← back to sabrinaeppensteiner.com
built & running

personal project · 2025–2026

Constant-Current Driver for a Diode Laser

I wanted a laser that can cut paper and burn things, so I got a 7 W blue laser diode. It needs a constant current to run, and I built the driver for it myself. Simulated in LTspice, soldered by hand.

roleCircuit design, simulation, build, testing

toolsLTspice, soldering iron, multimeter

statusBuilt and running, more measurements to come

The problem

If you just put a voltage on a laser diode, it warms up, draws more current, gets even warmer and eventually dies. That's why it needs a driver that holds the current constant.

Mine is a NUBM44 (450 nm, about 7 W) in a copper module. It runs at around 3 A.

The hand-soldered driver board, on perfboard, with a potentiometer for current adjustment

The driver on perfboard. The blue knob on the left sets the current.

The circuit

Three parts: a reference voltage, a divider that sets the target, and a control loop that holds the current there.

LTspice schematic of the constant-current driver: a 1N750 Zener reference on the left, an RC-filtered 11k/1k divider in the middle, an LT1006 op-amp driving an IRFP250N MOSFET, and a 0.1 ohm shunt with the laser diode in the drain path on the right. R4 is the potentiometer, stepped through five values.

scroll sideways to read the schematic

Simulated with an IRFP250N, the closest LTspice model to the IRFP260N on the board.
01 · referenceV2 · R2 · D2: a 4.7 V Zener gives a stable reference.
02 · targetR3 · R4 · R5 · C1 · C2: 11 kΩ and the 1 kΩ knob divide it down to 0 to 392 mV.
03 · control loopU1 · C3 · R6 · M1 · R1: the op-amp compares the target with the voltage on the 0.1 Ω shunt and adjusts the MOSFET until they match.
turn the trim This is what the blue knob does.
divider · R40.75kΩ
target across R1300mV
current through the diode3.00A

Calculated from the part values, not measured.

Simulated current through the laser diode over the first 20 ms after switch-on, for five potentiometer settings. It settles at 0.41, 1.03, 2.02, 2.96, 3.87 A. The 750 ohm setting, about 3 A, is the typical operating point. The lasing threshold lies at 0.15 to 0.35 A. lasing threshold 0.15–0.35 A 01234 A05101520 ms typical operating point · 3 A 0.41 A · 100 Ω1.03 A · 250 Ω2.02 A · 500 Ω2.96 A · 750 Ω3.87 A · 1000 Ω

hover or tap the plot to read off the current

Simulated switch-on for five knob settings. The current settles without overshooting. Low settings start later because the MOSFET gate has to charge up first. The dashed line follows the knob above.

Heat. At 3 A about 20 W end up in the MOSFET, so it needs a big heatsink too.

Stability. C3 and R6 slow the control loop down so it doesn't oscillate. D3 protects the laser against reverse voltage.

Getting it built

Soldering the board, with the schematic on the laptop.
01

Simulate first

I checked in LTspice that the current settles where it should.

02

Solder the driver

Everything by hand on a perfboard. The 10-turn knob lets me set the current afterwards.

03

Mount the diode

The copper module sits on a heatsink with a fan, otherwise it would overheat.

04

Tune and test

Turned the current up slowly and checked it with a multimeter.

What it's made of

Laser

  • NUBM44 diode, 450nm 7 W class
  • 12×30mm copper module
  • Heatsink with fan

Driver

  • LT1006 op-amp
  • IRFP260N power MOSFET
  • 0.1 Ω shunt
  • ZPD4.7 Zener reference
  • 12 V battery, 10 A fuse
  • Perfboard, hand-soldered

Design

  • LTspice simulation
  • 10-turn potentiometer trim
  • RC-filtered reference

Handling it safely

Laser class

7 W is class 4. Even reflections can hurt your eyes.

Eyewear

450 nm safety glasses on, always, before switching on.

Current limit

The knob can't go above about 3.9 A, and there's a 10 A fuse.

Beam block

The beam ends on a beam block. I only test briefly, with nobody else in the room.

What comes next

Set vs. actual current

Measure the real current over the whole knob range.

Drift

See how much it changes while everything warms up.

Switching on

Check for overshoot on the oscilloscope.

First cuts

So far it only works on paper. Next, it should actually cut some.