What's the PROPER way to drive LED's?

[schmellyfart]

Im going to be making some LED arrays in the near future and I've read a lot about "ben is wrong about the led mod. Dont use a 10ohm resistor. Thats not the right way. blah blah blah" So, I want to know the PROPER way to drive LEDs.

[Mario]

http://www.ledcalc.com/" onclick="window.open(this.href);return false;

[hailrazer]

meh everyone has different opinions it seems.

Some say it is fine for a single resistor and then series the leds. Others say a resistor for ever led and then parallel them. Others say it makes no difference either way.

Guess it might not matter.

[schmellyfart]

yeah thats why I was asking.

For kicks and giggles, I checked to see what resistor should be used for a ps1 screen LED mod with 3 LEDs. It said 10 ohms...

It appears to that the proper way to drive LEDs is rather relative.

[nevermind1534]

yeah thats why I was asking.

For kicks and giggles, I checked to see what resistor should be used for a ps1 screen LED mod with 3 LEDs. It said 10 ohms...

It appears to that the proper way to drive LEDs is rather relative.

But is that in series? I think ben suggest the 10 ohm in parallel, but I could be wrong.

[nitro2k01]

Ok n00bs, here's the theory. A diode differs slightly from a resistor in how it works. A resistor has a linear voltage->current curve. Double the voltage to double the current. Triple the voltage to triple the current. A diode however has an exponential voltage -> current curve. You need to vary the voltage slightly to double or triple the current, causing the the diode to break faster. If done carelessly, within seconds.
There's a useful range fo current values which is approximated by the forward voltage drop. When the current is within a useful range, the voltage drop over the diode will be around this value.

So, what you need to do is to control the current that goes through the diode to prevent it from dying from overcurrent. The usual way of doing this is by adding a current limiting resistor. (To understand any of this, it's important to know what voltage and current actually are)
Connecting a diode like this, with no limiting will make the current exponential with respect to the input voltage, ie, very high.


However, when connecting a resistor in series, like this, you can define a current that you want to go through the diode. (The current decides how bright the diode will shine) Let's say we're using a red LED, (2.0 V voltage drop) that's rated at 30 mA for the sake of this discussion.


This allows you to calculate a suitable value for the resistor using Ohm's law. First a word about the voltage and current values in the picture. The current will be the same through the resistor and the LED since no current can escape anywhere between +5V and ground.
As for the voltages... 2.0 V is the voltage drop value for the LED. The voltage drop for the resistor must be 5 V (The supply voltage in this example) - 2 V (The LED's voltage drop)
The version of Ohm's law suitable for this calculation is R=U/I:
3 V / 0.030 A = 100 Ohms

In this case however, you could also put two LED's in series.

The current is the same, however the voltage drop over the resistor is different. The total voltage drop over both the LEDs is 4 V, so the voltage drop over the resistor will be 1 V this time to add up to the 5 V supply voltage. Applying Ohm's law again you get:
1 V / 0.030 A = 33.3 Ohms

You can put LEDs in series in this way, so long the sum of the LEDs' voltage drops will leave some room for a resistor. For example, two blue diodes with a voltage drop of 3.3 V each will add up to 6.6 V, which is way above 5 V. These LEDs won't light up if connected in series even without a resistor since the voltage drop (Which really works like a threshold voltage) won't allow any current through the diodes.

Hope of all this helped you guys instead of confuse you even more.

[schmellyfart]

Ah, thanks Nitro!

Nevermind, its parallel.

[marshallh]

You one resistor in series for each led, with the sets arranged in parallel. Otherwise you'll overdrive LEDs due to manufacturing inconsistencies.

I have built two portables that have developed this problem due to following Ben's guide. Don't. Use a 33-ohm in series with each led, all wired in parallel.

[nitro2k01]

You one resistor in series for each led, with the sets arranged in parallel. Otherwise you'll overdrive LEDs due to manufacturing inconsistencies.

I have built two portables that have developed this problem due to following Ben's guide. Don't. Use a 33-ohm in series with each led, all wired in parallel.

