LED tutorial demystifies several control techniques

Controlling LEDs is really quite simple. As you know, they need to be current limited which is as easy as applying Ohm’s law to your given set of values. To make things even more even there’s a slew of constant current LED driver chips out there that can be had for a song. But do you have any idea how those constant current circuits work? If not, then [Giorgos Lazaridis’] guide on LED driving and controlling methods will bring you up to speed in no time.

He starts out with the most basic concept, how to light an LED using proper current limiting resistors. But from there he moves on to the juicy bits. He builds a transistor-based constant current driver, then adds voltage regulation for the circuit as seen in the schematic on the left. He moves on to the more robust and efficient method on the right which pairs a MOSFET with that transistor circuit. This is the technique found on each pin of many of those constant current drivers and functions well regardless of the voltage input level.

He’s been producing videos to go along with these articles. After the break you can watch the episode that accompanies the schematic on the left.

[via Reddit]

30 thoughts on “LED tutorial demystifies several control techniques

  1. The PWM circuit described there is not the greatest. The Disc pin on a 555 is not meant to sink very high current and the fact that the transistor begins to act as a current source means that it is kind of overloaded.
    Then, the pwm circuit, like the rest of them, will still be linear, meaning that when it delivers power tot the LED the efficiency still stays around VLED/Vsupply, so still about 20% for driving a green LED from 12V. It is still not a switching supply…
    I wrote about linear LED supplies a while ago.. http://www.electrobob.com/linear-led-power-supplies/

  2. I seem to be missing the point. The con of the resistor method is power wasted across the resistor, but every other design he demonstrated listed a con of power wasted across the transistor(s). Since he didn’t bother to quantify the amount of power wasted in each scenario I fail to see the benefit of adding transistors to the design so that you can waste power across them instea of across a simple resistor.

    Also, a pox upon everyone designs atrocious “open a new window for every page” websites like that. If I were in congress I’d be willing to sponsor a bill calling for mandatory jail sentences any time a web designer causes a browser to open a new window when the user didn’t select the “open in new window” option from the context menu.

    1. [shrug] didn’t happen with Firefox, out of curiosity didn’t happen with IE either, both opened links at eltrobob in the current tab. Personally I right click on a link to be able to open it in a new tab. While I have FF tabs options set to do that, it doesn’t do so all the time.

  3. Totally useless,no one controls LEDs this way.
    Low-power LEDs do not require constant current.
    And for high power LEDs, these designs are absolutely inefficient

    1. There is a case for linear regulation if you have several high power diodes in series and you operate it from high voltage.

      For example, your supply voltage is 120 volts DC, and you have a string of LEDs that need 350 mA and the sum of their treshold voltages is 110 volts at 350 mA.

      Now your linear regulator has to dissapate 10 V x 0.35 A = 3.5 W while the LEDs will see 110 V x 0.35 A = 38.5 W which means you dissapate only 8.3% of your power in the linear regulator, making your circuit 91.7% efficient.

      That’s much simpler than building a switching mode power supply for the purpose, or trying to source a suitable chip to do it for you. It’s probably much cheaper as well.

      1. Tim is correct. This was just a learning exercise. An simple in-line resistor is the only circuit actually used in production.

        That’s not even ‘constant current’. It’s the ol’ ‘slap-a-resistor-in-there” method that changes current with applied voltage and ambient temperature.

        Your example is a very specific case with no reason why the transistor circuits would have an advantage over the single high watt resistor.

        Don’t be scared of switchers.

        In your example application I would use a LinkIn LNK302DN offline switcher and the four other support components (under $1.00 total).
        Short-circuit, open loop fault, auto restart, significantly higher Vin breakdown voltage, significantly better current regulation, nearly zero waste heat compared to your example, significantly smaller footprint, can double as on/off switch, etc etc.

