Integrating a Microchip Laser: Drive, Timing, Thermal | ERDI

15.10.2026 г. Erdicn

Integrating a Microchip Laser Source: Drive, Timing and Thermal

*Most integration problems with a microchip laser are electrical or thermal rather than optical. This article

covers what the driver has to supply, why the drive pulse width matters more than it looks, how the timing chain

should be arranged, and what the published limits mean in practice.*

In short: the driver must be matched to the model, not designed generically. Drive current across this range

varies from 5 A to 100 A, voltage from 2 V to 14 V, and maximum drive pulse width from 0.3 ms to 4 ms. Exceeding

the drive pulse limit is the most common way to damage a source, and the optical pulse width has nothing to do

with how long the driver is switched on.

The three quantities the driver must satisfy

For every model, three numbers on the datasheet define what the driver has to do. They are not

interchangeable, and each is a limit rather than a target.

Drive pulse width — the maximum time the current may be applied. This is an energy-storage interval. The

pump diodes fill the gain medium during this window, and the Q-switch releases it at the end.

Voltage — the compliance the driver must reach, given as a maximum. The load is a diode, so the actual

voltage across it at operating current will be near the stated figure.

Current — the drive current at which the published energy is specified. Delivering less gives less energy;

delivering more risks the pump diodes and the coatings.

Model	Drive pulse width	Voltage	Current	Pulse energy
MINI-100	≤1 ms	2 V	12 A	≥100 µJ
ER100	≤2.4 ms	2 V	8 A	≥100 µJ
ER200	≤2.4 ms	2 V	10 A	≥200 µJ
ER300	≤2.4 ms	2 V	12 A	≥300 µJ
ER400	≤2.4 ms	2 V	15 A	≥400 µJ
ER500	≤2.4 ms	2 V	20 A	≥500 µJ
ER800	≤3 ms	2 V	30 A	≥800 µJ
ER900	≤3 ms	2 V	30 A	≥900 µJ
ER2000	≤4 ms	≤5 V	80 A	≥2 mJ
ER4000	≤4 ms	≤5 V	100 A	≥4 mJ
ER8000	≤4 ms	≤14 V	100 A	≥8 mJ
ER10000	≤4 ms	≤14 V	100 A	≥10 mJ
ER10	≤0.3 ms	2 V	5 A	≥10 µJ
ER20	≤0.3 ms	2 V	5 A	≥20 µJ
ER40	≤0.4 ms	2 V	5 A	≥40 µJ
ER70	≤0.4 ms	2 V	5 A	≥70 µJ

What varies, and why it matters

Read across the rows and two groups emerge, and they ask very different things of a driver.

The high-repetition family draws 5 A at 2 V. Low current, short drive pulse, high repetition rate. The

driver is a fast, modest-power design — but it has to fire up to five thousand times a second, so the switching

losses and the thermal design of the driver itself become the constraint.

The high-energy models draw 80 to 100 A, up to 14 V. That is over a kilowatt of instantaneous input. The

driver needs substantial energy storage, heavy current paths and careful layout. In exchange it fires a few

times a second, so average dissipation is manageable.

The middle tiers sit between: 8 to 30 A at 2 V, with drive pulses of 1 to 3 ms.

The practical consequence is that a driver designed for one tier will not serve another. A design intended

to cover a product line should either be specified across the full range or deliberately restricted to one tier.

Drive pulse width is not optical pulse width

This is the single point of confusion most worth clearing up, because getting it wrong damages hardware.

The drive pulse is the electrical pulse from the driver. It lasts hundreds to thousands of microseconds —

0.3 ms to 4 ms across this range — and its job is to pump energy into the gain medium.

The optical pulse is the light that comes out. It lasts nanoseconds, and it is produced when the saturable

absorber, acting as a passive Q-switch, suddenly releases the stored energy.

The two differ by a factor of several hundred thousand. For the ER800, the drive pulse is up to 3 ms and the

optical pulse is 7 ns.

Three consequences follow:

Lengthening the drive pulse does not lengthen the optical pulse. The optical pulse width is set by the

cavity — the length of the gain medium, the absorber and the coatings. It is a fixed property of the model. This

is why the same model keeps its pulse width whatever the driver does, and why pulse width cannot be tuned from

the electronics.

Exceeding the maximum drive pulse is a real risk. The pump energy has to go somewhere. Held on too long, it

becomes heat in the gain medium and stress on the coatings, in a package only a few millimetres across.

The optical pulse width is not a design variable at integration time. It is chosen when the model is chosen.

If the resolution requirement needs a different pulse width, that is a different model — the reasoning is set out

in pulse width and range resolution.

Matching the driver to the model

Current source, not voltage source

The load is a laser diode stack, so the driver must regulate current, not voltage. A voltage-mode supply will

deliver whatever current the diode's forward characteristic allows at that voltage, which varies with

temperature and between parts. The stated current must be regulated directly.

