connect 10,000 LEDs in parallel

Discussion in 'General Electronics Chat' started by mousepad, Nov 3, 2008.

  1. mousepad

    Thread Starter New Member

    Nov 3, 2008
    8
    0
    Hi :)
    I want to toggle (on/off) 10,000 parallely connected LEDs with NE555 timer IC. each LED connected with a 150 ohm resistor

    each LED gets 20 mA current (That means about 200 Ampere)

    I'm expecting to give power from PC power supply (5V)


    * is this possible, :confused:
    * is there any other way to do this :confused:
    * should I buy a Transformer with High Ampere:confused:

    Please help me.

    Thanks.
     
  2. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    That has to be the most ambition LED project in terms of LED count I have every encountered.

    If you were to use Hi-efficiency RED LEDs you may be able to get the total current down to 50 Amps (5 milliamps per LED). That is still a hefty amount of current to manage.

    50 Amps equates to 1 Volt of drop for every 0.02 Ohms. Perhaps you could ease things a bit if you used ten 5 Amp power supplies for your design. That would give you 10 banks of LEDs with 1000 LEDs in each bank.

    hgmjr
     
  3. mousepad

    Thread Starter New Member

    Nov 3, 2008
    8
    0
    hi hgmjr

    Thanks for your Idea.:)
     
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You have a LOT of soldering ahead of you! :eek:

    Since you specified needing a 150 Ohm resistor to get 20mA current through an LED, I calculated backwards:
    E = IR = 20mA x 150 Ohms = 3 Volts across the resistor
    5V-3V= 2V across the LED
    to get your LED's Vf.
    With your plan as it is now, each LED will be using: P=EI = 2V x 20mA = 40mW, and each resistor will be consuming 3V x 20mA = 60mW as heat! How wasteful!

    If you started off using a higher supply voltage and created "strings" of LEDs in series with a single current limiter resistor in each string, you would waste far less power in the current limiting resistors.

    Here's an example power supply that would work for a good number of them:
    http://www.mpja.com/prodinfo.asp?number=17429+PS
    24V at 12.5A.
    You could then connect 11 LEDs in series with a 100 Ohm current limiting resistor.
    How I calculated that:
    MaximumLEDsInAString = INTEGER( (Vsupply-0.5v)/LEDForwardVoltage ) = 11
    Rlimit = (Vsupply - TotalLEDForwardVoltage)/Desired LED Current
    Rlimit = (24 - (11 x 2v))/20mA
    Rlimit = 2/0.02A = 100 Ohms
    One of those 24v 12.5A power supplies could power 625 of such strings, or 6,875 LEDs.
    You could get two supplies, and balance the load between them. That way you'd be using about 9.1A per power supply.

    If you are in doubt about how to connect them up, try here:
    http://led.linear1.org/led.wiz

    Instead of 24v for the power supply, put in 23.9v.
    Use 2v and 20mA for your LED parameters.
    Just put in 100 LEDs for the number in your array, or it will take a LONG time to create the schematic for you.
     
    Last edited: Nov 3, 2008
  5. mousepad

    Thread Starter New Member

    Nov 3, 2008
    8
    0
  6. scubasteve_911

    Senior Member

    Dec 27, 2007
    1,202
    1
    Isn't there a problem with stringing many LEDs in series? Of course, you get to choose one resistor to limit the current and use lighter gauge connections, but what about variations in Vf?

    Statistically, would they average out? Or, could the error in true Vf and specified Vf be accumulated? I would imagine a specification reflects the mean value of Vf, which would be statistically cancelled out; however, what if that isn't on a lot-by-lot basis? Perhaps an entire yield of LEDs could be up 10mV or so, which can add up quickly when stringing 10s of LEDs together.

    Steve
     
  7. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    Maybe by using the string approach in concert with a constant current source via an LM317 would improve on the overall uniformity of the brightness of the individual LEDs on a given string.

    hgmjr
     
  8. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Actually, when you're wiring that many LEDs in series, statistics are very much in your favor, as each string would have it's own random distribution that would approximate the traditional Gaussian or "bell" curve.

    It would be far worse if you tried using single LEDs with individual current limiting resistors within a few % of the typical Vf; as you would be guaranteed to have some LEDs with far higher current (those with a low Vf; would be much brighter) and some with a far lower current (higher Vf; much dimmer).

    There's always the possibility that you'll wind up with a few strings with an excessively low Vf or excessively high Vf. However, that is a VERY low possibility - particularly if the LEDs were jumbled in together and well-mixed beforehand.

