A fast recovery diode is selected by its reverse recovery time (trr), and that number only means something inside a switching context: at high frequency, the charge stored in a PN junction must be cleared at every switching edge, and the clearing costs power, noise, and stress. This guide explains the recovery waveform, the loss formula, and the selection margins for high-frequency rectifier and freewheeling stages.
Why Recovery Time Matters at High Frequency
At 50/60 Hz, the stored charge in a PN junction has all the time in the world to clear, and recovery is an afterthought. At 100 kHz it repeats one hundred thousand times per second, and the effects stop being theoretical. Every time a conducting diode is forced into reverse bias, its stored minority carriers must be swept out before the junction can block voltage. The removal is not instant: the diode conducts briefly in reverse, current reverses sharply, and voltage snaps back with an overshoot that shows up on the drain-to-source waveform of the adjacent switch.
Three consequences matter to a power design:
- Efficiency. The recovery transient dissipates energy in the diode and in the switching transistor at every edge.
- EMI and ringing. A fast current reversal excites the parasitic inductance of the layout and produces damped oscillation that shows up in conducted and radiated emissions.
- Stress. The reverse-recovery current spike adds to the switch current, raising turn-on stress and junction temperature.
That is why a diode that is perfectly acceptable at line frequency becomes the wrong part in a 100 kHz output stage. The datasheet VF did not change; the switching context did.
Anatomy of a Reverse Recovery Waveform
The recovery waveform is the exclusive starting point for every fast-recovery selection. When the diode current ramps from forward (IF) to reverse at a controlled rate (di/dt), the current does not stop at zero. It overshoots in the reverse direction, reaches a peak reverse recovery current (Irrm), and only then returns toward zero as the junction regains blocking capability.
The datasheet divides that waveform into two intervals:
- ta (charge storage interval): current falls from zero to −Irrm while stored charge is removed; the diode still conducts.
- tb (recovery interval): current returns from −Irrm to zero as the depletion region rebuilds; the rate of this return defines the recovery characteristic.
- trr = ta + tb: the total reverse recovery time, stated in nanoseconds (ns).
- Qrr (recovered charge): the integral of the reverse current over the whole recovery interval; this is the number that drives energy loss.
Two parts with the same trr can behave very differently, because trr says nothing about how the current returns to zero. The shape of the return—abrupt or gradual—is the difference between hard and soft recovery, and it is often more important than the headline time.
The datasheet measurement states the forward current before the transition, the applied di/dt, and the temperature; recovery time depends on all three, so compare parts only at identical conditions. A number quoted at 25 °C and a gentle di/dt cannot be compared directly with one quoted at 125 °C and a fast transition.
Hard Recovery vs Soft Recovery: What the Spec Sheet Hides
Hard recovery means the reverse current collapses abruptly at the end of tb. The diode blocks quickly, but the sudden current change couples into layout inductance and produces a sharp voltage overshoot, ringing, and a noisier switching node. Soft recovery means the current returns gradually, trading a slightly longer recovery for controlled dv/dt, lower overshoot, and less EMI.
The spec sheet usually lists trr and Qrr, but the softness factor—the ratio of the two interval slopes—may not be printed. Two rules cover most designs: if the node is EMI-sensitive or the layout inductance is high, prefer a soft-recovery part and confirm the characteristic with the manufacturer; if the converter is hard-switched at very high frequency and losses dominate, the shorter trr matters more and the ringing must be managed in the layout or with a snubber.
The Switching Loss Formula, Worked Example
The recovery energy can be estimated as the recovered charge times the reverse voltage across the diode, repeated at the switching frequency:
Psw ≈ Qrr × VR × f, or equivalently Psw ≈ Vr × Irrm × trr × f / 2
Both forms are first-order estimates; the first uses the integrated charge, the second a triangular current approximation. Worked example: a fast recovery diode in a 400 V stage with Qrr ≈ 375 nC at the operating temperature, switching at 100 kHz, contributes roughly:
0.375 µC × 400 V × 100 kHz = 15 W
That is 15 W of commutation-path switching loss that never appears in the VF table. The actual split between the diode and the adjacent switching device depends on the circuit waveform—the estimate sizes the recovery event, not the diode’s dissipation alone. At 50 kHz the same event costs about 7.5 W; at 20 kHz about 3 W. Frequency multiplies the recovery loss linearly, which is exactly why the “efficient diode” conversation always starts with the switching frequency. The same estimate also explains why the recovery term belongs in the loss budget: it adds to the conduction loss in the commutation path, and the diode’s share must be included when sizing its heat-flow route.
Recovery in Context: Where Each Family Stands
The recovery characteristic is what separates the families at the switching node. A Schottky has no PN-junction minority-carrier storage, so it has essentially no PN-style reverse recovery; its turn-off behavior is dominated by junction capacitance and displacement current. A fast recovery diode has a controlled but finite recovery charge. A SiC Schottky combines the no-recovery behavior with a high voltage class. The full family comparison—voltage, leakage, capacitance, surge, and cost—belongs to the family selection guide; this article’s job is the recovery waveform and its loss.
