Multiple LEDs

Instead of using resistors, use a bank of current mirrors such that each string gets the specified current. Bu sure that the Vf is enough below 2V that your current mirrors have headroom, and you're good to go!

You can send your check for my share to: Richard Grise, Engineering Director ABI Engineering

12143 Rivera Rd. Whittier, CA 90606

Thanks! Rich

And if you have a burning desire to email me, I have a spamdump at snipped-for-privacy@yahoo.com - for my real email, elide ard. :-) R

Cool web site!

John

Asolutely. These are forward biased junctions whose conduction voltage is never going to be perfectly matched, so parallelling will result in unequal current sharing.

The real issues are: . exactly how "12V" is the 12V source; and . what is the actual forward voltage of the LEDs at the intended current level.

What you need to do is some simple sums for series strings of various counts and work out the minimum headroom at minimum supply voltage. Then select a device for providing constant current down each such string within that dropout voltage (headroom). The lower the device's dropout voltage, potentially (ouch!) the greater number of LEDs per series string and hence the lower number of strings/regulators.

Whether the "regulator" is a resistor, current mirror, LM317-type current source, bipolar transistor, or whatever, will come after those calculations.

As above re series/parallel connection.

I often use multiple LM317Z regs for series LED drive, but there is no "right" choice. Havibg said that, some solutions *are* better than others.

If the voltage across each LED varies even slightly with temperature ( or some other influence ), then a series resistor is a very unstable current limiter. ( by the way, white LEDs I've met drop rather more voltage than you suggest at full output )

there's no problem with using 6 in series, or even more, but you should use a controlled current source, which has the happy effect of dissipating zero waste heat.

Maxim and LTC do a whole pile of chips to do this stuff, though I don't think any of them will handle 120 at once. They can do chains of 8 or more.

Your best cct would be a switching step-up supply generating 120 times the drop of a single LED, and all the LEDs in series... sadly there would e a few other component selection and safety issues to solve.

Any time you parallel up 2 LEDs ( or stacks of ), the process differences will tend to mean one sucks a large the current and the other just looks sullenly on. But that might not matter to you, in your context.

David

Thanks! I wrote it myself. ;-)

Cheers! Rich

Thanks for all the comments, some interesting thoughts.

to

taken

don't forget that multiple leds generate heat -- and the millicandles you get out is inversely related to the amount of heat -- the big UV LED-emitters which are now used for curing polymers are air or water cooled.

second -- some of the recent UV LED emitters have a reflector and heat sink for each device -- this will improve your efficiency and the reflector (like a little microwave dish) can also serve as a heat sink.

Depending upon your application, you might get some interesting and surprising results by pulsing the LED's -- while output is generally a linear function of duty cycle, some processes have a reciprocity failure -- just like photographic processes -- more energy over a shorter period of time can be more productive.

I have wired 64 in series parallel with only one resistor. I have wired

9 high power LED's with no resistor and a variable supply for testing purposes. No problems. I suppose the LED's should be tested for consistancy. greg

The heat will cause a droop in light output when hitting them hard, but when you stack them up, its easy for heat to develop.

greg

No problems.

i think you've been lucky to have LEDs that were reasonably well matched in their forward drop characteristic. I've seen a five-in-parallel blue LED array count slowly around a ring - first one was alight, then the next, then the third, then the fourth, then the fifth. Then darkness.

I have never placed individual LED's in parallel. The ones I spoke of were in strings of 3 or 8.

greg

I suspect that the result would still depend on forward voltage matching - or the sum thereof for each string.