How to protect LED from overvoltage

Discussion in 'Analog & Mixed-Signal Design' started by LordOfThunder, Nov 24, 2018.

  1. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Hello!

    I am building a simple circuit to light an array of 9 LEDs in parallel using a signal generator.
    Since I will not be the only one to use the device, I want to insert some form of protection for the LEDs in case the input voltage is set too high by mistake.
    I have read that, in this case, you should use a Zener diode in parallel with the LEDs. Is that correct?
    The LEDs that I am using are these ones: Datasheet.
    The maximum forward voltage is 3.7V so I was leaning towards this 3.6V Zener diode: datasheet
    So if the voltage is over 3.6V the Zener diode would enter in the breakdown region and absorb all the excessive current maintaining the voltage drop constant at 3.6V. Is my understanding correct? In case could you point out a better-suited diode?

    PS I will also use some ballast resistors to account for the differences in the forward voltage among all the LEDs. The final purpose of this circuit is to make all the LEDs emit the same light amount (have the same luminosity) as possible.

    Thank you
    Jojo
     
  2. Sensacell

    Moderator

    Jun 19, 2012
    2,016
    632
    A zener has a very soft knee, and are generally not very accurate.
    The I-V curve of the LED also varies unit-to unit, trying to shunt current around them based on voltage would not work well.

    What you really want is a current control mechanism, the simplest would be a constant current regulator IC.
    https://www.digikey.com/product-det...ed/AL5809-20P1-7/AL5809-20P1-7DITR-ND/5030209

    One of these regulators in series with each LED would make it almost bulletproof, save for reverse polarity, which you could protect against with an additional silicon diode in series with the main power input.
     
    LordOfThunder likes this.
  3. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Thank you very much for the quick reply. I have briefly read through the datasheet but I have still a couple of questions:
    • With those current regulator ICs, will I be able to tune the LED luminosity?
    For our experiment (calibration of Multi-Pixel Photon Counters MPPC) we need to tune the LED luminosity so that we have on average ~20 photoelectrons hitting the MPPC and I cannot know which value of the LED current is optimal a priori, so I need a way to tune the LED luminosity. I was thinking about increasing or decreasing the signal generator voltage so to have a different Voltage at the ends of the LED and tune the forward current like that.​
    • Actually, I was thinking about putting some potentiometers instead of normal resistors as ballast resistors, so that I could tune the current of each LED individually. Is this a good idea? Are ballast resistors still needed if I use those current regulator ICs?
    Just to be clear. I want to evenly illuminate a flat surface of roughly 30x50cm. I have already bought a diffusor slab to diffuse the 9 LED light. Then I would like to individually tune the LED so that the target surface is illuminated evenly at roughly 5% accuracy.
     
  4. Sensacell

    Moderator

    Jun 19, 2012
    2,016
    632
    Are you pulsing or modulating the current? at what frequency?
     
  5. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Pulsing the current.
    The final goal would be one pulse every 580ns (so 1.7MHz if I am not mistaken).
    Actually, the final goal is to have only 8 consecutive pulses followed by a dead time of a few ms to let the electronics process the signal.
    The pulse width and the voltage are to be tuned so to have only ~20 photoelectrons hitting the MPPCs.
    These numbers are not randomly chosen but are set by other parameters of our experiment (particle physics).
     
  6. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Actually by giving it a little more though I would like the pulse to be as short as possible so that all the photons would arrive roughly at the same time on the MPPCs so that when doing the hitting time calibration we can assume that the photons arrive all together to the MPPC surface.
     
  7. Sensacell

    Moderator

    Jun 19, 2012
    2,016
    632
    Ok, let's start over.

    The requirements are rather demanding.
    You are out of the casual "LED and resistor" zone, you need a more sophisticated driver.

    Why are you worrying about over voltage?
     
  8. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Because this device may be used by other students in my laboratory (I am a Physics Ph.D. student but there are many undergrads).
    I am worried that one of them may apply too much voltage and fuse the LEDs.
    Anyway, if the overvoltage protection is detrimental to the luminosity uniformity, the latter has the priority.
     
  9. crutschow

    Expert

    Mar 14, 2008
    20,219
    5,717
    To limit the generator voltage, use a series resistor and shunt Zener at the output of the generator, before the LED resistors.
    How much current do you anticipate running through the LEDs and what is their maximum current rating?
     
    LordOfThunder likes this.
  10. danadak

    Well-Known Member

    Mar 10, 2018
    1,881
    393
    You are driving the leds in parallel. Why ?

    In general one does this in series to get more uniform lighting. LEDs are
    non linear, subject to falling Vf with increasing T. In parallel, unless you
    have current equalizing series R in series with each LED, you are asking
    for trouble.

    Regards, Dana.
     