    The other possibility is to measure all 10,000 LEDs, and sort them by their Vf @20mA, label, sort and then match them up. That would be a mind-numbing task of near-epic porportions. If it took you 30 seconds per LED to measure, label and record each one, it would take you 83.34 man-hours, or over two standard work weeks.

    Using semiconductor current limiters such as LM317L's would not be affordable. They have a built-in voltage dropout "overhead' of about 3v. So, you'd have a maximum of 10 LEDs per string - along with increasing the complexity of the board. You might do it with LM317L's, but even in quantity the cost would get prohibitive. A quick price check indicates you may be able to get them for around $0.40/ea in quantity from authorized distributors. Since 1000 would be needed for 1000 strings of 10 LEDs, That's $400 - and you'll still need a resistor to set the current for each LM317L. Since using the LM317L's would limit one to using 10 LEDs per string instead of 11 as with plain resistive limiters, you'd need about 91 more resistors.

    1,000 100 Ohm 1/4W 1% metal film resistors are under $20 from Digikey.
    http://search.digikey.com/scripts/DkSearch/dksus.dll?Detail?name=100XBK-ND
    Less than $0.02/each.
     
  9. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    $400 is definitely too high.

    I noticed that Digikey has LM317LZ parts in TO-92 packages for $103 for a quantity of 1000 units. That is still a bit stiff but a little easier to afford.

    The resistor approach is by far the cheaper approach and the compromises are not that severe.

    hgmjr
     
  10. mik3

    Senior Member

    Feb 4, 2008
    4,846
    63
    Mousepad why do you want to switch so many leds?
     
  11. scubasteve_911

    Senior Member

    Dec 27, 2007
    1,202
    1
    Maybe it's a PWMed LED tanning bed?!

    Steve
     
  12. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    I would recommend investing in protective eyewear without delay.

    hgmjr
     
  13. Wendy

    Moderator

    Mar 24, 2008
    20,764
    2,534
    I'm surprised no one has mentioned the 555 limitations. To me it sounds like a boat load of transistors as current regulators, along with a 555 to drive them. You don't have to draw a complete schematic, but a partial is definately in order. The shear quantity of LEDs is going to be expensive, the transistors will be relatively modest compared to that. If you have a chain of 10 or so LEDs (optomistic) then that is 1000 chains, with that many transistors and resistors.

    You could cut down on the transistor quantities a lot, but it will still be a large number.
     
  14. beenthere

    Retired Moderator

    Apr 20, 2004
    15,815
    282
    One 555, many FET's.
     
  15. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Actually, one 555 running on 12-15v should be more than adequate to drive the gates of a pair of IRFZ44's, one IRFZ44 per 24v power supply.

    IRFZ44's are spec'ed at Vdss=55V, Rds(on)=17.5mohm, Id=49A - actually, 5x what's required, since 5,000 LEDs in strings of 11 each will draw about 9.1A. But, the maximum overkill means minimal heat-sinking needed.

    This is assuming that our OP wants to turn all 10,000 LEDs on and off at one time. Since only a single 555 was mentioned, I have no reason to believe that they want to do anything other than flash the entire array on and off.

    Jameco has IRFZ44N's for $0.82/ea if you're buying them individually. You can get them for even less if you shop around.

    Current regulation will be provided by the 100 Ohm current limiting resistors, balanced against a reasonably accurately regulated 24v switching power supply that will be running at about 73% capacity.

    Each 11-LED string will have about 40mW consumed in the resistor, and 440mW consumed in the LEDs themselves; so about 91% of the power will be used emitting light. That's a lot better than what was started off with.
     
    Last edited: Nov 4, 2008
  16. Wendy

    Moderator

    Mar 24, 2008
    20,764
    2,534
    You know, if these are even half way modern LEDs this would be bright enought for a stadium light.
     
  17. hgmjr

    Moderator

    Jan 28, 2005
    9,030
    214
    Yes Bill,

    I am revising my suggestion up from standard laser protective eyewear to a welder's mask for all who might be tempted to look into the assembly.

    A quick back of the envelope calculation would indicate that 10,000 LED of the {T-1{\large \frac{1}{4}}} type would occupy an area approximately 2ft square.

    hgmjr
     
  18. Bernard

    AAC Fanatic!

    Aug 7, 2008
    4,170
    395
    My bet is on a 20" X 20" O'scope screen?
     
  19. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Now you're talking GENUINE nightmare to wire up! :eek:

    Of course, one could drive the x & y axis using just 20 LM3914 IC's...
     
    Last edited: Nov 4, 2008
  20. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    It's going to go on a rocketship. He's planning on lighting up the dark side of the moon. :)
     
Loading...