T&L Publications

Hello everyone! this is my first post, I hope i'm doing it right...

I am currently (and finally) slapping together my own take on the "freeze fountain" (as seen in N&V MAY 2008) but have a quick question about driving the LED circuit.

I see (as used in the article) using one 33 ohm resistor in series with 8 parallel LEDs will limit the current to about 20 mA per LED (with a 5V source).

20 mA is all safe and within the LEDs specifications, but I have heard that a person can drive an LED at (much) higher currents so long as it is being pulsed, which these LEDs are. ( these LEDs are pulsed on for 1 ms at a rate of 7 to 76 Hz)

MY QUESTION IS...
Can I "overdrive" these puppies to around 30 mA because of the short duty cycle? maybe get some more brightness out of them, to excite the fluorescent dye better? maybe use a 10 ohm resistor? Is there some special guideline to overdriving PWM-ed LEDs?

if anyone who is familiar with this project OR overdriving LEDs would provide some advice it would be greatly appreciated!
Thanks very much,
Greg

GregM WI NY wrote:I am currently (and finally) slapping together my own take on the "freeze fountain" (as seen in N&V MAY 2008) but have a quick question about driving the LED circuit.

I see (as used in the article) using one 33 ohm resistor in series with 8 parallel LEDs will limit the current to about 20 mA per LED (with a 5V source).

This arrangement is flawed. I wrote a letter to the editor,
and so far it hasn't been published. The parallel connection
of randomly selected LEDs or those of different colours
is incorrect. I noted that three articles in recent
issues have this problem. I heard from Brian K. (who runs
the store for kits) saying that he had a problem getting
LEDs (in the fountain kit) to be equal brightness.

If you have the current issue (Nov 08) the creators
of the PIC-mas tree reported trouble with matching
LEDs for their project (they are mixing LEDs of
different colours in parallel connections).

GregM WI NY wrote:20 mA is all safe and within the LEDs specifications, but I have heard that a person can drive an LED at (much) higher currents so long as it is being pulsed, which these LEDs are. ( these LEDs are pulsed on for 1 ms at a rate of 7 to 76 Hz)

Yes, the LED average current must be below the
manufacturer's Abs Max rating, as seen in the
datasheet. Its a thermal management issue, the
high power LEDs (LUXEONs, etc.) have metal
heatsinks.

If your driver circuit is MUX'd for, say, 4:1 the
LED current can be a peak of four times the
desired average. 80mA for 20mA DC in your example.
The upper limit is imposed by the fusing strength
of the die and bond wires, so LED data sheets
will state three parameters (max power, max
pulse current, max DC current). Where DC =
average. Further, these must be derated for
elevated temps.

GregM WI NY wrote:Can I "overdrive" these puppies to around 30 mA because of the short duty cycle? maybe get some more brightness out of them, to excite the fluorescent dye better? maybe use a 10 ohm resistor? Is there some special guideline to overdriving PWM-ed LEDs?

A definite "maybe". If you know the brand and
part number get the datasheet. In some cases
driving the LED harder is not helpful as the light
output is not linear to current, again there should
be a graph in the data sheet. Light is proportional
to LED current, the diode voltage is fairly constant
over the useable range.

GregM WI NY wrote:if anyone who is familiar with this project OR overdriving LEDs would provide some advice it would be greatly appreciated!

If it were my project I'd provided "ballast resistors"
for each LED, and drive the string from a low impedance
source. Better yet, reconfigure the LEDs in a series
chain so they share the same current and use a higher
voltage (which is a bit more work, but requires only
one ballast resistor per chain).

EDIT: DC = Direct Current, not Duty-Cycle that may
be used by others. Yeah, it's confusing...

I recently received the Heat Management DVD from Optek (the LED people).

Ignoring all the sale pitch stuff they ahd some interesting things to say about LEDs.

1. They are only 8% efficient once the heat and photonic losses are taken into account. This refers to high power white LED's

2. The most intersting thing was the light output specs. When manufacturers quote Lumens or Candelas the reference point is the output detected within the first few milliseconds .

