Connect with us

Forward voltage binned LEDs?

Discussion in 'LEDs and Optoelectronics' started by eem2am, May 16, 2013.

Scroll to continue with content
  1. eem2am

    eem2am

    414
    0
    Aug 3, 2009
    The below (example LEDs '1' and '2') demonstrates that Forward voltage binned LEDs are always around twice as much in cost as otherwise equivalent non-forward-voltage binned LEDs.
    Given this price difference, why do people use forward voltage binned LEDs?
    I mean, LEDs are always put in series (most usually anyway) so forward voltage binning shouldnt matter?

    1.....Cree, White, XPEHEW-L1-000-00BE7 = $1.54 (@ 5000 pces on digikey)

    XPEHEW family datasheet:-
    http://www.cree.com/~/media/Files/C...odules/XLamp/Data and Binning/XLampXPEHEW.pdf



    A cheaper alternative would be:-
    2......Cree, White, MX6AWT-A1-0000-000BE7 = $0.77 (@ 5000 pces on digikey)

    MX6AWT datasheet:-
    http://www.cree.com/~/media/Files/C...d Modules/XLamp/Data and Binning/XLampMX6.pdf
     
  2. Harald Kapp

    Harald Kapp Moderator Moderator

    10,582
    2,360
    Nov 17, 2011
    I have never heard of forward voltage binned LEDs. The datasheets you linked do not reference such a binnig.

    Typically LEDs are binned for brightness and/or color. This is to allow manufacturers of e.g. LED marquees to have a consistent brightness across the LED matrix or any similar purpose.
     
  3. eem2am

    eem2am

    414
    0
    Aug 3, 2009
  4. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,418
    2,788
    Jan 21, 2010
    You can bin things based on anything.

    In this case the bins are fairly wide. The bin will allow you to design something where the max and min Vf are pretty well characterised (certainly better than the datasheet for a non-binned part would be).

    This will allow you to (say) put a string of these LEDs together (let's say 10 H binned LEDs) and know the Vf will be between 32.5V and 35V. Without the binning, you may have to deal with (say) 30V to 37.5V.

    I'm not sure that I'd put two LEDs in parallel if one had a Vf of 3.25V and the other had 3.5V. I would certainly expect one to hog a significant portion of the available current.

    There's lots of discussion about LEDs in parallel on the internet. My LED tutorial has a number of references to manufacturer data to support it. Here is another discussion. Of interest is a post near the end which refers to problems of reverse voltages occurring when LEDs are used in long strings with PWM (practically I'm not sure that's a problem unless the input is effectively grounded during off times, but I am willing to be convinced otherwise)..
     
  5. eem2am

    eem2am

    414
    0
    Aug 3, 2009
    Thanks , that post you kindly referenced, from "grizedale" is the one that always comes up when you google "parallel leds".
    Ive read it many times, as well as the thread by sunnyskyguy, who seems to claim that leds can be paralleled.


    The LED manufacturers are amazingly silent over paralleling leds.
    That "reverse biased led" failure mode can also happen if you have series strings of leds in parallel, which one almost always does, so its not very relevant?

    Here is a parallel LED product for sale, and its on cheap FR4 which is unusual for parallel led products, where the thermal coupling has to be excellent, but they wont answer questions on how they get the led currents to share equally

    http://www.forge-europa.co.uk/uploads/FE-LCXXXXXXXX.pdf

    Back to the post you kindly referenced, ( i know its not you speaking in it) i didnt see the basis for making the series resistor equal to the ESR.
    Also, the LEDs ESR is not in the datasheet, and may vary, and certainly does vary as the current is varied in the led.

    LEDs in parallel offers huge simplifications in driver circuitry, since simple hysteretic buck converters can be used.......so why are led manufacturers not jumping on the "parallel led bandwagon"?
     
  6. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,418
    2,788
    Jan 21, 2010
    Here is one reference I pointed you to earlier.


    Because if any string were to maintain a large charge, it would act to forward bias the other LEDs. This would be protective.

    If we assume that all the LEDs in one string were "discharged" then this still imposes a combined Vf that is equivalent to the voltage across an equivalent number of LED's junction capacitances.

    If you had a string of (say) 10 LEDs in parallel with a string of only 1 LED, then it is possible that the Vf would be sufficient to provide a significant reverse current through the other string. However, no damage is going to occur to any LED until its junction capacitance has been fully discharged. This means all the other LEDs will have their junction capacitances partially discharged, thus reducing the available voltage.

    Let's keep throwing figures around. Let's assume that the capacitance of one LED is only 75% of the capacitance of the others. And let's assume the LEDs operate at 3V.

    OK, so there's 30 volts across the array, and 3 across the single LED. That's a differential of 27 volts. Now, let's make some bold assumptions that the voltage across the junction capacitance of the faulty LED falls to zero volts and the voltage across the other LEDs falls by 75% of this. So each of those 9 LEDs has 0.25 x 3V (0.75V) across them (that's 6.75V). Now, there's also a drop of 3V across the other LED in the short string, so that leaves 3.75V to reverse bias the low capacitance LED.

    It's probably not unreasonable that this could happen if the LEDs were of very different types, but it still requires that one string have a far lower Vf than another.

    Is there another mechanism which could cause this?

    It is entirely possible that they don't. But assuming they do, they may do it by specifically sourcing LEDs from the same manufacturing batch (the difference between such devices will be very small), and that the substrate they're on may be sufficient to keep them in good thermal contact. In short, it may be a discrete form of COB.

    The intuitive reason is that it causes the power dissipation in the LED to rise slower than the temperature, so that whilst self heating will cause an increase in current, which causes more heating, more current, ... the rate at which this happens slows so that an equilibrium is eventually reached (hopefully before the LED melts)

    Yes, it is the slope of the V/I curve at a particular point.

    Because it's in their interests that they don't sully their own name by recommending practices which will cause customers to lose faith in the product.
     
  7. eem2am

    eem2am

    414
    0
    Aug 3, 2009
    agreed, however, very few people in the world can design LED drivers, unless its a hysteretic buck type of led driver.....since its buck, if they want to drive many leds theyve got to parallel them........so if the led manufacturers want to get leds out in the market, they have to explain to customers how to sucessfully parallel leds (eg thermal coupling , Vf binning etc etc)

    There are companies out there who are stalling on led products because they dont have the staff who can design buckboosts/boosts/sepics/flybacks etc.

    So why arent led manufacturers putting app notes on their websites explaining how to do parallel leds sucessfully?
     
  8. (*steve*)

    (*steve*) ¡sǝpodᴉʇuɐ ǝɥʇ ɹɐǝɥd Moderator

    25,418
    2,788
    Jan 21, 2010
    It's simple. Make many strings of LEDs, each with a resistor and place them in parallel.

    It's so simple everyone is doing it.

    edit: or use COB techniques. But that's essentially a manufacturing step.
     
  9. eem2am

    eem2am

    414
    0
    Aug 3, 2009
    The resistors will dissipate too much...so people are doing it like this....with no resistors
    http://www.forge-europa.co.uk/uploads/FE-LCXXXXXXXX.pdf

    The resistor value would also depend on the degree of thermal coupling between the leds.....the sizing of the resistor is only simple if you make it way too dissipative.
     
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

-