A pulsed laser diode is not an LED you drive with a resistor. It is a current device with a low forward voltage, a sharp threshold, and a hard ceiling on pulse width and duty cycle. The diode does the photon work; the driver's job is to feed it a fast, flat, well-protected current pulse and keep the thermal envelope. Most field failures we are asked to diagnose are driver problems, not diode problems — so this is where the design effort should go.
What the driver must do
- Source current, not voltage — regulate the peak current into a low-impedance load.
- Switch fast and flat — sub-nanosecond to few-nanosecond edges, minimal overshoot, low jitter.
- Protect the junction — clamp reverse voltage and ESD; no inductive kick.
- Hold the thermal envelope — respect pulse-width and duty-cycle derating.
1. Current source, not voltage source
The forward voltage of a pulsed 905 nm edge emitter is low — roughly 2 V for a single emitter, higher for multi-junction high-power parts — and it has a threshold current below which it is just a diode and above which it lases. A fixed voltage with a series resistor does not hold current steady as the junction heats (the voltage actually falls with temperature), so current climbs. That is how a board that is fine cold destroys itself warm. The driver regulates current: a sense resistor, a comparator or controller, and a switch that holds the set pulse flat regardless of the diode's V-I curve.
2. The loop is the hard part
At tens of amps and nanosecond edges, parasitic inductance is the enemy. Every centimetre of lead is inductance that rings and overshoots when the switch turns off, and that overshoot is what avalanches the facet. The rules are unglamorous and non-negotiable:
- Minimise loop area — the diode, switch, sense resistor and decoupling capacitor sit in one tight loop.
- Stiff local decoupling — a low-ESR capacitor bank right at the diode so the pulse is drawn from the cap, not from a distant supply trace.
- Clamp the reverse side — a Schottky diode across the laser diode so any inductive kick is caught at a fraction of a volt instead of breaking down the junction.
- Snub the gate — controlled gate drive avoids the overshoot that comes from slamming a big MOSFET on or off.
3. Edges and the ToF timing budget
For time-of-flight, the rise and fall time of the optical pulse is part of your timing budget. A slow, rounded edge adds uncertainty to the flight-time measurement. That is where the switch choice bites: silicon MOSFETs are cheap and switch the current well at wider pulse widths, but their gate charge limits the fastest edges. GaN FETs switch faster and with lower loss, at the cost of a tighter layout and a higher part price. Pick the switch for your pulse width and repetition rate, not for the diode's headline number.
4. Thermal derating is not optional
The published peak power assumes a pulse condition — typically 50–100 ns at a low duty cycle (ROHM's RLD90QZW3, for example, is rated at 50 ns and 0.05 % duty). Push the pulse width or the repetition rate and the same die derates differently, because average power — not peak — sets the junction temperature. The driver has to enforce that envelope: limit duty cycle, monitor case temperature, and back off current as the can warms. A driver that lets a customer run 1 % duty when the diode is rated for 0.1 % is a driver that ships returns.
Driver spec checklist
The parameters that should be fixed before a diode is chosen, not after:
| Parameter | Why it matters | Typical range |
|---|---|---|
| Peak current | Sets optical output with the diode | 2 – 40 A (up to 100 A+ stacked) |
| Pulse width | Hard ceiling on safe operation | 5 – 100 ns (some to 200 ns) |
| Repetition rate | Drives average power and heating | kHz – MHz |
| Rise / fall time | Part of the ToF timing budget | < 1 – 5 ns |
| Reverse-voltage clamp | Protects the facet from kick / ESD | Schottky, < 1 V drop |
Send us pulse width, repetition rate and peak current and the sample set comes back against those conditions — not against a generic datasheet pulse. The diode and the driver are one optical source, and we quote them that way.
The bundled source + driver offer
LumeLiDAR quotes the emitter and the driver as one optical front-end rather than two line items. The reason is practical: a 905 nm emitter is only as good as the pulse it is fed, and a driver is only as good as the diode it drives. For the replacement programs — the RLD90QZW3, SPL TL90AT03 and SPL UL90AT08 second sources — the same logic applies: the sample set is qualified against your actual drive conditions, not a nominal pulse.
Frequently asked
Why can't I drive a pulsed laser diode like an LED?
Because its forward voltage is low and it has a threshold. A few volts of over-drive pushes current far past the safe peak, and a single over-current pulse can facet-damage the die. A laser diode is a current device with a hard ceiling on pulse width and duty cycle, so the driver has to be a regulated current source with fast edges and hard protection — not a voltage source through a resistor.
What kills a laser diode in the field?
Over-current and reverse voltage, in that order. Reverse bias above a couple of volts (and ESD) avalanches the junction; an inductive kick from a poorly laid-out switch ring above the rating. The fixes are a Schottky clamp across the diode, a low-inductance loop, and gate drive that does not overshoot. Thermal runaway from ignoring the duty-cycle derating is the slower, quieter failure.
Do I need a GaN FET in the driver?
For sub-5 ns edges and clean timing it helps. Silicon MOSFETs switch the current fine at higher pulse widths, but their higher gate charge and slower edges limit the fastest ToF systems. GaN gives the edge speed and lower loss; the cost is a tighter layout. Match the switch to your pulse width and repetition rate, not to the datasheet headline of the diode.
Related reading
How to choose a 905 nm pulsed laser diode
Five decisions — power, package, beam, drive and temperature drift — that pick the right emitter for LiDAR, rangefinders and sensors.
905 nm vs 1550 nm for LiDAR: the real trade-offs
Safety budget, receiver cost and BOM trade-offs between the two dominant LiDAR wavelengths.
VCSEL vs EEL for ToF: which emitter to choose
Where each emitter wins on beam shape, wafer-level test and peak power for time-of-flight systems.
Send us the pulse you need to hit.
Peak current, pulse width, repetition rate and package is usually enough to quote a matched emitter and driver — and to say whether a sample is available from stock.