This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM3445 LED driver question regarding low end dimming

Other Parts Discussed in Thread: LM3445

This is a repost of sorts and I am interested if anyone experiences this or has a solution.

I am using the circuit topology very similar to the application data sheet.

At the low end of the dimming, the LM3445 chip at some point switches modes from PWM control of the power switch FET to turning the switch on full. 

This is described in the datasheet as

"The transition from dimming with the DIM decoder to headroom or minimum on-time dimming is seamless. LED currents from full load to as low as 0.5 mA can be easily achieved."

-Page 14

"www.ti.com.cn/.../lm3445.pdf"

The problem is that there is a noticeable pop-on phenomenon where the LED will have a brighter level at a lower dimmer level. In bringing up the level from zero, the light output has a high spot and and then steps down with a further slight increase in dim level. From there the LED has a smooth dim curve to full. I believe this happens where the DIM decode takes over from simply leaving the FET switch on full.

Is there a way to control this point a which control is handed over to the decoding of the driver chip?

Has anyone seen or solved this?

Thanks

search keywords

LED, pop-on, dim curve, LM3445, dimming, dim

  • Hello,

    I assume this is using a triac dimmer. What I believe is happening is a function of the triac dimmer. One way to test this is to add a resistive load or an incandescent bulb in parallel with the LM3445 circuit and see if this goes away. There are EMI components in the dimmer. One of them is a capacitor. If the load is very light the actual perceived duty cycle of the LM3445 will be higher until the circuit adds enough load to use this capacitor energy.

    Thanks,
  • Irwin,

    Thanks for replying.

    So I am not correct in assuming that at some point in the dimming curve the lm3445 changes in switching mode from PWM to fully on, when very low dim levels are present?

    It seems from scope traces that it indeed has this behavior but that could be the dim detect/angle detect erroneously sensing a full on condition and then attempting to bring the current up to the limit set by the ISNS pin resistor.

    Thanks 

  • Hello,

    The method the LM3445 regulates current will leave the gate high if there is no voltage present or if it is lower than the LED stack (assuming a buck converter) so if the triac dimmer is off during a portion of the line cycle the gate will be high.  It will remain high until the current can rise to the Isns threshold but this will only be possible when input voltage is present and higher than the LED stack voltage.

    So it will look like it is fully on (gate high) but that is because current cannot ramp up in the inductor and trip the LM3445 Isns threshold.  Also no current will go to the LED string if the gate is high.

    For a buck the input voltage has to be higher than the LED stack, for a flyback it does not.

    Thanks,

  • Irwin,

    I resolved the issue it seems.

    I had a 0.47uF capacitor on the FLTR2 pin that was probably not discharging enough to bring the voltage down in the negative input to the PWM comparator block. A 0.15uF seems to clear up the problem.

    I used the higher capacitance to slightly slow the change in dim level to replicate an incandescent bulb behavior. I will add this slight lag somewhere else in the circuit.

    But I am correct in assuming that the capacitance on the FLTR2 is causing this problem as this and the mosfet connected to the internal 370K resistor converts the PWM 'DIM' signal to a voltage.

    Thanks

    Wade 

  • Hi Wade,

    The FLTR1 pin is compared with a ramp waveform from 1 to 3 volts which is turned into a 5.9 KHz PWM. FLTR2 is pulled up via the 370 Kohm along with FLTR2 capacitor make the DC reference that the controller uses to regulate peak current.

    If it is discharged it shouldn't matter what the value of FLTR2 capacitor is. I would watch FLTR1, DIM and FLTR2 as it powers up to see what is going on. If you have capacitance across the LEDs it's possible that the smaller FLTR2 capacitor causes the flash of current but the LED capacitor is large enough to absorb it when FLTR2 capacitor is 0.15 uF and maybe not large enough when FLTR2 capacitor is 0.47 uF.

    It's easiest to tell by what the signals look like. Also note that a cold start board can look completely different than one power cycled. Capacitors across the LEDs can hold their voltage for a long time.
  • Irwin,

    I was premature in thinking I had resolved this issue.

    I will capture the waveforms and post.

    Thanks

    Wade