The temperature dimension belongs to this article: recovery time and recovered charge grow as the junction warms, so an FRD that settles comfortably at 25 °C can ring at the hot working condition. Read trr and Qrr at the operating junction temperature—not the 25 °C number—and re-check the recovery characteristic after the thermal design closes. Softness is not always printed on the datasheet; when the node is EMI-sensitive, ask the manufacturer for the recovery waveform or a softness factor at the operating condition.
Selecting an FRD: Voltage, Current, and trr Margins
Three margins close the selection:
- Voltage. Apply 20–30% margin over the maximum repetitive reverse peak, including transients. The recovery overshoot itself adds voltage stress, so margin the same way you would for any rectifier.
- Current. Size IF(AV) at the actual case temperature, not the 25 °C headline, and add the recovery loss to the thermal budget. A part that passes the current check on paper can still fail when the recovery term is added at the hot junction.
- Recovery time. Keep trr small relative to the switching period—a common engineering habit is trr below roughly 10% of the off-time—and match the softness to the EMI budget. For a hard-switched 100 kHz stage, that points to a part whose recovery settles well inside the dead time.
Read trr at the operating junction temperature: recovery time and recovered charge grow as the junction warms, so the 25 °C datasheet number is the optimistic one.
Softness deserves its own margin in EMI-sensitive designs: a part with an acceptable trr but an abrupt recovery can fail the emissions test where a softer part passes, so confirm the recovery shape with the manufacturer whenever the switching node is noisy or the layout inductance is high.
Good-Ark Fast Recovery Portfolio
Good-Ark’s fast recovery rectifier diodes category lists the family with trr, IF(AV), IFSM, VF, and package columns, so the selection above maps directly onto the catalog: filter by voltage class and trr before opening individual datasheets. The official FRD Application Guide on the site covers the same waveform and loss analysis in application form. For part numbers not fully specified on the site, contact Good-Ark sales with your voltage, frequency, and EMI requirements to confirm the datasheet and qualification documentation. The catalog rows also carry the package and thermal data needed to complete the selection—RθJC, mounting style, and derating—so the trr decision is checked against the case-temperature budget in the same step.
Engineering note. The waveform model above—ta, tb, trr, Irrm, and Qrr—follows the standard reverse-recovery measurement used in fast-recovery datasheets, and the loss estimate Psw ≈ Qrr × VR × f is a first-order engineering approximation for selection, not a substitute for a full loss model. trr and Qrr are temperature-dependent; read them at the operating junction temperature and confirm the softness characteristic with the manufacturer before finalizing the BOM.
Frequently Asked Questions
What is reverse recovery time (trr)?
It is the time a diode takes to clear stored charge and regain blocking after being forced into reverse bias, divided into a storage interval and a recovery interval. It is stated in nanoseconds and is the most-quoted number for fast recovery diodes.
Why does recovery loss grow with frequency?
The recovery transient happens at every switching edge, so its energy repeats at the switching frequency: Psw ≈ Qrr × VR × f. A 15 W recovery loss at 100 kHz becomes 7.5 W at 50 kHz, which is why frequency is the first question in any FRD selection.
What is the difference between hard and soft recovery?
Hard recovery returns the reverse current to zero abruptly, causing voltage overshoot and ringing; soft recovery returns gradually, trading a slightly longer recovery for controlled dv/dt and lower EMI. Choose by the node’s EMI budget and layout inductance.
Can a fast recovery diode replace a standard silicon diode?
Only where the circuit needs the faster recovery and the voltage, current, and thermal ratings fit. In a 50/60 Hz rectifier there is no recovery benefit, and the FRD usually costs more—match the device to the switching context.
When should I choose a Schottky instead of an FRD?
When the Schottky meets the reverse voltage, leakage, junction capacitance, surge, and thermal gates, it can win in both hard- and soft-switched stages because it has no PN-style recovery. The FRD becomes the alternative when the Schottky’s voltage class, leakage at temperature, surge capability, or cost does not fit. The full family decision is covered in the dedicated comparison guide.
How much trr margin should I leave?
Keep trr small relative to the switching period—roughly below 10% of the off-time as a starting habit—and verify at the operating junction temperature, because trr and Qrr grow as the junction warms.
Conclusion
Fast recovery selection is frequency-first: quantify the recovery loss at the real switching frequency, read trr and Qrr at the operating junction temperature, match the softness to the EMI budget, and apply the same voltage and thermal margins you would use for any rectifier. The datasheet number that matters most is the one at the hot working condition, not the 25 °C headline.
Browse the fast recovery rectifier diodes category to compare parts by voltage class and trr, and contact Good-Ark with your operating voltage, frequency, and EMI requirements to confirm the part, datasheet, and samples.