Current Mirror For Parallel LEDs

Discussion in 'Digital Circuit Design' started by Gamer91, Sep 23, 2016.

  1. Gamer91

    Thread Starter New Member

    Sep 23, 2016
    10
    1
    All,

    I'm new to hobbyist level electrical engineering, and need some help.

    I'm designing a circuit to power high power leds, instead of individual drivers and I want to power several in series/parallel.

    I am using a design I found on the net, it's a current mirror, but also has protection for open circuits so that one string doesn't get
    all the current if the other string opens.

    The way it's supposed to work. Pnp transistor Q1 is supposed to protect the LEDs on the right side, if an LED were to open on the left, because D1 will stop the flow of base current from pnp transistor Q1. Npn transistors Q2 and Q3 protect the LEDs on the left side, if an LED were to open on the right side, because there will be no current flow through the bases of npn transistors Q2 and Q3.

    The issue I'm having is when I set this up in LTSpice, and open one side, all 750ma flows through the opposite side, not working like it's supposed to is this a problem in spice or did I do something wrong?

    [​IMG]

    Thank you!
     
  2. Gamer91

    Thread Starter New Member

    Sep 23, 2016
    10
    1
    Correct photo, other one was an older one with an incorrect diode, sorry.

    [​IMG]
     
  3. #12

    Expert

    Nov 30, 2010
    16,321
    6,818
    There is something wrong with that circuit. I can't figure out why it's supposed to work.
     
    Gamer91 likes this.
  4. GopherT

    AAC Fanatic!

    Nov 23, 2012
    6,061
    3,823
    When you build this, where do you plan to get a constant current source to power it? Most people use batteries or a voltage regulated supply.

    With a constant 0.75a current supply, where do you expect the "other" 0.375 amps to go when you kill one string? Of course the voltage supplied to Q1 will get jacked up until 0.75 amps are forced through it no matter what happens.
     
    Gamer91 and #12 like this.
  5. #12

    Expert

    Nov 30, 2010
    16,321
    6,818
    I agree. Even if I could see how that current mirror works, that current source, I1 completely screws the pooch.
     
    Gamer91 and GopherT like this.
  6. crutschow

    Expert

    Mar 14, 2008
    13,028
    3,237
    As noted, the ideal current source in Spice will output 0.75A no matter what, unlike a real current source.
    You need to use a current source with a reasonable voltage limit, which would be no greater than the supply voltage for a typical current source.
    This can be simulated with a diode to a voltage source from the current source output.
    With that, the constant current voltage can go no higher than 1 diode drop above the voltage source.
     
    Gamer91 likes this.
  7. DickCappels

    Moderator

    Aug 21, 2008
    2,658
    632
    The transistor is not connected as a current mirror, it is connected as an always-on switch.

    If you put D2 from the emitter to the base with the anode connected to the emitter, it would be sort of a mirror.

    Using a resistor in series with each string of LEDs is much simpler, would work very well and would not be likely to kill the LEDs as much as this discreet current source.

    If you really want to drive the LEDs with a constant current, try using LM317 regulators.

    [​IMG]
    I found this circuit and the description of it at http://www.bristolwatch.com/ccs/LM317.htm
     
    Gamer91 likes this.
  8. Gamer91

    Thread Starter New Member

    Sep 23, 2016
    10
    1
    Thanks guys!

    Current source was causing the maximum voltage to go over the threshold of collector to emitter limit.

    It can balance the currents because of the equal voltages developed on 1.5Ω emitter resistors R1 and R2.
     
    GopherT likes this.
  9. Gamer91

    Thread Starter New Member

    Sep 23, 2016
    10
    1
    I'll be using a constant current uC to supply the LEDs, but haven't gotten that far, ha ha.

    Will be a string of 50 leds.
     
  10. Alec_t

    AAC Fanatic!

    Sep 17, 2013
    5,801
    1,105
    Here's a simple (albeit not very accurate) way of providing each LED string with its own 'constant' current source, so that failure of one string doesn't compromise the other.
    ParallelStrings.JPG
     
    Gamer91 likes this.
  11. atferrari

    AAC Fanatic!

    Jan 6, 2004
    2,648
    764
    All the fifty in series?
     
    Gamer91 likes this.
  12. #12

    Expert

    Nov 30, 2010
    16,321
    6,818
    Yeah. I can imagine this one going bad, real fast.:(
     
    Gamer91 and atferrari like this.
  13. Gamer91

    Thread Starter New Member

    Sep 23, 2016
    10
    1
    Don't have enough voltage for that, ha ha. Making a light bar for my truck. 50 or 40 leds, in a combination of series and parallel.
     
  14. #12

    Expert

    Nov 30, 2010
    16,321
    6,818
    Gamer91 likes this.
  15. Gamer91

    Thread Starter New Member

    Sep 23, 2016
    10
    1
    Thanks for the input guys, always helps to have different voices and opinions to show things I wouldn't have thought of.

    But I think I've found something better, using an LT3755 because it has an option to set a over voltage limit and uses a mosfet to shut it off if it passes the limit!

    I also plan to use PWM to make the bar able to dim as well.
     
  16. #12

    Expert

    Nov 30, 2010
    16,321
    6,818
    It would have saved some time if we'd known that. The current mirror circuit doesn't have PWM. That eliminates it immediately.

    I can't understand why you need an over-voltage limit. There isn't any over-voltage problem to consider with LED circuits. LEDs don't puncture from excess voltage because they are always on (never try to stop the current). The only way an over-voltage event could be imagined is if an LED burns out, at which point it stops the whole series string, and, if it shorts, the other LEDs are still forward conducting and don't care about voltage because they are current driven devices.

    The LT3755 chip is a current driver. Still no over-voltage problem available.
    The only over voltage situation possible is created by that chip in it's feedback loop, and then the chip shuts itself off.

    There are simpler ways to create PWM, but high-tech is fashionable, so, happy soldering on your 16 pin surface mount chip.

    https://www.digikey.com/product-search/en/integrated-circuits-ics/pmic-led-drivers/2556628?k=LT3755
     
Loading...