I'll have to review the DVD again, to get the exact figure.

The DVD is free by the way, but it is really more of a sales pitch to buy Optek's LED's - with some technical info thrown in to stop a techy beign bored.

Colin

Hi,


I always use average current to get the operating light output of an
LED, even though it's less efficient at higher current.
This is easy to calculate with pulse power too:
Iavg=Ipk*DC
where
DC is the duty cycle (0.25 is 25 percent 'on' time).

For many small LEDs the average current is 20ma, but they
do work up to 30ma without too much trouble (room temperature)
but pushing them up to 50ma starts to heat them up too much
and they start blinking on and off and then die.
The max peak current is typically 100ma (5 times as high as average I)
for these small 5mm LEDs, but you have to check the data sheet.

For larger LEDs like the 1 watt Luxeon Star, the average current is 350ma
but the peak is only 500ma, so you have to be more careful about
the pulse amplitude even if the duration is short.

Hey, what exactly is flawed with the circuit bigglez? I'd guess paralleled LEDs of the same specification would be ok, no?

GregM WI NY wrote:Hey, what exactly is flawed with the circuit bigglez? I'd guess paralleled LEDs of the same specification would be ok, no?

LEDs are current operated devices. Putting them in parallel
does not share the current equally unless the LEDs are matched
(for forward voltage drop).
A limited matching comes from using the same wafer for
the die, but even high volume vendors grade the LED parts
by light output matching (i.e. for multi-segment displays).

A much better design is to drive the LEDs from constant
voltage and add a ballast resistor per LED.

Placing the LEDs in series means they do share the
same current and only one ballast resistor is needed
per chain of two or more LEDs.

> This arrangement is flawed.

Maybe... if you were sending a UV LED array to Mars to image certain rocks against a six order of magnitude scale, but for this project the scheme is just fine.

I find this constant imperialist vibe toward "pure" electronics a true waste of time. Too much energy has been wasted on unnecessary precision and I won't have it, Mr. Bigge Lez.

Dan

I totally disagree. There is a very good reason for what big said. What part of "had a problem getting LEDs (in the fountain kit) to be equal brightness." don't you understand?

Your example is actually completely backwards. The mars mission - they can afford to match the LEDs. (though I doubt they would do sloppy engineering.) Hobbyists buy what has a reasonable cost. What's the harm in using individual ballast/limiting resistors??? It has nothing to do with some misguided need to lord over the hobbyists. If you want to use sloppy design, do so but guys like us "imperialists" will be quick to correct it when you tell others it's ok.

Maybe... if you were sending a UV LED array to Mars to image certain rocks against a six order of magnitude scale, but for this project the scheme is just fine.

Dan is the man!

I'm with Bigglez and Philba. Its just constructive criticism. There are multiple ways to do just about anything. Naturally there are ways to bias LEDs and there are other ways. Some have advantages, one was being pointed out.

many people may build this project using parts from their parts collection and from assortment packs rather than buying all new LEDs from the same source (in hopes they will be somewhat matched). In such a scenario, some LEDS can be overdriven while others may be barely lit. Adequate for educational purposes for those learning about construction and that circuit. Many readers are not novice and will benefit from discussion about why resistors can improve such a design

One thing you learn when you start building projects is that what looks adequate on a schematic or in simulation may have problems in actual construction. For example due to issues such as the matching of Vf and Rf on LEDs from actual stock on hand. If you own a curve tracer, matching parts is easy but few people (not including me) do not own a curve tracer

BTW. If I come off as imperialistic (expanding ones territory through conquest), I prefer Pontification (What I say is the final word and that's the end of discussion) ha ha

"If you own a curve tracer, matching parts is easy but few people (not including me) do not own a curve tracer "


If few people do not own curve tracers, does that imply that most peole do own curve tracers?

If few people do not own curve tracers, does that imply that most people do own curve tracers? Very Happy

Oops, I doubled up on too many negatives

"Two wrongs do not make a right but three lefts do"