Questions about a Darlington (ULN2803) Circuit

Discussion in 'Programmer's Corner' started by blarg, Nov 13, 2012.

  1. blarg

    Thread Starter New Member

    Nov 12, 2012
    8
    0
    Hello everyone,

    I am designing a ULN2803 circuit for LED strobing, for a photic stimulation project.

    From what I can gather from the datasheet, the ULN2803 can sink up to 500mAh of current per channel, and I am driving eight LEDs that are independently PWM modulated for fiber optic coupling with this setup.

    What I am not clear on is how power can be distributed between all of the LEDs in this configuration. If the intended output current for each LED is 500mAh, how can power be equitably distributed from a single power rail to all the LEDs without overloading a single LED if that's the only thing lit up?

    Do I need separate 500mAh power distribution for each LED power leg to get the full intensity from each LED channel, should I be current limiting each LED to a common power rail that provides the total power required for all LEDs?

    Thanks in advance for your help, this is my first embedded project and I don't want the magic smoke to escape these LEDs.
     
  2. t06afre

    AAC Fanatic!

    May 11, 2009
    5,939
    1,222
    I do not think you can have a 500mA load on all 8 drivers at the same time. That will overheat the chip in no time. Take a look at this datasheet http://www.datasheetcatalog.org/datasheet/allegromicrosystems/2801.pdf and the graph at page two. Then all drivers have a load equal to 500mA. The chip has to dissipate 7 Watt. It will for sure burn. And the ground pin 9 will not be able to handle that kind of current either
     
    blarg likes this.
  3. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Yep, that's a great point. In terms of pins soldered and PCB real estate used, a ULN2803 is not much better than 8 FETs anyway. And if you used 8 FETs they can be good for far more than 500mA each, and/or will dissipate a lot less heat, many times less if you use good FETs.

    And i think to try to answer your question, yes each "leg" (all 8 of them) needs to have it's own resistor to set the current through that leg. All should work independently.
     
    blarg likes this.
  4. blarg

    Thread Starter New Member

    Nov 12, 2012
    8
    0
    Thanks for the feedback, I've been looking at passive cooling using some DIP heatsinks, but I wasn't aware that ground pin 9 would be problematic.

    Any idea what the maximum recommended current would be in this configuration?

    The LEDs are pulsed between 1Hz and 100Hz, so they wouldn't never all be on at the same time. However, I want as much intensity as possible for optical coupling with several light pipes that have a 90 degree bend in them which will absorb a good portion of the injected light.

    Also, any recommendations on individual FETs for each channel? Precision timing is key for this application, so the PWM output on the control board would have to be conveyed all the way to the LEDs with no delay and precision timing for the LED strobing.
     
  5. t06afre

    AAC Fanatic!

    May 11, 2009
    5,939
    1,222
    The power dissipation will depend on the duty cycle and current of course. Do you have any numbers for those. I have a feeling that a ULN2803 will not be the correct circuit in this application. But I am sure we can find another solution for you.
     
  6. blarg

    Thread Starter New Member

    Nov 12, 2012
    8
    0
    Duty cycle is random, as are the pulse intervals. Little if any synchronization between channels, each PWM output is independently pulsed. Most of the LED strobing would be done at 50% duty cycle in the 10-20Hz range. In this design there are two ULN2803s, for a total of 14 LEDs that are in turn optically coupled to light pipes. For the LEDs I am using, 100mAh is the minimum current, 750mAh max. My plans were to tie a current limiting resistor to each anode, to cap the current consumption at 500-750mAh max for each LED, and then feed power to all 14 LEDs from the same power rail.
     
  7. blarg

    Thread Starter New Member

    Nov 12, 2012
    8
    0
    A friend suggested the CAT4104 for this project, it has a total of 400mAh of output with four channels. No idea if the four channels can be tied together like the ULN2803 can. What sucks is that part is only packaged TFDN and SOIC, so I would likely have to spin a board just to start prototyping with it. It was also suggested to me to use aluminum clad instead of copper clad for heat dissipation, but I haven't researched any of that yet.
     
  8. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,670
    804
    That´s mA, not mAh.
     
  9. t06afre

    AAC Fanatic!

    May 11, 2009
    5,939
    1,222
    As said before. It would help if the OP. Did show us some simple schematic. It is badly needed.
     
Loading...