  11. ebp

    Well-Known Member

    Feb 8, 2018
    2,213
    774
    If you want to use individual ballasting resistors, which is probably the best compromise method given that you need fast response, then the larger the voltage across the resistor the better. This can serve two functions - limiting the maximum current if the signal generator is turned up (e.g. simply scale the resistors to the maximum voltage from the generator at the maximum allowable current), and it reduces the variation in current with the temperature of the LEDs. It's simply a matter of response of the resistors versus that of the LEDs. The LEDs are non-linear and the forward voltage varies with temperature. Resistors of any reasonable quality are very linear and have low temperature coefficient of resistance. The larger the fraction of the input voltage that appears across the resistor the lower the fractional uncertainty due to the characteristics of the LED. Think of a constant current source providing 1 ampere. If you use 1 ohm and 1 volt, an extra half an ohm or a half voltage change is a big error. If you use a million volts and a million ohms, you can the resistance by hundreds of ohms or the voltage by hundreds of volts and still produce only small error in the current.

    Any attempt to use active current regulation for each LED is going to be something of a problem with the required speed. The requirement for individual "tuning" makes a common current source largely unsuitable. A current regulator that is based on an op amp will be forced into "large signal" condition (that is, where the amplifier is unable to maintain the inputs of the amp at virtual short-circuit) which leads to saturation of transistors in the amp and slow response. This is not insoluble, but tricky. "Open loop" current regulators that rely on simple bipolar transistor circuits won't likely be nearly adequately stable. Current sources or sinks based on current mirrors might work well, but require a good deal of care in design and use of matched pairs of transistors that thermally track each other (not too difficult to obtain as long as matching only needs to be "OK" and not really good).
     
    LordOfThunder likes this.
  12. ebp

    Well-Known Member

    Feb 8, 2018
    2,213
    774
    Any likely method you use to apply and remove power from the LEDs is going to have rise and fall times equal to a travel distance of a few feet for a photon. Achieving sub-nanosecond rise and fall is possible, but not easy, especially if you need clean transitions with minimal aberrations (undershoot, overshoot, non-monotonic settling, etc.)

    Also remember to consider the effects of terminating a connecting cable between the signal generator and the LEDs in anything other than a pure resistance at the characteristic impedance of the cable. Many signal generators are series terminated at the source, which is helpful.
     
  13. KeepItSimpleStupid

    Distinguished Member

    Mar 4, 2014
    2,918
    510
    You probably should pre-select the LEDs based on Vf (Forward Voltage) and you MIGHT be fine for uniformity.

    Uniformity ..... Yuk Yuk! Yuk!

    I can only describe what I did for a non-LED uniformity. We wanted 100 mW/cm^2 within 5% over a 4x4" area.
    We used four 120 V 300 W ELH lamps. These have dichroic reflectors, so the heat goes out the back of the lamp.
    The filaments do have vibration issues.

    In these lamps the spectral response shifted with voltage and that will happen with LED's too. Thus the area to be illuminated was about 2-3' away from the lamps. Height adjustment moved the lamps only. All 4 at the same time. There was a ground glass filter that helped the uniformity. The lamps could be clamped at nearly every angle.

    We never developed a good way to do a uniformity test except manually measuring for a sensor that was 1 cm x 1 cm square.

    I know more than what I'm saying, but it does not apply to LED uniformity.
     
  14. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    I don't know how much current will flow through the LEDs. That is to find during the calibration. The maximum current rating is 80mA.
     
  15. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Our current signal generator can only output 5 volts. So it cannot even power two LEDs in series.
     
  16. crutschow

    Expert

    Mar 14, 2008
    20,219
    5,717
    Then for likely resistor values to give the current you want, you will not be able to overdrive the LEDs by cranking the generator to its max output.
     
    LordOfThunder likes this.
  17. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Thank you for the very detailed reply. As far as the passive components explanation (resistors in series) is concerned, it corresponds to what I have read online and that is why I have chosen this design.

    For the current sources or sinks based on current mirrors, it may be worth a little bit of research. Could you point me out in the right direction?

    Consider that every physicist that I know that has done something similar to what I am doing, has only used one les with one resistor of 100 ohm without even knowing why the resistor is needed. In general physicists are not so experts in electronics as it may seem.

    I would like to break this cycle and do something more accurate but I am not sitting for the Moon either.
     
  18. wayneh

    Expert

    Sep 9, 2010
    14,966
    5,434
    There's a lot going on here with all your other specifications, but this one has many solutions. Some are as simple as clear labelling, using connectors that prevent the "wrong" hookups, and so on. For instance if you intend 5V to be the supply voltage, I'd be tempted to use a USB plug. No matter what USB source they might plug in, they're all 5V.

    If you still think you need your device to internally protect against >5V being applied, there are several solutions. One would be to control a relay or a transistor to switch on the power to the rest of the circuit, and only do that when the supply voltage is within your specified range. The problem with a simple shunt circuit is that you don't know how much current you might need to shunt. For instance if you attached a 12V car battery, you'd have to shunt an enormous current. Of course there are ways around that but adding more and more complexity is probably not what you want.
     
    LordOfThunder likes this.
  19. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Thank you for the answer. Anyway could you please explain a little bit more why a
    I know but in future someone may try to connect something else. Do you think I could get by without the resistor in series with the shunt zener? That is because I would like to maximize the resistance in series with each led to improve uniformity.
     
  20. LordOfThunder

    Thread Starter New Member

    Jun 27, 2018
    22
    2
    Thank you for your advice. Anyway I need to pulse the LEDs and the USB connection that is compliant with the USB specification. You are right complexity is not something that I am very happy to consider at the moment.
     
Loading...