  4. @Paul, right about the new window thing. And the power dissipation quantification. The fact is, for all linear circuits it is almost the same, weather a simple resistor or more complicated. The simple resistor does its job well when used from a stabilized supply(such as a microcontroller running at 5V) and when the LED is not prone to overheating.
    The other circuits try to maintain a constant current through the LED when the supply varies (except the one with a single transistor which is no better than the single resistor variant).
    When efficiency is at matter, switching supplies are needed(usually this is the case for high power LEDs used for lighting, not merely indication/display)

    1. The only advantage of the single transistor circuit seems to be that you can throw in any LED without calculating a new resistor value for it.

      It’s less stable to voltage variations than the simple resistor, because the circuit acts as a voltage-current amplifier.

    2. right, i don’t see the point with the first single transistor either: it looks like you just replace a resistor with a transistor set to a fixed resistor value.
      but the second single resistor circuit does regulate: the zener holds the base on a constant potential, while the potential on the emitter depends on the voltage-drop of re, which changes with current. v_be and therefor v_ce change with current. you basically have a resistor that adjusts to different voltages or leds.

  5. I don’t see how one controls more than one LED with the simple transistor circuit, or why it is necessary for that purpose.

    Surely if I put the LEDs in series, I can still use a simple resistor just by summing up their operating voltages together and using that to calculate the required series resistance.

  6. couple of things to add:
    -linear regulators aren’t that bad once you have a “few” leds in series. an ideal example: 8 leds 3.6v eachin series on 30v (old printer power supply) would be 28.8v and at 350ma they’d consume 10watt, with only a half watt wasted at the regulator: 95% efficency

    -pwm becomes interesting once you add a feedback and the pulse width can be adjusted by the “needs” of the led. without feedback it is basically like a linear regulator switching on and off, or worse no regulator. the typical arduino pwm is like his first example, led with series resistor, just that the power is switched on and off. which is of course more efficient than dimming with just a resistor, but most often not a good way to dim power leds.

  7. People, please, it is a THEORY page, not circuit demonstration, nor project worklog. It is there to learn basically.

    Some said that a series resistor is all needed to do the job and can’t find the point for the other circuits. Forget about the “learning” point. A friend of mine starts a project with a 10 Watts LED for his motorcycle (will be posted in the site soon). The motorcycle produces 12-14 volts (depending on the rpm) and the LED needs around 10-12 volts @ 1A current to operate (something like that). Try to make this work with one resistor.

    Others said that the single transistor with voltage divider is the same as a single resistor, and they cannot find a point why use this, since it has (indeed) no voltage regulation. Well, here is one point: You can turn this into a linear supply simply by replacing one resistor with a potentiometer and i explain how in last pages. But mainly, the reason was to learn how transistors work.

    One posted a comment with his page (Bogdan first comment) which is nice indeed. I had post this link already to this site because i found the 2-diodes alternative from there. I found it as a nice thinking, although it did not work that well, since diodes are not intended to work like this. Nevertheless, i felt that i had to give the required credits to this site and i did.

    Finally, about the “open in new window” thing. The only thing that opens in new window are the images and the datasheet – hope that you people understand why. Clicking on “next article pages” wont open in new window. So don’t rush pushing me any legal charges :D (just kidding).

    1. “Try to make this work with one resistor.”
      I guess if you want to cobble things together…ok… use a 2ohm resistor. With or without transistors that’s 2W of wasted heat.

      If you want to do it correctly..I would recommend a switcher. Greater than 95% efficient, smaller package and likely cheaper. Additionally the motorcycle’s voltage will see spikes upwards of 30Vdc and brown-out conditions near 8V. If someone tries to jump the battery incorrectly it can easily see a continuous 24V. Probably enough to blow a single resistor or single transistor solution. If it doesn’t blow, the LED current/intensity will change during these situations…. well not with the switcher.

      1. it won’t change brightness if a zener and a single transistor is used. With a single resistor it will certainly drop brightness. I do not really know where you found those 8 volts, i said “12 to 14″. I do not doubt about the power dissipation, but 2watts is an amount that my friend is probably willing to spend……..
        As for the SMPS, it is indeed a better solution thanks for sharing with us (along with all others who suggested it), and 95% is a nice number.

