Peak power, pulse width and duty cycle: the three numbers that define a pulsed laser diode

The relationship

Three numbers, one identity:

Everything else follows. Pulse energy sets reach. Duty cycle sets heat. Pulse width sets resolution.

Peak power is a marketing number

A 180 W diode and a 90 W diode at the same 100 ns are very different; a 180 W diode at 50 ns and a 90 W diode at 100 ns are the same system - both deliver 9 uJ. Peak power alone tells you nothing about range.

This matters commercially, because datasheets lead with peak power. When you compare parts, convert to energy first. Our own catalogue does this: every pulsed laser diode page lists pulse energy at 100 ns next to the peak power figure.

Pulse width: resolution, blind zone, and driver cost

Range resolution is set by pulse width:

For t_p = 100 ns, ΔR = 15 m. For 10 ns, 1.5 m. For 1 ns, 15 cm.

The same number sets the receiver bandwidth you need (BW ≈ 0.4 / t_p), which sets your noise floor. It also sets the blind zone - a receiver cannot see a return until its own pulse has finished, so a 100 ns pulse cannot measure anything closer than about 15 m without special techniques.

Consequences:

Note the trade: a narrower pulse gives you resolution and a worse noise floor. At 1 ns you are integrating 100x more noise bandwidth than at 100 ns, which costs roughly 10 dB of sensitivity. That is why the range equation holds energy, not peak power, as the figure of merit: a 1 ns pulse at 100 W gives you the same reach as a 100 ns pulse at 1 W, at 100x the driver difficulty.

Duty cycle: the number that decides survival

The catastrophic failure mode in pulsed laser diodes is not peak power - it is average power. A TO56 can with a 110 x 30 µm emitting area can dissipate a few hundred milliwatts of heat. Push it further and the facet coating degrades, threshold current climbs, and the part dies in tens of hours instead of tens of thousands.

Worked example with PLD905-180W-M:

Comfortable. The datasheet allows up to 0.1 %, which at 3 kHz means you could stretch the pulse to 300 ns - but then you would be running 54 mW of average power and a longer pulse. Most designs sit well inside the rating and treat the remaining duty cycle headroom as thermal margin.

What breaks designs:

Forgetting that duty cycle scales with both variables. Doubling the repetition rate and keeping the pulse width doubles the average power.

Ignoring wavelength drift with temperature. At 0.28 nm/K, a 40 K rise moves the peak wavelength 11 nm - enough to walk off the edge of a 10 nm bandpass filter and silently lose half your signal. Low-drift parts like PLD905AL-70W-M hold 0.10 nm/K specifically for this reason.

Assuming the receiver filter is wide enough. It usually is not. Match the filter passband to the diode's temperature-widened spectrum, not to its 25 °C value.

Practical selection rules

Start from energy, not power. Decide the range you need, convert through the range equation, then choose a pulse width from the resolution requirement, and only then read off the peak power you need.

Pick the longest pulse your resolution allows. It is easier to drive, needs less receiver bandwidth, has a better noise floor, and puts less stress on the driver electronics.

Keep average power at half the rated duty cycle. It buys thermal margin for high-temperature operation and for the wavelength drift that comes with it.

Check the emitter count and aperture. The same 75 W is available as three 200 x 10 µm emitters or four 90 x 18 µm emitters. The first gives you a wide, thin line; the second a more compact spot. That choice interacts with your collimation optics - see beam shaping.