Laser safety classes for ranging products: how the eye-safety limit overrides your energy budget

The component is not the product

Every one of our 905 nm datasheets carries the same notice, and it is the most important sentence in the document:

Depending on the operating mode, these devices may emit highly concentrated invisible infrared light that may be hazardous to the human eye. Products that incorporate them must comply with the protective requirements of IEC 60825-1, Safety of Laser Products.

Read that carefully, because it is a statement about where responsibility sits. A laser diode is a component. It has no class. The classification is a property of the finished product - the diode, its driver, its optics, its housing and its firmware, evaluated together. The same 180 W diode can sit inside a Class 1 product or a Class 4 product. Nothing about the diode changed; the optics, the pulse width, the repetition rate and the enclosure did.

That is why this article is written to the product designer rather than the component buyer.

What the classes mean in practice

The classes run from least to most hazardous. IEC 60825-1 is the international standard; individual regions adopt it with local variations.

905 nm is invisible, so Class 2 is not available to you. A 905 nm product falls into Class 1, 3R, 3B or 4, and the difference between them is mostly a matter of how much energy arrives at an eye under the standard's assumed conditions.

What sets the permitted energy

The standard defines an accessible emission limit (AEL) for each class. The AEL is not one number; it depends on all of the following:

• Wavelength. The eye's transmission and the retina's absorption vary strongly across the spectrum, and the limits vary with them.
• Exposure duration. The permitted energy for a single pulse depends on how long that pulse lasts, and it does not fall in proportion to duration. The relationship is not something to memorise: look it up for your pulse width in the current edition of the standard.
• Pulse structure. A pulsed source must satisfy the limit twice - once for the energy in a single pulse, and once for the pulse train considered as a whole. For a nanosecond source the single-pulse test is usually the binding one.
• Measurement geometry. Classification is defined in terms of what passes through a specified aperture at a specified distance from the source, not the total power emitted. A wider beam and a smaller beam can therefore have different limits at the same power.
• Apparent source size. How large the source looks from the measurement position matters, and the limits are stricter for small apparent sources.
• Collection by optics. The assessment must also hold when the beam is collected by a lens - the standard's way of representing binoculars or a telescope. "The beam diverges so much that nobody could get it all into their eye" is not an argument that survives the standard.

Two consequences follow for 905 nm designs:

Collimation makes classification harder, not easier. A fast-axis collimator concentrates the beam and reduces the apparent source size. Both effects push toward a lower permitted energy. A design that passes as Class 1 with a bare emitter can fail once a FAC is added - see FAC and fibre coupling. This is a real and commonly missed interaction between the optical design and the regulatory one.

Short pulses do not escape the limit. A 100 ns pulse is a very short event, but the standard evaluates it as it is - it does not average it away against the time between pulses. This is why short-pulse 905 nm systems are limited to small pulse energies, and why the limit, not the diode, is usually what caps the range.

Why 1535 nm is different

At 1535 nm the cornea and the lens absorb the light before it reaches the retina. The maximum permissible exposure is correspondingly higher - roughly a million times higher in terms of radiant exposure, which is why the phrase "eye-safe wavelength" exists at all.

That difference is the entire reason the LZ1535 family can put 40 µJ, 300 µJ or 1000 µJ into the air from a 25 mm or 40 mm aperture and still be specified as an eye-safe transmitter, while a 905 nm design is fighting for single microjoules. It also explains the trade: an eye-safe transmitter needs an InGaAs detector, a slower repetition rate, and a solid-state laser instead of a diode.

None of that makes a 1535 nm product automatically Class 1. It moves the limit by orders of magnitude; it does not remove it.

Practical design rules that reduce the burden

1. Start the safety assessment before the optical design is frozen. The aperture position, the housing geometry and the collimation all change the class, and retrofitting controls is expensive.
2. Put the exit aperture where an eye cannot reach it. A recessed window, a baffle, or a beam that exits upward or sideways removes the worst case entirely.
3. Use the shortest pulse your resolution allows, and the lowest energy your range equation allows. Both directions reduce the hazard, and both are usually cheaper than adding Class 3B controls.
4. Consider the failure modes. A stuck driver that holds the diode on, a broken window, a lens that falls off - the standard requires you to consider reasonably foreseeable conditions, and a driver fault that turns a 100 ns pulse into a continuous waveform is a foreseeable condition. It is also a thermal failure mode for the diode, so the protection serves both purposes.
5. Re-assess after any optical change. Adding a collimator, changing a window material or replacing a diffuser changes the classification.
6. Measure the finished product. Classification is performed on the complete unit, at the worst-case operating point, with a calibrated instrument, by someone who knows the standard. Do not infer it from the laser's datasheet.
7. If you ship Class 3B or 4, do the paperwork properly. Warning labels, apertures, key control, interlocks, eyewear specified by wavelength and optical density, and a person responsible for the safety programme.

When the limit forces a different design

If the energy you need exceeds what your target class permits, the options are, in the order we usually see them work out:

1. A larger receive aperture. Range goes as the square root of aperture area, and this costs nothing in laser safety. Doubling the diameter of the receive lens buys 1.41x the range and no change in class.
2. A better receiver. Moving from PIN to APD, or from a 5 mm to a 16-channel array, buys sensitivity that can be traded for lower transmit energy. See APD versus PIN.
3. Multi-pulse averaging. Averaging N returns improves signal-to-noise as the square root of N, so range can be bought with time instead of energy.
4. Duty-cycle economics. If the product can accept a slower update rate, more averaging per measurement is usually cheaper than more energy.
5. Move to 1535 nm. This is the honest answer when the requirement is genuinely multi-kilometre, and no amount of 905 nm engineering will close the gap. It costs the detector, the repetition rate and the source complexity - and it buys the energy budget.

This page is engineering background, not a classification. The standard is the authority, and every product must be assessed on its own hardware. Where this article and the current edition of IEC 60825-1 disagree, the standard wins.