Which values, voltage and LED colour?
There are two potential pitfalls.
You always need a resistor, and if the resistor is too small, manufacturer inconsistencies may indeed become noticeable. This is especially true if you drive the LEDs directly off of a non-regulated voltage source, such as a battery. Using a nonregulated source together with a small resistor will make the current limiting effect less effective. In that case, it's better to wire them in parallell. And each parallell LED need its own resistor, or you'll risk that one of the LEDs won't light up because the other LED steals all the current, again due to manufacturer inconsistencies.

[eurddrue]

You always need a resistor

Where's rekarp now!? Thank you nitro, we had an argument about that. And, I use 10 ohm resistors in front of all my LEDs and they've lasted almost a year now.

[Rekarp]

You always need a resistor.

<-- btw love that smiley

Manufacture inconsistencies for LEDs fall in the very small range. Hell slight temperature differences are going to affect your resisters (and the +-5% resistor value) are going to be way more inconsistent then the LEDs specifications. If your supply is a steady (regulated source) and is equal to the forward voltage then a resistor is not needed.

The always need a resistor is one of those old timey sayings that doesn't really hold true now days.

Here is an experiment for yall that doubt this. Wire a voltage regulator to be the forward voltage of the LED. Wire the LED in series with the regulator. Now turn it on. Did it explode? Check the voltage drop across the LED. Now check the current running through it. Zomg Physics works? No way!

I do agree that running LEDs in parallel is better then series. As you sometimes get a bad LED and it could short out causing all your other LEDs to go pop. Manufacture inconsistencies in specs is not the real reason to chose parallel over series its bad LEDs ( I guess thats kind of a Manufacture inconsistencies)

[nitro2k01]

You always need a resistor.

<-- btw love that smiley

Ok, you're right. There are instances when you do not need a resistor. It was a simplified way of saying "Do not connect LEDs directly to a voltage source". I still withhold that it is good practice to use some sort of current limiting circuit, however it is designed.

Manufacture inconsistencies for LEDs fall in the very small range. Hell slight temperature differences are going to affect your resisters (and the +-5% resistor value) are going to be way more inconsistent then the LEDs specifications. If your supply is a steady (regulated source) and is equal to the forward voltage then a resistor is not needed.

The difference though is that a LED has an exponential V->I curve whereas resistor has linear curve. (And thus works as a current limiter) If you use a resistor which value is 10% off, (Btw, can you even get 10% tolerance resistors anymore?) the (linear) current limiting effect will still be enough to protect the LED. If you're using a regulated voltage supply and the LED's voltage drop is, say, 100 mV off you might get a rampant increase in current, which might shorten the lifespan of the LED. (You can of course wire up any common linear regulator as a current limiter as well, but that's another story.) And the risk I described here is theoretical, but on the other hand, driving a LED with a voltage regulator is kind of a theoretical venture as well. There's no real point in it? The regulator is bulkier than a resistor and doesn't have any other advantages. The voltage drop over the regulator, if equal to that in an equivalent resistor configuration, would still burn just as much power as heat.

I do agree that running LEDs in parallel is better then series. As you sometimes get a bad LED and it could short out causing all your other LEDs to go pop. Manufacture inconsistencies in specs is not the real reason to chose parallel over series its bad LEDs ( I guess thats kind of a Manufacture inconsistencies)

The bad thing about running them in parallell is that you waste more energy as heat. Especially if your original source voltage is higher than 5 V... Let's say your source voltage is 7.2 V (Seems to be a popular battery voltage) In a series configuration that would be enough to drive 3 red LED's or 2 white ones. If you only connect one in each branch that additional energy is going to be wasted as heat in resistors and/or voltage regulators. Perhaps that doesn't matter much for a power LED, but it's a it more wasteful for a backlit screen running on a battery...

Of course, the really proper way of doing it these days is to use a dedicated PWM LED driver with an LC filter on the drive output. That way your LED will be safe, and a minimum of energy is wasted. (As a bonus you can adjust the brightness real easy too)