Using PWM to drive LED - What Do I Gain By Using a Regulator vs a MOSFET?

Discussion in 'The Projects Forum' started by Vyperion, Feb 3, 2013.

  1. Vyperion

    Thread Starter New Member

    Feb 3, 2013
    I'm currently working on a battery-powered project, in which I'm using a PWM signal to control an LED. For the first prototype I used the PWM signal to control a MOSFET, but also calculated and ordered parts to make a circuit based around the TI LM3405 Constant Current Buck Regulator.

    My question is whether it's worthwhile to use the LM3405 driver for the LED in this system, where the LED is the only load. Since I'm already driving both with a PWM signal, I should be getting a much better efficiency, so am I going overkill using this dedicated driver IC?

  2. Sensacell


    Jun 19, 2012
    The answer depends of what the power supply voltage is and the LED forward voltage.

    If you are driving the circuit from a higher voltage, (12 volts, for example) using a linear (resistor) circuit to limit the LED current becomes inefficient.

    Using a switching regulator will give you higher efficiency at the expense of circuit complexity and cost.

    If the supply voltage is close to the LED forward voltage, you might be better off using a resistor to limit the current.

    Post your schematic please.
  3. crutschow


    Mar 14, 2008
    Controlling an LED with PWM doesn't improve efficiency unless you also have an inductor (instead of a resistor) in series with the LED to control the current, in which case you then have a rudimentary switching power supply.
  4. Vyperion

    Thread Starter New Member

    Feb 3, 2013
    I'd like to thank you both for helping to lead me to more accurate information. Up till this point I was under the impression that PWM was a more efficient route for dimming, partially due to how often it's discussed as the solution to the LED dimming problem (I've seen it overwhelmingly more often than current dimming, which may be because I'm mainly an embedded systems guy or it may just be coincidence), but also not fully considering the full system power. I know LEDs like constant current much more, but never really came across much discussing 'constant current dimming' in particular.

    For what its worth, here's a couple sources I've found with a very thorough discussion of the efficiency of PWM and Constant Current dimming. The second one in particular has clear measured results for PWM vs Constant Current dimming, making it quite plain the benefit of the latter.

    The circuit is running at 5V for the moment, while it's on the bench. Due to all this new information, changes are a-comin', so I'll probably bump that to 6V (or 9V, but that's a bit more of a jump). This schematic is for the MOSFET circuit; MultiSim doesn't have the LM3405, but basically for that one I just fed the comparator (PWM) output signal into the EN/DIM pin of the chip. The CON voltage is a simple voltage divider so I could use the parts I have on hand. At max Duty Cycle (~90%), the LED sees a bit over 80mA.
    Last edited: Feb 3, 2013
  5. crutschow


    Mar 14, 2008
    You circuit has no current limiting so the LED peak current will be whatever the MOSFET can supply. You need a resistor in series with the LED to limit the current.
  6. Audioguru


    Dec 20, 2007
    A Mosfet should NEVER be used as a follower as you show it because for it to turn on well its gate must be maybe 5V higher than the +5V power supply.
    If the gate is +5V then its source might be only +2V then the LED will be VERY dim.

    Instead use the Mosfet as a switch with the LED and its series current-limiting resistor from the drain to the positive power supply voltage.
  7. ErnieM

    AAC Fanatic!

    Apr 24, 2011
    OK so you have a battery so pick one that's close to the LED voltage so you don't convert anything and get good efficiency for free.

    What is the PWM control for?
  8. JMac3108

    Active Member

    Aug 16, 2010
    Using your terminology, PWM dimming vs Current dimming of LEDs ...

    Be careful if you're using white LEDs. They will change color (usually they get blueish) if dimmed by reducing the current. Its best to PWM them if you want to keep the true white color.

    But, this doesn't preclude using a switching power supply to drive a constant current through the LEDs. When I designed products with white LED backlights for the LCD, I used a constant-current switching regulator and PWM'ed the ENABLE pin on the switching regulator. This allowed me to get an efficient constant current source and dim the LEDs without affecting the color. But be aware that the PWM frequency has to be low enough to account for the startup time of the switching power supply.
  9. Vyperion

    Thread Starter New Member

    Feb 3, 2013
    @crutschow: Err, umm...*blush* heh, knew I forgot something... >.<

    @AudioGuru: Thanks for the tip - I'll make the change.

    @ErnieM: Right, I'm trying to use as low a voltage as is convenient - 5V is needed for the chips, which isn't a convenient amount for batteries (and doesn't give much room once it starts dying), so a 6V pack was the plan. As is, the circuit actually behaves pretty much the same from 6V to 5V since the voltages scale.

    @JMac: That's actually more or less what I'm doing with the LM3405. I'm not terribly concerned about the LED color in this application - maybe further on down the road I will be, but for now I'm placing a greater emphasis on simplicity and efficiency.