LED Driver bias problem

Discussion in 'The Projects Forum' started by calegatter, Aug 26, 2013.

  1. calegatter

    Thread Starter New Member

    Aug 26, 2013
    6
    0
    I am using an LM324 and three MOSFETs to drive three 44V 40W led's. The problem I am having is the LED's don't shut of all the way. If I use only one op amp (out of the four on the chip) and MOSFET, the LED turns off properly. If I use three of the four op amps, the LED's do not turn off.

    I have tried placing a resistor (20K) in line with the negative feedback to see if I was having a current bias problem, but that didn't help.

    One thought was my issue comes from the fact that the LM324 isn't rail to rail, but I am not sure about that since I can get it working with one op amp. If the input voltage is zero, I am seeing about 1.5V on the output of all three op amps.

    Any help would be greatly appreciated. Thanks.
     
  2. wayneh

    Expert

    Sep 9, 2010
    12,086
    3,024
    Do your +12V and +44V supplies share the same ground? How about the source of the input?

    You mention MOSFETs but have drawn darlingtons?
     
  3. #12

    Expert

    Nov 30, 2010
    16,247
    6,744
    The spec sheet says this should work. The input range includes zero and there is a sink transistor inside the chip. Maybe you have a sick IC?
     
  4. calegatter

    Thread Starter New Member

    Aug 26, 2013
    6
    0
    I switched to MOSFETs. I thought the base-emitter current may be causing the trouble so I tried the MOSFETS. However, the results were the same.

    Yes, I am using a common ground for all supplies. I have tried several LM234 IC's with the same results. The only success was disabling two of the three LED's. Then the driver works as expected. I also tried buffering the input signal by using the fourth op amp as a voltage follower. This made the problem worse.

    Do you know of a rail to rail drop in replacement for the LM234 that can handle a 12V power supply? I may try dropping in another quad op amp and see what happens.

    Is there any reason that the op amps would be interacting with each other, within the IC? I find it strange that this works in a one op amp configuration but not three.
     
  5. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    Another test. I know you said 0 volts in still shows about 1.5 volts out, but what happens if you ground the inputs?
     
  6. calegatter

    Thread Starter New Member

    Aug 26, 2013
    6
    0
    Same thing with grounding.
     
  7. LDC3

    Active Member

    Apr 27, 2013
    920
    160
    Are the inputs to the unused amp connected to ground?
     
  8. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    Why not try a different configuration:
    Move the resistors to the top side, ground the Mosfet sources. Maybe try it with the the Darlingtons also. BTW: What voltage is the input signal?
     
    Last edited: Aug 26, 2013
  9. calegatter

    Thread Starter New Member

    Aug 26, 2013
    6
    0
    Are the inputs to the unused amp connected to ground?

    Is it common practice to tie unused op amp pins to ground? In my case, the original circuit used a voltage follower so all op amps were used. In my test with one LED (the one that worked) I left the pins floating and the circuit worked. I know this is not good practice with microcontrollers, but it didn't occur to me that this would apply to op amps.

    Tubeguy, do you mean creating a high side current source like the one show here? I'll give it a try. Is there an advantage to doing this?

    Are there any other recommendations for current sources other than this circuit? This circuit seems to be quite common which is why I selected it. I am not concerned about efficiency. Simplicity and speed is of the greatest importance for my application.
     
  10. tubeguy

    Well-Known Member

    Nov 3, 2012
    1,157
    197
    What is your "On" input voltage ?
    No, I wasn't talking about a constant current source in my last post. But likely that is what you do need.
    What is the supply voltage for the 44v Leds?
     
    Last edited: Aug 26, 2013
  11. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    Tie the unused amp inputs to ground, then put a decent sized cap (2.2uF to 10uF tantalum) to ground on the main input that goes to the amp + inputs. It is likely to be an oscillation issue.
     
  12. LDC3

    Active Member

    Apr 27, 2013
    920
    160
    For most CMOS ICs, it is recommended to connect the inputs so they are not floating. For TTL ICs, it does not matter. Most MPUs have tri-state data lines so it does not matter. I think someone else had a similar problem until they connected all their inputs.

    Added: And added a capacitor.
     
    Last edited: Aug 26, 2013
  13. JoeJester

    AAC Fanatic!

    Apr 26, 2005
    3,373
    1,157
    I think you need to change your topology. As it stands now, with the 1.5V or even .7 volts out of the LM324 will allow some current to flow through the Darlington transistor and not turn off the LED completely. You need to reduce the output of the LM324.

