1
\$\begingroup\$

everyone!

I would like to know if I can avoid to use current limiting resistors by pushing my power supply to its limits (it already sounds not like a good idea... btw...).

Say I have a 3.3V power supply with 200mA current limit.

Question 1: is it safe to suppose that connecting 10 LED in parallel with no resistors (positive to 3V3, negative to ground) each one gets 20mA of current? (200mA/10)

Question 2: how much can I trust the power supply limit?

Question 3: pushing a power supply to its limit for a long time can cause any issue?

Thanks!

\$\endgroup\$
  • 3
    \$\begingroup\$ Are you sure your power supply is actively limited to 200 mA, or is it simply rated for that value? \$\endgroup\$ – Dave Tweed Feb 15 '15 at 17:36
1
\$\begingroup\$
  1. No, it is not safe to assume that the current will be divided equally between LEDs in parallel, as there will invariably be differences between the LED dies. They might be closer if from the same batch, but it is still a bad idea. What would be closer to working would be connecting them in series, using a higher supply voltage. Some cheap flashlights actually work that way, and depend on what could be crudely modeled as the Thevenin resistance of the batteries to limit current.

  2. You cannot trust a supply to not source more current than it is rated for, unless it is explicitly rated as having a controlled current limit. If it does, there will be a tolerance specification on the parameter which will tell you how much higher or lower it could be while remaining within specification. In contrast, if it is merely something such as a transformer core in a linear supply starting to saturate, that would be a fairly imprecise and unreliable limit unless that core were carefully manufactured and tested for saturation current.

  3. If the current limit is an intentional feature, then it may be designed to operate in current-limited regime - check the data sheet to be sure. If the current limit is a design artifact, then it may not be appropriate to operate there for an extended period of time (could overheat, etc).

In short, what you want to do is not a good idea in this form. However, connecting your LEDs in series and using some sort of active current regulation would be a good idea - non-trivial to get right, but potentially much more energy efficient than using resistors. I believe if you do some searching you will find designs for things like LED bicycle safety lights which do current regulation the right way, in order to maximize battery life.

\$\endgroup\$
  • \$\begingroup\$ Since I have to connect like 100 leds (the example I gave was, as said, just an example) it is almost not possible to connect them in series. I was quite sure that this was not a good idea, but I would like to know the main technical issues involved. \$\endgroup\$ – il_mix Feb 15 '15 at 17:53
  • \$\begingroup\$ I'd expect that for that you would probably end up organizing them as a number of serial strings. They might be driven with some sort of multichannel PWM regulator, possibly incorporating a step-up to a higher voltage than you were contemplating (thus allowing more in series). I remember seeing a voltage warning label inside a large LED backlight TV, which suggested they might be doing such. \$\endgroup\$ – Chris Stratton Feb 15 '15 at 19:05
0
\$\begingroup\$

If ten identical 20 mA LEDs are connected in parallel and all held at the same temperature then you'll have the same current through each and the total current will be 200 mA. This could work for small 5 mm devices at low current, less than 3 mA per LED, say, since everything will be close to ambient temperature. If you increase the current some of the (usually non-identical) devices will heat up more than the others and the current through them will increase even more, again heating them up even more. This will stop either when the current limit of the power supply is reached or when one of the LEDs fails (and then depending on how it fails other bad things can happen).

Power LED are even harder to control. Standard arrangement is one constant current supply per emitter. LED manufacturers make modules with many emitters in a single device placed close together for good thermal coupling. Such modules look like a single device but they have much higher voltage drop than a single emitter ones, 10-40 V as compared to ~3.6 V for white (blue).

\$\endgroup\$
  • \$\begingroup\$ The problem was not really to have all led with same current (while having "almost" the same current was the idea) but to avoid hardware failures with a non-standard configuration. \$\endgroup\$ – il_mix Feb 15 '15 at 17:56
  • \$\begingroup\$ Thermal runaway is the issue for parallel-connected LEDs. If you can keep current low enough you won't have the issue. Many LED backlight assemblies are constructed this way and work just fine even with appreciable temperature gradient over the screen area. \$\endgroup\$ – Oleg Mazurov Feb 15 '15 at 18:01
-2
\$\begingroup\$

There are several chips on the market specifically for driving large numbers of LEDs. Many of these will take a lower voltage and boost it to a higher voltage so you can string many LEDs in series. Some of these chips can manager multiple strings with just a single chip. The advantage of these chips is that they have low external component counts, are very efficient, and they are design with the current requirements of LEDs in mind.

Check at Maxim-IC and TI as a good starting point. I also use DIGIKEYs search function to find chips across manufacturers to find chips that serve a specific purpose. Often you can implement solutions for for less than 20 dollars that would extremely difficult to implement with discrete components.

\$\endgroup\$
  • \$\begingroup\$ This isn't really an answer to the actual question asked. All your answer essentially says is "Get an LED driver" \$\endgroup\$ – Doodle Apr 25 '17 at 15:17

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.