I want to drive multiple strings of XTE LEDs (total of ~36 LEDs). The relevent (-ish) specs are as follows:

  • I_f,max = 1.5 A
  • Desired I_f = 0.5 A
  • V_f = 2.94 V (0.5A @ 85C)

According to some calculations, I found that the larger the voltage dropped across the resistor the more stable the LED current is with shifts in supply voltage. Voltage drop ratio between resistor and all LEDS in a single string determines some part of current stability. Is this generally true when looking at voltage source deviations? I have looked around and haven't found something that explicitly confirms this.

But I don't want to use a large resistor as too much is dissipated in the resistor. For simplicity, I want to choose a small [resistance] resistor and use a power supply with tight voltage tolerance (1%, according to some Meanwell HLG specs). Will the right PSU be enough to maintain a fairly consistent voltage and thus consistent current?

Here is my follow-up concern, will thermal runaway be a large concern because of the small resistors? Say I use 2 strings x 18 LEDs = total of 36 LED. This is using a constant Voltage source, not a CC source.

How big of a voltage difference should I expect between each string of LEDs due to dissimilarities at the per LED basis?

Would splitting up into smaller series strings like a 4 x 8 LEDs while still using the same ratio of (Voltage drop over Resistor)/(Voltage drop over all LEDs in a string). Is thermal runaway better/worse/no different in this scenario?

  • 2
    \$\begingroup\$ You really should consider using a CC supply for this if you can. \$\endgroup\$ – Ignacio Vazquez-Abrams May 19 '16 at 20:12
  • \$\begingroup\$ You can use a CV source with a resistor (a large one) to limit the current based on your source voltage, but if an LED shorts out then the current increases and the other LED's get brighter. Also the brightness will change with the supply voltage. A CC supply accounts for all of these variations. \$\endgroup\$ – Sparky256 May 19 '16 at 20:17
  • \$\begingroup\$ What is your planned power source? AC to DC? An off-the-shelf DC supply? \$\endgroup\$ – Sparky256 May 19 '16 at 20:19
  • \$\begingroup\$ With a CC supply and parallel strings of LED, wouldn't thermal runaway draw the current away from the other string of LED? You would still need a limiting resistor correct? AC to DC \$\endgroup\$ – Mr. Poopsicle May 19 '16 at 20:23
  • \$\begingroup\$ That's why sane people don't use one CC with multiple parallel strings \$\endgroup\$ – Asmyldof May 19 '16 at 20:27

The LT3517 may be used with 30 Vin and a string of 9 LEDs. For 36 LEDs four strings with an LT3517 for each string may do.


You ought to consider a switching LED driver like this: -

enter image description here

Just google "1 amp buck led driver" and look at the thousands of images of solutions. It's a constant current driver with very little heat loss. You can get them in various shapes and sizes and some even have a boost regulator on board to drive several LEDs from a low voltage source.

  • \$\begingroup\$ Not really useful since he wants 18 LEDs in series so 55 volts. Fortunately boost LED drivers are fairly common, unfortunately they are more difficult to design as current stress on components is more than you'd expect. \$\endgroup\$ – Barleyman May 19 '16 at 21:28
  • \$\begingroup\$ In a comment he mentioned his source was AC power, so he's bucking presumably. \$\endgroup\$ – Harper - Reinstate Monica May 19 '16 at 23:58
  • \$\begingroup\$ @Barleyman of course it's useful - I didn't say he should use this one - I said he should consider a switching driver like this. I'm not going to track down the exact one that meets his requirements. I helped him by telling him what to google so I disagree with you. \$\endgroup\$ – Andy aka May 20 '16 at 8:23
  • \$\begingroup\$ @Harper that comment ought to be in the question and anyway, he probably made it after I made my answer. \$\endgroup\$ – Andy aka May 20 '16 at 8:24
  • \$\begingroup\$ @Harper I've never heard of someone using buck for AC/DC, that's what flyback is for. \$\endgroup\$ – Barleyman May 20 '16 at 8:28

What is your power (voltage, current) availability? There are online calculators that will give you the optimal parallel/series connection scheme for the LED parameters and voltage available.

Example: http://led.linear1.org/led.wiz

  • \$\begingroup\$ I guess what I am trying to get at is how thermal runaway effects should be considered. Based on those considerations, I can pick a supply voltage, a resistor value, and LED string configuration for the best performance. \$\endgroup\$ – Mr. Poopsicle May 19 '16 at 20:49

18 LEDs in series would be about 55 volts, at 0.5 amps that's a fair amount of power in one string. You probably want to use active current controller, simplest would be a mosfet with a series resistor and a transistor pulling the gate closed when 0.5A is achieved. Be aware that the forward voltages can vary a fair bit and with 0.5A you can end up dissipating a lot of heat in the current regulator. I actually designed fully analog LED controller circuit with feedback to boost converter to control excess voltage recently.

It can all be done with exploiting voltage drops over components to "trigger" transistor base to achieve pull-down.

I'd try to find LED boost controller first however which will do it all for you.

Edit here is an example of the kind of circuit I'm talking about. They both do essentially the same thing but the 1st one is much easier to understand. Constant current circuit with transistors/MOSFETs design comparison

If you don't need PWM then just add a pull-up resistor. That 100k is somewhat excessive though. Bear in mind most FETs are limited to 10 or 20 volts at the gate so you have to do voltage division if you're going to use your 55 volt supply.

  • \$\begingroup\$ Thanks. How reliable are transistors in current control, especially to drive strings of LEDs? I understand a little bit of the type of circuitry it may involve, but the transistor would be switching almost constantly? If the transistor were to close at 0.5A and then open when its base-emitter voltage is reached at a rate which I don't really know, would the effective current be 0.5A? \$\endgroup\$ – Mr. Poopsicle May 19 '16 at 21:54
  • \$\begingroup\$ @Mr.Poopsicle it actually "seeks" equilibrium, there's a feedback loop but it's very simple and it won't give you any trouble if you have the transistor and mosfet side by side. I added a link to similar question. \$\endgroup\$ – Barleyman May 19 '16 at 23:43
  • \$\begingroup\$ These sound like LED's that need to be epoxied to a metal plate or they will burn up. \$\endgroup\$ – Sparky256 May 20 '16 at 19:09
  • \$\begingroup\$ @Sparky256 nah, that's about 1.5W a pop, you definitely have to manage the heat somehow but it's not too bad presuming you've got a proper PCB with generous helpings of surface copper and it's attached to a heat dissipating structure like a metal frame. This is presuming the LEDs actually have reasonable separation. I'm worrying about my 200W led boost circuit as there you can get 3-4W in a single mosfet if someone changes the binning mid-way through the display. \$\endgroup\$ – Barleyman May 20 '16 at 21:05
  • \$\begingroup\$ @Barleyman. Surface copper may be good enough, but they do fall into the category of 'High-Power' LED's. I always put mine in series so only one supply and constant current sink is needed, and one place to adjust LED brightness. 36 in a row needs about 110 VDC to turn on. 120vac converted to DC is about 170VDC. Bit of a shock hazard until it is done. Even 55VDC will cause a mild shock. \$\endgroup\$ – Sparky256 May 20 '16 at 22:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.