Connect with us

Driver for luxeon leds

Discussion in 'Electronic Basics' started by booth multiplier, Dec 1, 2004.

Scroll to continue with content
  1. Dear All,
    I'd like to built a driver circuitry for 8 Luxeon leds in series.
    Each of them has a forward voltage at 3.5 V at 350 mA. 8 in series
    makes 28V. I have applied the 28 volt works fine, but if I want to
    make them flash it's not working, they simply don't flash. The ON
    OFF sequence is like this: 1 sec OFF, 0.4sec ON, 0.1 sec OFF,0.4sec
    ON, 0.1 sec OFF AND SO ON. Do I need a higher voltage for flashing?
  2. Your thinking voltage when its better to think current. Vf of LEDs can be
    fairly variable, wireing then is series is fine, but you want to think of
    350mA roughly at whatever the Vf settles at.

    1W LS will take up to 1A at low duty cycle, look up the data sheet at
    Lumileds , there is an application sheet for using them as a flash for
    mobile phone cameras.

    400mS is possibly a bit long tto go as high as an amp for longetivity but
    genuine Stars will take a suprising amount of abuse.

    Look to use a constant current driver not voltage though.

  3. I agree with Adam that you should use a current source or at least a
    current regulated power supply instead of a voltage source. However,
    that does not explain why your current system will not flash.

    Does it not turn off or does it not turn on? Also, how are you
    flashing the LEDs? Do you drop the voltage to 0 when they are supposed
    to be off? Can you describe the circuit you are using?

    Vic Roberts
    To reply via e-mail:
    replace xxx with vdr in the Reply to: address
    or use e-mail address listed at the Web site.
  4. You treated the LEDs like resistors which they are not. According to the
    manufacturers information you always has to use a series resistor to limit
    the current through the LEDs. Look at:
    for instance.
    As LEDs are current driven devices you'd better look for a current source.
    See simple schematic below. The 350mA you mentioned is the maximum current
    the LED can handle without being damaged. With a lower current the light
    production will slightly decrease but the life expectance of the LED will be
    much better. Another important thing is temperature and heat production. Too
    high a temperature will also decrease life expectance of the LED or may even
    damage it. The example below has only two LEDs but you can add more if you
    increase the supply voltage accordingly. T2 and R3 make up the on/off
    switch. A 5V on R3 removes the base current from T1 so its collectorcurrent
    vanishes and the LEDs go off.

    +------+ 9V
    | |
    .-. V
    | |R2 - LED
    | | |
    '-'1k5 |
    | V
    | - LED
    | |
    | |
    | |
    | |
    +-----------+ |/ T1
    | +----| BC635
    | | |>
    | Si V |
    R3 | - |
    ___ |/ T2 | .-.
    -|___|--| BC546 | | |R1
    4k7 |> Si V | |2.2 Ohm
    | - '-'
    | | -
    ---------+-----------+------+ GND
    created by Andy´s ASCII-Circuit v1.24.140803 Beta

    petrus bitbyter
  5. First off, he said 8 LSes in series, yours shows only 2 and 9V supply.
    For more than 28V at more than 1/3 amp, I would say that the BC635 will
    quickly butn up. You need a TO-220 type transistor like the TIP31 on a
    heatsink. The two diodes should be 1N4002 or better. Remember these
    Luxeon Stars are $15 or so apiece! If you give him the wrong advice and
    they burn up, that's well over a hundred dollars!!
  6. You're really quick with your comment but did you read what I wrote? I gave
    an example, not a full fledged design. The components in the example are
    used within the specs. If you really think I'am wrong, prove it.
    Nevertheless, if you want to stay on the safe side a power transistor may
    have its use, but a heatsink for less then 0.5W seems a little bit
    exaggerated to me. Both power transistor and heatsink are only required if
    you want to increase the power voltage wihout adding LEDs in the collector
    chain. I also wonder what special properties requires the 1N4002 or better.
    Why should a 1N4148 (for instance) not be good enough?

    petrus bitbyter
  7. Rich Grise

    Rich Grise Guest

    It's because Watson's afraid of anything over 24V. ;=)

    Actually, I'm pretty much afraid to use LEDs that cost fifteen bucks! =:-O

  8. This circuit doesn't have very good current stability to start with.
    The LED current depends on the forward voltage drop across those two
    diodes; this Vf changes with temperature. Using heftier diodes with
    greater dissipation will give some improvement in stability.
    Nope. I've been installing a PBX all day today, Neg 48V (more
    like -56VDC) all over the place. Ten gauge wiring, etc. Each phone
    line has enough current, maybe 150mA, to make a nice blue spark when you
    short it. Scares the hell out of some of the contractor's techs. I get
    a kick out of seeing them go "WHOA! What was that?" when that happens!

