How to choose a 905 nm pulsed laser diode

905 nm is the workhorse of time-of-flight. It is cheap, it is mature, and it sits under the 1400 nm retinal-damage threshold that lets you pump serious peak power into a handheld optic. The families here span 15 W to 600 W, in plastic and metal cans, bare and collimated. That range is the problem: picking the wrong one is a board spin, not a parameter tweak. This guide is the order in which those decisions should be made.

1. Peak power: the lowest that meets your range budget

The temptation is to buy the highest wattage on the shelf. It is the wrong instinct. Peak power sets two things you pay for whether you use them or not: drive current and thermal load. A 600 W emitter at 55 A and 80 V is a different power supply and a different heatsinking problem from a 75 W part at a few amps.

As a rule of thumb by application:

• 15–75 W — handheld rangefinders, golf lasers, robot-vacuum obstacle avoidance. This is the volume centre of the market.
• 70–140 W — automotive and industrial LiDAR, long-range scanning.
• 180–600 W — airborne mapping, multi-stripe and stacked emitters where aperture and range both push high.

Specify the power against your range equation with margin for fog, optics loss and end-of-life degradation — then stop. More is not better; it is just hotter.

2. Package and footprint: plastic, metal, or collimated

Within the same 905 nm emitter family, the package is the biggest cost and reliability lever. From cheapest to most sealed:

• Plastic two-pin (P) — cost-optimised, silver lead frame. Every handheld rangefinder and golf laser on the market uses this format. Lowest price, lower solder-temperature ceiling.
• Metal TO56, three-pin (M) — hermetically sealed, isolated header, high-temperature solder. The default for LiDAR modules and industrial rangefinders.
• Metal TO56, two-pin (MP) — same metal can as a drop-in for a plastic two-pin, adding high-temperature solder without a footprint change.
• Fast-axis-collimated (MCL) — a cylindrical lens built into the can cuts the fast axis from 22–28° to 6–8°, simplifying your receive optics.
• Aspheric-lens / collimating-pin (MS, MZS, MZ) — a 3° × 5° beam at lower cost than a FAC stack; confirmed at enquiry because they sit outside the standard price list.

The emitter is the same wavelength family in each. The package is a manufacturing and reliability decision, not a photon decision — which is why a plastic part and a metal part can carry the same peak power and the same emitting area.

3. Beam divergence: match it to your receiver

A bare die emits at roughly 28° (fast axis) by 10° (slow axis). That is fine if your collimating and focusing optics were designed around it. If they were not, you are choosing a collimated build, not a different diode:

• Bare — 28° × 10°. Lowest cost, you do the optics.
• Fast-axis collimated (FAC) — 8° × 10°. Removes one barrel lens from your assembly.
• Aspheric collimated — 3° × 5°. The tightest beam; lowest component count downstream.

The far-field is the number your optical engineer actually cares about, and it is confirmed on samples against your receiver — the datasheet headline is a starting point, not a sign-off.

4. Drive conditions: pulse width is the constraint

These are pulsed parts. The published peak power assumes a pulse condition — typically 100 ns at a low duty cycle (ROHM's RLD90QZW3, for example, is rated at 50 ns, 0.05% duty). Two things follow:

• Pulse width and duty cycle set your thermal ceiling. Push either and the same die derates differently.
• Threshold current is low (sub-amp on the 75 W parts) — the drive design is a current source, not a voltage source.

Send us pulse width, repetition rate and peak current, and the sample set comes back against those conditions — not against a generic datasheet pulse.

5. Temperature drift: when range must hold across the band

905 nm wavelength drifts with temperature — typically 0.2–0.3 nm/°C on a bare part. For most rangefinders that is invisible. For airborne mapping and long-range scanning, where the receiver filter is narrow and the operating band is wide, it is not. The low-temperature-drift parts (TC < 0.1 nm/°C, case at negative potential) exist precisely for that case. If your spec sheet has a wavelength-stability line, that is the part you want; if it does not, the standard drift is fine and cheaper.

The 905 nm vs 1550 nm choice you will also face

905 nm is not the only choice for time-of-flight, and the alternative is worth naming honestly. At 1550 nm the permitted eye-safe exposure limit is orders of magnitude higher than at 905 nm — the cornea absorbs the longer wavelength instead of focusing it onto the retina (see the MPE tables in IEC 60825-1). That lets a 1550 nm system run higher average power for the same safety class. The cost is price and maturity: 1550 nm and 1535 nm erbium-glass sources are more expensive per watt and the ecosystem is smaller. The decision is a safety-budget trade, not a performance one — 905 nm wins on cost and supply, 1550 nm wins on eye-safety headroom. Choose on where your product's safety classification lands.

If you are replacing an existing part

Two cases come up constantly, and both have a dedicated page rather than a paragraph:

• The ams OSRAM SPL PL90_3 is discontinued (last orders 1 April 2024). The pin-to-pin drop-in is PLD905-75W-P.
• The ROHM RLD90QZW3 is active but single-sourced. The functional second source is PLD905-75W-M.

Both map to the same 905 nm family — the package and the supply story are what differ.