    Attached it a comparison of two topologies.
     
  14. calegatter

    Thread Starter New Member

    Aug 26, 2013
    6
    0
    @Tubeguy, I made a mistake in my original post. The LED's are 40V and the power supply is 44V. My "On" input voltage is 3.4V, the digital output level of a DSPIC i/o line.

    @THE_RB, in this case I don't think I can bypass the input. I am using this driver for a flash system.

    @JoeJester, What simulator is this? I am a bit confused on your use of the op amp. Can you explain it please? I am using this device in harsh arctic winters and I want to use a topology that allows for great stability over a large temp swing. Bias current is greatly dependent on temperature and the Beta value using your topology. Using my topology, I source high tolerance, low PPM resistors. Other than that, I need to control my voltage input and I should have a quite stable current source for my LED's. At least this is how it plays out in my head. I didn't actually sit down and do the calculations for comparison. I could be completely wrong and don't hesitate to say so.

    The other topology I was using was this. It seemed to work all last winter, but I was still concerned with stability. The op amp solution seems so elegant. I may default back to the transistor solution, if I can't get this working soon.
     
  15. DickCappels

    Moderator

    Aug 21, 2008
    2,647
    631
    The TIP122 has a 0.5 ma collector cutoff current. Is that about how much current you are getting through the diodes? If so, you might want to switch to a MOSFET for the switching.

    Also, the LM324 cannot sink more than 50 ua (1/10th leakage current from the collector of the TIP122) below about 0.6 volts, so that is probably driving the base of the first transistor in the Darlington, which will result in additional current through the LED. To fix this, you can put a diode or two in series with the base of the Darlington (cathode to the base) or use a resistor from the output of the LM324 to ground -1k to hold the base of the Darlington to at 50 mv should do it.
     
    Last edited: Aug 27, 2013
  16. JoeJester

    AAC Fanatic!

    Apr 26, 2005
    3,373
    1,157
    I use TINA, the free version is TINA-TI available at www.ti.com
    All I did was compare your original circuit looking at why the LED's didn't "turn off completely." I only used one-third of your circuits to see what was going on. In fact, I was confused on why you had the capactior in parallel to the LED as it leads to current spikes when the LED is initially energized.

    I used it as a differential amplifier. Mostly because I didn't know your input requirements, nor the operating enviorment.

    So, we now know you are designing a constant current source to provide ____ Amps to a 40V LED. This is to operate in the Artic environment where the temperature can range from ______ C to _____ C. The input to this circuit is the 0 to 3.4V level signals from a DISPIC I/O. This circuit must be completed before ___________ to ensure deployment to the Artic.

    If the topology used last year worked well, why change?
     
  17. calegatter

    Thread Starter New Member

    Aug 26, 2013
    6
    0
    @ JoeJester, The capacitors across the LED are a mistake. Thank you for catching that. I meant for them to be bypass caps like in your diagram.

    I'd like to drive the LEDs at 950mA each or slightly less based on available resistors (for my design). Assume -40 to 0 C for a temp swing. This needs to be completed in a couple of weeks.

    I am changing my driver design in response to complaints that the LED light output had degraded over time. I don't know if this actually true, but said I'd try a different driver this year. I know that individual transistors can be quirky. Op amps have precision trimmed components and use proper current mirrors, etc. to come as close as possible to an ideal op amp. I also prefer calculations dealing with op amps and the calculation for this driver is about as simple as it gets. Also, this op amp circuit driver is listed in my engineering texts and all over the web so it seems well documented and used. More so than say a high-side current source.

    BTW, I prefer an inefficient driver because it helps keep this device ice free.

    @DickCappels, your suggestions make sense but didn't work.
     
  18. THE_RB

    AAC Fanatic!

    Feb 11, 2008
    5,435
    1,305
    I work with TIP122 darlingtons in a product, they won't turn on until Vbe is around >1v.

    The problem here is likely to be oscillation (or noise on the signal line).

    It should really have a cap on the opamp inputs to ground! Otherwise any significant + noise on that signal line causes the LEDs to turn on.

    If you have a cap there the voltage on the + inputs will be an average voltage with no significant + noise content. You can add a cap there and still "flash" it at a few Hz, just make sure the cap is not excessively large.

    The cap will also provide a "soft" turn on and off feature to extend the life of LEDs and switching components.
     
Loading...