    I was using four 1N4002s and maybe a 1N5818 schottky in series for a
    dummy while I was experimenting with my LSes. Better safe than sorry at
    that price.
  9. This circuit does not have a very good stability. But IMHO it's much better
    then connecting the LEDS in series to a battery (or other power supply)
    without any current limiting at all. If you want to improve it's stability a
    little bit you'd better first increase the current through the diodes by
    lowering R2 then use beefier diodes. But why make things difficult? I'm sure
    you can calculate and measure the variations in the LED current due to
    temperature changes but I doubt if you will *see* that as variations in the
    light with the naked eye. If you nevertheless want more control you'll have
    to make a different design. Maybe you have a proposal?

    petrus bitbyter
  10. [snip]
    Yeah, the following commonly seen circuit, posted here many times by
    myself and others, is used more often than the above because it's more
    stable, but still not very good. As with your circuit, it has to be
    scaled up in current and power to meet the 350 mA needed for the Luxeon
    Star. Use a BD135 or TIP31 for Q1, on a heatsink if it gets too hot to
    touch. And the 10k to 47k needs to be lower also. With the 33 ohms
    changed to 2 ohms, it should give about 300 mA.

    You can also use the LM317 as a current limiter, see the circuit in the

    +--------------------+------- Positive
    | | Supply V.
    | |
    | --- LED
    10k \ \ / =====>
    to 47k / ===
    ohms \ |
    / |
    | / Q1
    | | / Gen'l
    | | Purp NPN
    +----------------| 2N3904 or
    | | 2N2222A
    | | \ E
    Q2 \ \
    Gen'l \ | |
    Purp NPN | 470 ohms |
    2N3904 or |------/\/\/\----+
    2N2222A | |
    E / | |
    / \
    | 33 ohms /
    | for \
    | 18ma /
    | |
    | |
    | |
    Negative Supply V.
  11. I built a mains power supply for a Luxeon star as follows:-

    Live / hot --------------
    O 40 watt light bulb.
    V luxeon A 1N4007 diode
    Neutral / return -----

    It lights, but doesn't flash like Luxeons are meant to. At least it's
    not flashing slowly. It appears to be flashing very quickly at a speed
    that only registers with peripheral vision.

    Have Luxeon had a bad batch of their LEDs or do I need to use a
    different light bulb? A friend says that Luxeons don't flash, but I've
    seen them do it in cheap disco lights so he's just talking complete
  12. I built a current regulator based on an LM2941 that drives one Luxeon
    with about 270MA from 4.2 to over 18V (as high as my bench supply
    goes). It's got an on/off pin that you can use for logic-level

    [I'm building a solar powered beacon, with a PIC for control,, but
    don't hold your breath. 8*]
  13. Are you talking about these? I believe they are meant to flash at a high

    Now, your bulb thingy will work, but the initial current will be
    painful. Light bulbs start out with a very low resistance, which quickly
    increases as the bulb heats up. See

    Thus, your luxeon is being subjected to hideous currents when you flip
    on the light. I just measured the resistance through an 'off' 100W light
    bulb (USA). The resistance was 10.3 ohms. Now, if you put 120V across
    10.3 ohms, the current will be

    120/10.3 = 11.65A

    Once it heats up, the resistance will go up to 120^2/100 = 144 ohms.
    That is more than a 10x increase in resistance.

    You might be able to snub the initial rush of current using a soft-start
    device, as outlined in Don Klipstein's tutorial given above. However, I
    guess it would be easier and safer just to use a 5V wall wart.

    Robert Monsen

    "Your Highness, I have no need of this hypothesis."
    - Pierre Laplace (1749-1827), to Napoleon,
    on why his works on celestial mechanics make no mention of God.
  14. Why should it flash slowly? Using your circuit the LED will be on for
    one half cycle and off for the other. So, in Europe it will flash 50
    times a second, in NA 60 times a second.
    Any LED will flash if used on a "flashing" power supply.

    Vic Roberts
    To reply via e-mail:
    replace xxx with vdr in the Reply to: address
    or use e-mail address listed at the Web site.
  15. ask Linear Technology for Application Notes and Driver circuits for
    Luxeon LEDs, Maybe LT1618 and LTC3441 do the job for portable Battery
    systems with torch and Flash functionality.
  16. Charles

    Charles Guest

    I recently designed a driver for Luxeon III, about 3400 of which are
    now being installed on the pyramid shaped Luxor in Las Vegas. I used
    the National LM2592-ADJ, with a current feedback for 850 ma Each
    driver is for 2 in series, but it can work with 8. Across the 2 in
    series is a 220 uf cap. The feedback resistors are not bypassed.
    There is also a 5K resistor across each LED pair to make them turn off
    quickly. If this is omitted, there is a visible fading "afterglow"
    effect as the cap discharges slowly.

    The circuit turns on/off in a few millisec. Supply voltage should be
    more than 20% over the load voltage (this app uses 42 v anf the "HV"
    version of regulator)

Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day