        But with very quick and dirty calculations, the 2-transistor circuit (or single with 2 diodes) has total power dissipation (on transistor +re+rb+diode) about 1.1 Watt, for powering a 10W LED with Vf=10V If=1A (worst case scenario). That is 90% efficiency. It is not as slick as an SMPS, but it certainly won’t be the cause of global warming.

      2. sorry for the second reply. right now i’m designing a buck dc dc converter with the a6210, which is considered to be one of the most efficient buck constant current led drivers. I will power the chip with 14volts to supply a 10V 1A LED array. The max efficiency i could theoretically get was 93.53$. I’m not sure if the 95% you mentioned is an easily achievable and realistic efficiency. Nevertheless, as you see, i won’t use a single or dual transistor driver for my next project (kitchen lights) but the A6210 – NOT because it is more efficient, but because i like QFN components, tiny PCBs and slick designs ;)

    2. “Finally, about the “open in new window” thing. The only thing that opens in new window are the images and the datasheet – hope that you people understand why. Clicking on “next article pages” wont open in new window.”

      Starting from the page linked in this article brings you to the presentation page, you only have the option of “Go to This Page” links which DO open in a new window. However, if you click “Worklog” at the top, you get the option of clicking the next page, which is displayed as an image at the top of the worklog.

      I see no link titled “next article pages” anywhere.

      1. damn you are right. my bad JamieWho. I will fix it. You see, i’ve designed the new CMS some 2 months ago and still has some bits’n’bites to fix… As for “next article pages” i meant the indexing images on top which change pages.

    3. @Gio
      I don’t understand where you are getting anything less than 4W of wasted power. A good understanding of Kirchhoff’s laws will tell you are going to waste AT LEAST 4 Watts (71% efficient) trying to linearly drop 4V at 1A.

      As with most automobiles, a motorcycle’s supply voltage is awful. If you wish for your LED project to work well consider the supply more like 8 to 19Vdc with a potential 24V continuous non operational and 30V transients… just based on experience and testing.

      “a6210, which is considered to be one of the most efficient buck constant current led drivers”
      …by who? Look at what’s out there. There are plenty of better solutions.

      I have a modest hourly rate if you would like me to make further recommendations.

      1. Well, he’s right. Your 10W LED will need about 10.5V and your constant current circuit will have to drop the difference from the supply voltage.
        If you feed 14V into the circuit then the 3.5V difference will become heat in some way or another.
        I don’t even want to imagine a scenario where you feed 24V into that circuit…
        BTW, this phrase “If the maximum forward voltage is exceeded by little, the current is increased logarithmically” is weird. I think you mean “exponentially”

      2. @Dumitru Stama yes i know, but i kinda rushed and did not explain correct. The led is 10W, and i measured about 800mA at 12V (it can go higher but has no really brightness increment). The maker states 10-12 @1A but we measured it. The supply is measured NOT on the poles of the battery nor at the output of the alternator, but at the wires that arrive at the lamp that the bike has, so it is almost stable between 12 and 14V. A mistake i made was that i did not add the dissipation on a diode that was calculated to use (but probably won’t) for reverse polarity. Without this diode the power is about 1.6 watts (86%). When the bike is at low rpm (not moving) the voltage drops at 12V so the LED will be dimmer, but that is no problem.

        The most serious problem in the make that my friend will have to find a cool solution (and i really mean “cool”), is how to dissipate the heat from the LED. It’s been a long time when he began and i really do not know what he figured out, but believe me, this LED turns VERY hot, even at 800ma. Imagine what happens at 1A!!!

        Thanks for the “exponentially” as well, this is exactly what i meant …. :D

      3. @Gio
        I can not tell. You do realize that with 14V applied to a 10V 1A load you have to dissipate 4W (71% efficient) using any of the non-switching circuits, correct?

        My goal wasn’t to entertain, but rather educate. I hope I was successful.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s