I have a simple circuit that uses a MOSFET to switch on and off a ~100W LED spotlight (30V, 3A). It works for this application.

I have been asked to modify the circuit so that the LED can be dimmed instead, but without using PWM (the reasons are beyond the scope of this question, but I can explain if needed).

So my naive solution was to drive the gate of the MOSFET with the output of a DAC, per the schematic below:


simulate this circuit – Schematic created using CircuitLab

The DAC I am using is the MCP4725 (specifically the breakout board from Sparkfun). The MOSFET is a logic level one, IXFP5N50P3, and it is properly heatsinked.

After some testing with the DAC output at voltages other than 0 or 5V, I notice the MOSFET starting to fail. In some cases they end up letting some current flow even if the gate voltage is zero. In other cases they fails by never letting enough current flow for the LED to reach full brightness.

I suspect I am using the MOSFET in a mode of operation that is not safe, but I am not sure why, or how to tell by looking at the datasheet.

Is this MOSFET only usable for switching a load, and if so are there any other parts that would be suitable for this application ?

  • 1
    \$\begingroup\$ If the MOSFET works well for turning the LED on and off but fails when you try to dim by controlling the gate voltage then the cause of failure is most likely heat. Convince us that you have provided an adequate heatsink and show your calculations of the junction temperature. Also, provide a link to the datasheet for the LED. \$\endgroup\$ Aug 15, 2018 at 11:32
  • 2
    \$\begingroup\$ When not using PWM your MOSFET will need to operate as "variable resistor". It will then 'burn' all the unused power and get very, very , very hot. Besides a lot more difficult to control, compare it so start/stopping a rope or slowing it down, letting it slide through your hands. \$\endgroup\$
    – Oldfart
    Aug 15, 2018 at 11:36
  • \$\begingroup\$ Look at the datasheet of the MOSFET: Fig. 1. Output Characteristics @ TJ=25ºC. At Vgs=5V: Id=0.4A, so the FET's Vgs(th) is way too high. Your LED will never turn fully on, even at maximum DAC output. \$\endgroup\$
    – MartinF
    Aug 15, 2018 at 11:47
  • 2
    \$\begingroup\$ "I have been asked to modify the circuit so that the LED can be dimmed instead, but without using PWM (the reasons are beyond the scope of this question, but I can explain if needed).", I would love to hear the reason. I doubt that it is justified. \$\endgroup\$ Aug 15, 2018 at 19:12
  • \$\begingroup\$ @HarrySvensson the LEDs are used to provide patterns of light on an object that is filmed by a high speed camera. The system has 30 LEDs and each needs to be individually addressable. PWM'ing this many channels at the right speed did not seem doable with an Arduino. Perhaps with a different MCU, like a Parallax, it might be possible. \$\endgroup\$ Aug 16, 2018 at 8:23

1 Answer 1


It is not recommended (especially for 3 A) to turn an LED off and on for dimming purposes. You should use a constant current LED driver that delivers the current in continuous conduction mode (CCM). CCM means that the LED current never goes to zero between switching cycles. An inductor is required.

enter image description here

enter image description here

This is a simple Buck CC LED driver circuit using a 3 Amp Didoes Inc. AL8849.
Max current is set using RSET.
You would use the CTRL pin for PWM dimming

enter image description here

The easiest and arguably the best way to do this would be to use an AC powered $45 Mean Well 42V, 120 Watt HLG-120H-42B Constant Current Driver with three dimming methods utilizing a single pair of wires. This driver provides 93% efficiency from wall to LED.

This is a high quality driver with a 7 year warranty. There are cheaper knock off drivers available as well. I recommended the 42V where the constant current range is 12V-42V and max current is 2.9A. The 36V HLG-120H-36B is 18V-36V up to 3.4 A.


  1. 0-100K Resistance
  2. 0-10VDC
  3. 10V PWM

Dimming methods 2 or 3 would work in your application.

enter image description here

  • \$\begingroup\$ using a MeanWell driver I would need a control signal in the 1-10V range which will require additional circuitry to convert the 0-5V signal I have available from Arduino+DAC. So it's probably simpler to use an AL8843 as you suggest, though I am worried that it won't be enough as in the datasheet it claims "up to 60W of output power" and the LEDs require more than that (around 100W) \$\endgroup\$ Aug 21, 2018 at 22:09
  • 1
    \$\begingroup\$ If you look at the block diagram there is only a MOSFET between SW and ground. That MOSFET is rated at 3A. I do not see how the output voltage would put limitations on the 3A. I have not yet used this chip personally so I cannot test and verify for you. The 60W "output power" does conflict with the 3A and 40V spec so I understand your concern. There are not many 3A drivers with an internal MOSFET available. I see it as a matter of time. Do you have the time to build a prototype and test. If not then a driver with an external MOSFET, like an LM3409, may be a better way to go. \$\endgroup\$ Aug 21, 2018 at 23:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.