Triac-Based LED Dimmer by National Semiconductor

Discussion in 'The Projects Forum' started by Tom Kay, Mar 5, 2009.

  1. Tom Kay

    Thread Starter Active Member

    Feb 10, 2009

    I'd like to discuss the new chip and circuit by National Semiconductor. It claims to dim LEDs very smoothly with any basic Triac, using line voltage. This appears to be done with the addition of their circuit, which uses the LM3445 chip.

    I will probably buy a sample of the board soon, to find out how many LEDs I can power with the circuit. As well, I will probably need some help in altering the resistors and capacitors on their board to get the maximum number of LEDs connected as possible. They claim it's good up to an amp of output, and around 70 volts, or more. However, both figures are a bit uncertain, as several different websites show several different values for the outputs. So it's a bit nebulus to say the least.

    Please have a look at the attachments and see what you think. The theory is well beyond me. I just want to make it work.

    So, to narrow it down to a question or two, what changes would be needed to drive, let's say, 6+ LEd bulbs. The bulbs would be custom made with something like 2 parallel sets of 20 LEDs in series. Each LED would be a 5 mm, 3.4 volt, 30mA, warm white, as an early guess. This concept needs lots of fine tuning, no doubt.

    They do have a couple video clips on their website, including a demo of the board.

    And, does the method seem sound? Any obvious concerns or weaknesses about their approach to LED dimming?

    Thanks, Tom.
    Last edited: Mar 5, 2009
  2. thatoneguy

    AAC Fanatic!

    Feb 19, 2009
    It looks good.

    In simple terms, a Triac light dimmer is a PWM dimmer, but with an AC voltage rather than DC. The IC "measures" the on/off time at 60Hz created by a triac dimmer. The driver then genereates and outputs it's own PWM at 5.85kHz to change the apparent brightness of LEDs by the same amount an incandescent light would dim, but without causing the LEDs to flicker.
  3. Tom Kay

    Thread Starter Active Member

    Feb 10, 2009
    OK, so we've got one yes vote for the circuit. Good to hear.

    I have also read that the frequency is variable (there's no pot in the circuit) up to around a few hundred thousand hertz. I would assume that some item(s) would have to be swapped out to achieve this.

    No matter, really, because all you have to do is fool your human eye with 1khz or a bit better, so cancel out the flicker effect. And, I suppose, keep the frequency substantially different from IR remotes lying around the home theater.

    Any thoughts on maximizing the output on this beast?

    Thanks oneguy, Tom.
  4. Spudgunster

    New Member

    Jul 7, 2009
    BTW- your pulse width modulator may put out the right electrical frequency, but what matters is the wavelength of the light the LED emits. Additionally, if your drove an IR LED with this circuit, you would then have to precisely tune it to cause interference in the right frequency range, unless it was bright enough to just completely overwhelm the IR receiver.