905 nm vs 1535 nm vs 1550 nm: choosing a wavelength for ranging and LiDAR

Why wavelength is the first decision, not the last

Engineers often pick a diode by peak power and treat wavelength as a consequence. It is the other way round. Wavelength determines four things that constrain everything downstream:

1. Which detector you can use, and therefore the noise floor of your receiver.
2. How much energy you are legally allowed to emit, via the eye-safety standard.
3. How much of your light survives the atmosphere - and whether rain is your enemy or your friend.
4. What the target looks like at that wavelength (foliage, camouflage, glass, water all behave differently).

The three bands at a glance

905 nm: the workhorse, and the cheap one

905 nm is where the volume is. A stacked-tunnel-junction edge emitter in a TO56 can - our PLD905 series runs from 15 W to 180 W - costs a fraction of anything at longer wavelengths, and it pairs with a silicon APD that costs a couple of dollars.

The advantage is not just component cost. Silicon APDs have very low dark current, which sets a low receiver noise floor, and silicon has excellent responsivity out to about 1000 nm. That combination is why a 180 W / 18 uJ part can reach 1.6 km on a reflective target.

The limits are real, though:

Eye safety. 905 nm is invisible and focuses on the retina. Above a few microjoules the finished product is Class 3R or worse, and the beam has to be engineered so it cannot reach a human eye - shutters, interlocks, divergence, or a mechanical housing the beam physically cannot escape. This single constraint stops most high-energy 905 nm designs.

Energy ceiling. Because you cannot emit more energy safely, 905 nm tops out around 60 uJ in a practical handheld device. That caps diffuse-target range at roughly 1 km. See the range equation article for where that number comes from.

Sunlight. 905 nm sits in a strong solar background band. A narrow optical filter (5-10 nm) is mandatory; without it, daytime range collapses. It also means 905 nm is the wrong choice for a receiver looking into the sun-facing direction.

A subset of the supply chain - notably around 905 nm VCSEL arrays - now reaches 70 W in a surface-emitting device, which removes the fast-axis collimation problem entirely. That is covered in VCSEL vs EEL.

1535 nm: the eye-safe band for real range

At 1535 nm the cornea absorbs before the light reaches the retina, so the maximum permissible exposure is roughly a million times higher than at 905 nm in terms of radiant exposure. Practically, that means you can put 100 uJ, 300 uJ or 1 mJ into the air without making the product a laser hazard.

That is why every serious multi-kilometre handheld rangefinder is eye-safe. Our LZ1535 erbium glass modules span 40 uJ at 1 kHz up to 1 mJ at 10 Hz, and a 300 uJ module reaches about 10 km on a retro-reflector and 3.5 km on diffuse terrain using the calibration in the range equation article.

The costs are equally clear:

Detector. InGaAs APDs are 5-10x the price of silicon, need temperature compensation (their breakdown voltage drifts), and have higher dark current and excess noise. Your receiver budget grows.

Source. Erbium glass is a passively Q-switched solid-state laser, not a diode. It is bigger, slower (10 Hz to a few kHz, not 30 kHz), and needs its own power module - see drivers and power modules.

Weather. 1535 nm sits inside a water absorption band. Fog and heavy rain attenuate it noticeably more than 905 nm. In clear conditions it is excellent; in coastal fog it can be the wrong choice.

Repeat rate. 1 kHz is achievable for scanning LiDAR, but 10 kHz and above is not, which rules 1535 nm out of high-frame-rate flash architectures.

1550 nm: the fibre-laser wavelength

1550 nm shares the eye-safety advantages of 1535 nm and adds one more: it is the wavelength at which erbium-doped fibre amplifiers and telecom-grade components exist in vast quantity. If you need tens of watts of average power - for long-range automotive LiDAR, for example - 1550 nm fibre lasers are the only practical route.

For pulse energies under about 100 uJ, and for anything that has to fit in a handheld housing, 1535 nm usually wins: erbium glass is a simpler, smaller, cheaper source than a 1550 nm fibre MOPA chain.

A decision procedure

If the product is a handheld or consumer unit under about 1 km, use 905 nm. You get the cheapest optics, the cheapest detector, and a supply chain with dozens of qualified sources.

If the requirement is beyond 1.5 km on real terrain, use 1535 nm. The eye-safety margin is what makes the energy budget possible at all. There is no clever 905 nm design that closes that gap.

If you need more than 10 W of average power, or an all-fibre architecture, use 1550 nm.

If the application is 3D sensing at a few metres to tens of metres, use a VCSEL at 850-940 nm. Circular beams, wafer-scale manufacture, and no eye-safety headache at the energies involved.

The one case where 905 nm beats everything

Short range, high repetition rate, low cost. A robot vacuum, a docking sensor or a 20 m proximity detector running at 30 kHz is a 905 nm design, and will stay one. The PLD905-25W-P exists for exactly this: 25 W, three emitters, plastic package, cents not dollars.

Use the selector to check a specific combination against the catalogue, or read the driver design article before you commit to a pulse width.