are some LEDs expected to burn in???

[Processaurus]

There's a calculator in your brain too!  It's just Ohm's law, V=IR, it's just that V is what's left over from the supply after subtracting the turn-on voltage of the LED (in other words, minus the voltage drop of the diode). Supply Voltage minus LED forward voltage= desired current times current limiting resistor.
V=IR

V_supply -V_led=IR

(V_supply -V_led)/I=R

Example:

5v (supply) minus 3v (blue LED turnon voltage) = .020 amps (20mA is a crankin bright LED) times the unkown resistor value.

(5v-3v)/.020A=R
100=R
R is 100 ohms

Boom.  Now you don't need the internet!

[darron]

hmmmm... well put out...

would you call the typical forward voltage the 'turn on' voltage also? in this case it's 3.3, which makes problems with the the 3.3v supply...


(3.3-3.3)/0.020A=0

so maybe your suspect about 3v is closer:

(3.3-3.0v)/0.020A = 15r


the datasheet suggests 20mA max current.... is it safe to run it pretty much on that? it doesn't suggest typical... or should i run closer to 15mA?

(3.3-3.0v)/0.015A = 20r


thanks for the LED theory guys! free education rocks... sorry if you paid for it....

[PRR]

> the 'turn on' voltage also? in this case it's 3.3, which makes problems with the the 3.3v supply...

Did you see what that old-LED-abuser R.G. wrote?

"...this works better when the voltage is at least 100% higher than the LED voltage, and best when it's several times the LED voltage."

If your LEDs are about 3.3V, then by R.G.'S Rule you want 6.6V to 10+V of power supply voltage.

If you have found a 3.3V supply "for LEDs", then it is made for high internal resistance LEDs. This gives lowest system cost but uncertain brightness and life.

The LEDs you have are specified for 2.8V to 3.8V at the maximum 20mA. Feed them a fixed 3.3V, some will be under 20mA, some will be over (and smoking).

And you can NOT safely run the LEDs in simple parallel. The 2.8V LED will hog the current and die, then the 2.9V one hogs and dies, and so on. You must either run Series (a good plan; and less wire) or use individual resistors.

9V supplies are common and  "at least 100% higher" than 2.8-3.8V so qualify under R.G.'s Rule.

The lowest-voltage LEDs, near 2.8V, leave 9.0V-2.8V= 6.2V across the resistor. 6.2V at 20mA is 6.2V/0.020A= 310 ohms minimum. Assume 5% tolerance resistors, you must order 325.5 ohms to be sure of getting over 310 ohms. 330 ohms is a standard value.

Now check the other extreme: a 3.8V LED and a 330+5% or 346.5 ohm resistor. 9.0V-3.8V= 5.2V, across 346.5 ohms, is 0.015 amps or 15mA. To the eye, 20mA and 15mA look pretty nearly The Same.

So: 9V supply with 330 ohm resistor in series with each LED.

[darron]

shame i made some strip boards already...

maybe i'll make a new sheet of them with strings of maybe 10x leds in series per row and use a higher supply with a resistor per string. one smd resistor per makes for a lot of work when i was spacing them 10mm apart, but i also don't like to string too many leds because the voltage tolerance for what's left to regulate starts becoming very tight.

would it be completely unsafe to run them in parallel with a lower voltage, say 3.0v? would that just be asking for a slow death in the weeks/months or one year to come? i've seen lots of crappy lights come from china where all the LEDs are different brightness and dying. i think you'll tell me not to do it... time to make some new boards i'm pretty sure....

[amptramp]

shame i made some strip boards already...

maybe i'll make a new sheet of them with strings of maybe 10x leds in series per row and use a higher supply with a resistor per string. one smd resistor per makes for a lot of work when i was spacing them 10mm apart, but i also don't like to string too many leds because the voltage tolerance for what's left to regulate starts becoming very tight.

would it be completely unsafe to run them in parallel with a lower voltage, say 3.0v? would that just be asking for a slow death in the weeks/months or one year to come? i've seen lots of crappy lights come from china where all the LEDs are different brightness and dying. i think you'll tell me not to do it... time to make some new boards i'm pretty sure....

Your assumption is correct.  We will tell you not to run LED's in parallel.  In fact, we already did.

[PRR]

For larger numbers of MAX-bright LEDs, you should use serious techniques.

[darron]

righto.. back to the computer then... sigh... good news to have direction though


i'll save that one PRR... i was also considering taking straight from the mains through a 220n cap, but nahh... DC is nicer

that would mean that i'd have to put the pnp and diodes off the board though away from the LED strip.

to keep things simple, though more soldering, i think i'll do   6V -    <1K limit resistor per LED (probably 220r)       then led.... this way i can string as many as i want without grouping them

[R.G.]

For power voltages less than 40V to start with, the LM317 makes a nice constant current source. It operates by forcing the voltage across the output and adjust terminal to be 1.25V nominal. If you ground the adjust terminal, it pulls a constant current in the input pin, and also is thermally limiting if it gets too hot. The 317L will do up to about 100ma, the TO220 version up to about 1A if they are not forced to operate too hot by the voltage across them.

Notice that this can also run on full wave rectified AC at power line frequency, as the LM317 is fast enough to follow the changes in a half-wave-rectified sine. From zero, nothing happens until the voltage rises enough to start the string conducting, then the 317 eats enough voltage to force the current to be constant. The only limitation in running it from line is that the voltage to the 317 must never exceed 40V. Some special provisions are needed to keep that from happening.