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.

Lm3464: Lm3646

Part Number: LM3464
Other Parts Discussed in Thread: LM3553, , LM3643, LM36010

Support Path: /Product/Help with Device Selection/

I am currently using lm3553 for two IR LEDs flash light.

We are looking to go to 4 LEDs with I2C configuration but there is not many option. If we use the 2 LED driver version for example LM3464 1.5A max, how can we synchronize between two devices?

Maybe we have to use Strobe pin but there is still a delay difference from rising edge of strobe to LED firing across the chip.

Any suggestion?



  • I was meant to say LM3646.
  • Hello,

    To help me provide the best solution, I would need a few answers to some application questions.

    1. What is the max current per string that you need to drive?  Is it 1.5A per channel or 1.5A total?
    2. Are you looking for a high current pulse or is a DC current needed?
    3. What is the forward voltage of the LEDs you are using?

    Depending upon the answers a device like the LM3643 or LM3644TT might be a better solution. 


  • Currently we drive 1.5A total thru IR illuminator in series.

    Hight pulse is needed but programmable to adjust intensity. The I_forward is only 500mA DC, 1A/100uS at 4V Vf max, 3.3V typical.

    My concern is able to synchronise them, turning all 4 IR light on and off at the same time.

    We are looking at OSRAM OSLUX IR illumintator.

  • The LM3643/4 has an IR mode that can greatly reduce the lag between the strobe signal and the LED turn on. the following application note highlights the timing constraints associated with this mode. If you were to use the LM3643 and LM3643A (same device with different I2C addresses) you would then have independent control of each output and by using the strobe pins tied together the sync should be pretty good (1-2us if I had to guess due to deglitch filters on chip).

    Please give this document and part a look and let me know if you have questions. Thanks.

  • Also what is the max V_forward it can handle if we stack 2 or 3 IR with 1.8V Vf? I read some chip has I2C command to turn on and off, but we do have flash enable signal currently.

    With the LM3553, we stack 3 lower Vf IRs.

  • You should be able to stack two in series if VF = 1.8V. 3 LEDs in series would be too close to the over-voltage protection of the device. 4V is typically the max LED voltage we recommend.
  • Can you use this or similar 1 LED driver to turn on and off a UV light?

    This way we can control intensity as well. The UV light is turn on so you can read a passport ID or allow the builtin mobile phone like camera to take a picture.

    UV light example:



  • I'm also using them on the same board, so I2C address conflict is issue since I don't want to use mux.

    2 + 2 IR LED
    1 UV LED

  • The LM3643 device has two different I2C addresses available so there is not a need for a mux. IF the VF's are 3V or higher you could put one LED per channel and use the two different address versions
    If the IR LEDs are 1.8V, you could use two in series on each output channel.
    For the single UV LED, you could use the LM36010 which is a very small single channel device that has adifferent I2C address.

  • The LM36010 is new so only stock at Ti which I assumed we can purchase.



  • The LM36010 is new so only stock at Ti which I assume we can purchase.



  • You are correct in that it might only be available through at the time being due to being newly released.

  • I am doing the schematic based on LM3643 evm.

    I am not sure why there is a 49.9K in parallel with NTC 100K  on TORCH/TEMP?

    We are planning to use NTC on nearby location to monitor the temperature of LED.

    Also it looks like they double up on the 10UF capacitors on both input and output instead of one 10uF per datasheet.

    Since we have a tiny board, we might use one.


  • I figure the math for the parallel 49.9K due to the 2.3V NTC open detect threshold.
    Let you know if you think we need to double up on the capacitor. it is running off an Li battery similar to mobile phone.

    "Additionally, the NTC input looks for an open NTC connection and a shorted NTC connection. If the NTC input falls below 100 mV, the NTC short flag is set, and the device is disabled. If the NTC input rises above 2.3 V, the NTC Open flag is set, and the device is disabled. These fault detections can be individually disabled/enabled via the NTC Open Fault Enable bit and the NTC Short Fault Enable bit."

  • The parallel resistor is to help linearize the NTC temperature coefficient.  In picking the resistor and thermistor, you need to size the two to get the correct temperature trip point for your system.  The NTC faults are more for production testing purposes to make sure a connection is made.  Once verified in the test environment, you can disable the faults as they do not help much in the real application.

    With regards to the caps, the key is to keep the effective capacitance as high as you can in the area allowed (must account for DC bias effect of ceramic caps).  I would recommend that at least 4uF of capacitance to keep the led current ripple to a minimum and maintain stability of the boost at high currents.