Energy storage close to the load

At 100 A, the current path is the design. Every centimetre of loop between the storage capacitor and the laser

adds inductance, which lengthens the rise time and wastes energy as heat. The storage should be as close to the

source as the layout allows, with a short, wide return path.

Rise time

The drive pulse has to reach full current in a time short compared with the pulse itself, or part of the pumping

interval is spent ramping rather than pumping. At 0.3 ms and hundred-amp currents, that is not automatic and

needs designing for.

Repetition rate and duty

The maximum repetition rate differs by tier: 10 Hz for the compact and standard models, 5 Hz for some, up to

5 kHz for the high-repetition family. Two limits apply at once:

Average power. Drive pulse width × current × voltage × repetition rate gives the electrical power going

in. What is not emitted as light becomes heat, and the heat has to leave the package.

Thermal recovery. The gain medium must return to its operating condition between pulses. Firing faster

than it can recover reduces the energy per pulse, and the published minimum energy is specified at the stated

conditions.

For the high-repetition models the duty is low — 0.3 ms at 5 kHz is a duty cycle of 1.5% — but the average power

still matters because the package is 17×8×7 mm³.

Thermal design

The gain medium is phosphate glass, which conducts heat roughly an order of magnitude less well than a crystal

host. That has a direct consequence for integration: **the heat path is the integrator's responsibility, and it

is the constraint that limits repetition rate more often than the electronics do.**

Three points to design for:

Mount to a thermal mass, not to air. The mounting ears on the package are the intended heat path. They

should contact a surface with real thermal capacity, through a compliant interface material if the surfaces are

not flat.

Do not insulate the package. Potting compounds and conformal coatings that are convenient elsewhere may

reduce the heat path more than expected in a package this small.

Check the temperature at the source, not at the board. The published operating range is −40 °C to +65 °C for

the low-repetition models and −45 °C to +65 °C for the high-repetition family. What matters is the temperature of

the gain medium during operation, which in a small package can rise well above the ambient the sensor reads.

The mechanical drawing for each model is published on its product page, and the

3D model can be used to plan the mounting and the thermal interface in CAD before hardware

exists.

Timing and the receiver chain

Trigger and delay

The Q-switch fires when the stored energy reaches threshold, not at a fixed delay after the drive pulse begins.

The delay from the start of the drive pulse to the optical output is therefore not a fixed constant — it varies

with temperature, with pump energy and between units.

Two ways to handle it:

Detect the optical pulse with a photodiode tap and start the time measurement from there. This removes the

jitter from the measurement, at the cost of a small optical pickoff.

Characterise the delay over temperature and compensate in firmware. Simpler optically, but the compensation

table has to be built for the specific unit.

Which to choose depends on the resolution the instrument needs. If the pulse is 4 ns and the timing chain

resolves to 10 ns, the answer will be dominated by the electronics regardless of which scheme is used.

Receiver bandwidth

The detector and analogue chain must preserve the pulse. As a working rule, the bandwidth should be comfortably

more than the reciprocal of the pulse width:

Optical pulse width	Bandwidth to aim for
3.5–4 ns	500 MHz or more
5–6 ns	300–400 MHz
7 ns	~300 MHz
12–13 ns	~200 MHz

A receiver faster than needed collects more noise bandwidth and makes detection harder. A receiver slower than

needed discards the resolution the laser provides. Matching the two is worth doing deliberately.

Time-to-digital conversion

The TDC resolution should be finer than the pulse width, or the measurement is limited by the electronics. For

the 3.5 ns MINI-100 that means sub-nanosecond resolution; for the 12 ns high-energy models, considerably less

demanding.

What to check before design freeze

Driver current and voltage against the model's published figures, not against a similar model.
Maximum drive pulse width, enforced in hardware if possible rather than only in firmware.
Repetition rate and duty, against both the average power and the thermal recovery of the gain medium.
Thermal path from the mounting ears to a real thermal mass, measured at the source.
Trigger scheme — optical pickoff or characterised delay — decided against the resolution requirement.
Receiver bandwidth matched to the optical pulse width.
TDC resolution finer than the pulse width.
Operating temperature confirmed at the gain medium, over the full ambient range the instrument will see.

Next steps

Every model page publishes its datasheet, which carries the full electrical and environmental specification, and

a 3D model for mechanical integration — see Laser Sources. Dimensioned drawings and the

controlled STEP source are released on request so the correct revision is issued.

For integration questions, include the model, the supply available and the operating conditions in a

Contact / RFQ enquiry. Where a standard model does not fit, the customisation scope

— package, optical parameters, electrical interface and environmental requirements — is described under

OEM / ODM.

Обратно към блога