Understanding forward current

I bought some LEDs on eBay (see graph at the bottom for specs, particularly the white ones which are the ones that I performed this test with).

5V --- LED --- 200? resistor --- GND

When I measured the current, I got: 8mA. According to Ohm's law this is correct: (5V - 3.4V) / 200? = 8mA.
I then changed my resistor with a lower value; 150? and I measured again. This time I got 12mA. According to Omh's law I should have gotten 10mA (close enough: I'm assuming my resistor wasn't exactly 150?). So far, so good...

Now, in the datasheet I can see this a bunch of times "If = 20mA". What exactly does this value mean and when should I take it into account?

The reason I'm asking is that I need to stick a bunch of these LEDs behind a BC547 NPN Transistor (datasheet) and as I understand it, I can only "run" up to 100mA through this transistor. I need to calculate my LEDs' circuit current so that I don't exceed this and damage the component.

Thanks!

Ohm's law is a theoretical, linear relationship that doesn't work for diodes of any sort.

The voltage across any solid state diode (LED or rectifier) is a logarithmic function of the current, or to put it the other way, the current through the diode is an exponential function of the forward voltage. The values of the typical forward voltages reported in the table you posted are given for 20 mA diode current.

You need a current limiter when working with diodes, because applying a voltage even slightly greater than the typical forward voltage can lead to catastrophic current flow.

ardilla:
Now, in the datasheet I can see this a bunch of times "If = 20mA". What exactly does this value mean and when should I take it into account?

It means that the forward voltage readings listed are what you can expect when you put 20mA through the LED. You can take it as a hint that 20mA might be a good forward current to design for. The maximum forward current should be listed on the datasheet.

ardilla:
The reason I'm asking is that I need to stick a bunch of these LEDs behind a BC547 NPN Transistor (datasheet) and as I understand it, I can only "run" up to 100mA through this transistor. I need to calculate my LEDs' circuit current so that I don't exceed this and damage the component.

A BC337 transistor would be a better choice, it is rated at 800mA, and can reasonably switch up to about 400mA if you give it 20mA base drive from an Arduino pin and series resistor.

LEDs work by controlling the current, not the voltage.

The reason is their resistance varies wildly as you approach the sweet spot (usually 20mA). Trying to pick a resistor isn't easy. You can get big variance in brightness between LEDs.

This is why you see special chips just to control LEDs.

I need to stick a bunch of these LEDs behind a BC547 NPN Transistor

For switching large numbers of LEDs I like to use an6884 chips (or ka2284 which work identically). Yes, they're "VU meter" chips but if you put the input pin HIGH all the LEDs switch on.

Think of them as a BC547 with five input legs, each with proper current regulation (no resistor needed!) Just connect the Arduino output pin to pin 7 of the chip. You can connect dozens of them to a single Arduino pin.

fungus:
The reason is their resistance varies wildly as you approach the sweet spot (usually 20mA). Trying to pick a resistor isn't easy. You can get big variance in brightness between LEDs.

This isn't the real story - if the forward voltage of the LED is a lot less than the supply,
choosing a resistor is very easy. For instance a single red LED from 5V supply. Any
brightness variation will be due to differences in LED efficiencies, not currents in
this situation.

If the forward voltage is only a little less than the supply then there are problems -
the forward voltage is not constant, it varies with temperature and possibly the age
of the device and different LEDs of the same type have slightly different forward
voltages for the same current "manufacturing spread". You have to work out the
worst case to determine the resistor value that will safely work with all LEDs of that
type across the temperature range anticipated - and there may then be a spread of
brightnesses (because of the spread of currents).

The example is a blue or white LED from 3.3V supply (such LEDs have typical
forward voltages of 3.0 to 3.2V, and higher power ones go upto 4V).

MarkT:
This isn't the real story.

It was the short version

fungus:
LEDs work by controlling the current, not the voltage.

Givez fungus a 150a/h deep cycle 1.2v battery.

You have huge amounts of current, you could even place the led across the terminals... now if current controlled the led why is it not on?

Voltage requirements need to be met for the diode to begin to conduct right? So it can't be purely current controlled.

I suspect that English is not fungus' native language, and that he meant "LEDs work best when the current is controlled, not the voltage".

In any case, diodes conduct current at every nonzero voltage, forward or reverse. There is no discontinuity in the applicable voltage/current relationship in either direction. So, if you connect an LED to a 1.2 V battery, a small amount of current will indeed flow.

Using the following Spice model for a red LED connected to a 1.2 V battery, I calculate the diode current to be about 24 microamps.

*Typ RED GaAs LED: Vf=1.7V Vr=4V If=40mA trr=3uS
.MODEL LED1 D (IS=93.2P RS=42M N=3.73 BV=4 IBV=10U

• CJO=2.97P VJ=.75 M=.333 TT=4.32U)

Um

Sure... but I bet the battery will self discharge quicker lol

jremington:
I suspect that English is not fungus' native language, and that he meant "LEDs work best when the current is controlled, not the voltage".

I hardly ever speak English in real life.

jremington:
In any case, diodes conduct current at every nonzero voltage, forward or reverse. There is no discontinuity in the applicable voltage/current relationship in either direction.

No, but there's a point on the curve where it suddenly goes exponential and it's usually very close to the "forward voltage" in the datasheet.

An important point about LEDs - they don't behave like an ideal diode at low currents
when light falls on them - they are photodiodes. Only in the dark will the exponential
carry down to ultra low currents, and of course no diode is really ideal so there are
other leakage current mecahnisms.

Any sufficiently small leakage current can be called "zero" for practical purposes!

No, but there's a point on the curve where it suddenly goes exponential and it's usually very close to the "forward voltage" in the datasheet.

That is neither correct, nor helpful to others in understanding diode behavior. The normal exponential diode equation quite accurately models LED behavior at voltages significantly less than the data sheet "forward voltage" and for several orders of magnitude in current. You can verify this for yourself with a couple of multimeters, an LED and a regulated power supply that can be accurately adjusted in the range of 1 - 2 volts. But see MarkT's post above.

MarkT's point about LEDs being a photodiode (in fact any PN junction is potentially a photodiode) leads to a couple of interesting points: the photoinduced current is in the opposite direction of the normal forward current. Also, LEDs are reasonably efficient photodiodes only in a narrow wavelength range, blue shifted from the emission maximum, so they can be used as color-sensitive detectors. The following example plot was taken from one of Forrest Mims research publications. It is for IR LEDs but the same behavior is true of all "single-color" LEDs .

Some useful facts about LEDs and photodiodes can be found here: www.physics.unlv.edu/~bill/PHYS483/LED_PIN.pdf?