabeled zeners.
Recovery time is the hard one to test without a scope, which is exactly why fast-recovery fakes are the most dangerous — the 75ns spec is where they fail.
Volume pricing across the diode families is more compressed than engineers expect. Standard rectifiers price in the low cents; ultrafast and Schottky parts sit a tier up; the 1.5W zener and CSP0201 ESD parts carry the premium of the group.
Across all families, the spread is roughly an order of magnitude — and the cheapest part in the wrong family costs more in rework than the right part ever did.
Contact ICMASS for current pricing on your specific quantity — volume pricing on the whole family set (rectifier + zener + optocoupler for a PSU feedback loop) ships together from Shenzhen.
A: Frequency first, voltage second, efficiency third. Line frequency → standard rectifier. Switching supply → fast/ultrafast with trr matched to the frequency. Low-voltage, high-current, efficiency-critical → Schottky. Voltage above ~150V pushes back toward ultrafast silicon.
A: No — that's the classic rectifier failure. The 1N4007's recovery time is in the microsecond range against a 100kHz switching period; it overheats or fails. Use a 75ns-class ultrafast part (UF4007, MURS360B) or a 30ns part for fast edges.
A: Yes — leakage is their trade-off for low forward drop. Reverse leakage climbs exponentially with temperature, which matters in warm enclosures and battery circuits. The 395mV drop of an MBRS130L saves power at the output; the leakage shows up when the circuit is hot and idle.
A: Match trr to the switching frequency — 75ns covers most flybacks, 30ns is for fast edges. Buying 30ns when 75ns suffices adds cost without benefit; buying 75ns for a 500kHz converter adds loss. The recovery charge, not the recovery time alone, is what heats the diode — datasheets give both.
A: Only for tiny, constant loads — otherwise use an LDO. A 500mW zener holds its voltage at the 5mA test current; at 100mA it drifts with heating and impedance. For real regulation, the AMS1117-class LDO does the job with proper line and load regulation.
A: Energy class — ESD diodes handle human-body discharges, TVS handles surges. The ESD200-B1's 6.5pF protects signal lines from IEC61000-4-2 events; a power TVS (SMAJ-class) absorbs surge energy but loads high-speed lines with its capacitance. Signal line → ESD diode; power input → TVS.
A: CTR grade, isolation voltage, and speed — in that order. The CTR grade determines whether your feedback loop works over temperature and aging; isolation voltage must cover the safety requirement; and the 30kHz-class bandwidth limits these parts to slow signals. PC817 and SFH615A differ mainly in isolation (5000 vs 5300Vrms) and CTR grading.
A: Power and assembly — the package derates the die. DO-41 through-hole handles ~1A-class with a lead as the heatsink; SMA, SMB, and SMC are SMD steps with increasing pad area and power. A 3A part needs SMB/SMC; a 1A part fits SMA. The footprint is a thermal decision, not just a mechanical one.
A: Measure the electrical, don't trust the marking. Forward voltage at rated current catches undersized dies; zener voltage at the 5mA test current catches relabeled parts; recovery time is where fast-recovery fakes fail. Volume SKUs like 1N4007 and 1N5819 are the most counterfeited.
A: At least 2× the nominal peak — more for switching circuits. A 240VAC mains rail peaks near 340V, which is why the 1000V-rated 1N4007 is the standard choice. In switching supplies, the flyback voltage spikes add to the rail — the 600V and 1000V ultrafast parts exist for exactly that reason.





