Designing around Vf?

[MrNaz]

I read in another thread from a modder that because Vf is variable from emitter to emitter and that in a multi-emitter flashlight one could not guarantee consistent Vf, it is best to design your circuits in such a way that Vf is irrelevant.

This sounds to me like a good idea, but how do you do it? I was under the impression that Vf is the voltage required to make an LED emit, so if you supply much more then then LED is at risk, muchless and it won't light.

Is this correct or is my understanding flawed?

 

[Gunner12]

From my understanding Vf is the forward voltage of the LED at what ever current, so a LED might have a Vf of 3.4v at 350 mA.

Also I read that wiring all the LEDs in series and using the right driver would remove the Vf problem. The LEDs all receive the same current.

 

[georges80]

I read in another thread from a modder that because Vf is variable from emitter to emitter and that in a multi-emitter flashlight one could not guarantee consistent Vf, it is best to design your circuits in such a way that Vf is irrelevant.

This sounds to me like a good idea, but how do you do it? I was under the impression that Vf is the voltage required to make an LED emit, so if you supply much more then then LED is at risk, muchless and it won't light.

Is this correct or is my understanding flawed?

LEDs should be driven with a current regulated driver. i.e. you maintain a constant current through the LED and the Vf becomes a non-issue.

Think of Vf versus current as a curve that increases nearly exponentially at the right side. At higher current the Vf is higher than at lower current. The Vf is also temperature dependent, as the LED warms up the Vf (at a given current) would become lower. Vf also varies from LED to LED (variations in the process, wafer etc).

Given the variables above (especially the temperature dependence and device to device variation) using constant current ensures the LED is driven to spec regardless of how the Vf varies.

The Vf versus current curve is VERY steep at the higher end (as stated above), so even a small change in applied voltage can make a very large change in current, yet another reason to use a current regulator versus a voltage regulator.

You can see these curves if you look up a datasheet for a LED.

cheers,
george.

 

[asdalton]

The Vf from a spec sheet is a nominal parameter that represents the voltage at a specified current, often 350 mA for power LEDs.

In general, Vf is the forward voltage across an LED that varies with current.

If you use a current controller to vary the current to a LED, you will find that the measured Vf across the LED varies a few tenths of a volt at most, even if the current is varied from 50-1000 mA.

Conversely, if you try to regulate the voltage, you will find that the current through the LED is very sensitive to changes in applied voltage. When you add this sensitivity to the known part-to-part variations, you can see why Vf is not a parameter to design around.

Strictly speaking, you often can use Vf in a circuit design, but only if the circuit allows modest variations in the specific Vf of the LED without causing huge variations in current. The simple resistor regulation scheme is an example of this.

Let's say that you have the following:

Voltage source: Vs = 6.0 V
Intended LED current: I = 350 mA
Approximate Vf = 3.4 V

Then the resistor value should be:
R = (Vs - Vf)/I
R = (6.0 - 3.4)/0.350 = 7.4 ohms

(Power dissipation check: P = I^2*R
P = 0.350^2*7.4 = 0.91 W
So the resistor should be rated at at least 1 watt.)

Now, while current for the bare LED is a steep function of Vf, in the resistor protected circuit,

I = (Vs - Vf)/R,

and so the current depends linearly on the quantity (Vs - Vf) which is a much more predictable (and weaker) function of Vf that you would get by applying a voltage directly to the LED alone.

 

[MrNaz]

So let me see if I have this right:

Using a buck driver to drive a set of 3 series LEDs will ensure that they all get the same current, and that even if Vf varies over time (say as the lamp heats up), the buck driver will alter the voltage so that the same current flows through it?

To rephrase that question; Does a buck driver alter the voltage it passes through to ensure the right current, or does it pass through the same voltage as the source voltage, but limit the current it allows to pass?

 

[Greg G]

Vf is important when choosing the battery pack, or if that is fixed, the driver, buck or boost.

 

[georges80]

So let me see if I have this right:

Using a buck driver to drive a set of 3 series LEDs will ensure that they all get the same current, and that even if Vf varies over time (say as the lamp heats up), the buck driver will alter the voltage so that the same current flows through it?

To rephrase that question; Does a buck driver alter the voltage it passes through to ensure the right current, or does it pass through the same voltage as the source voltage, but limit the current it allows to pass?

The former, i.e. yes, the Current Regulated buck driver will step voltage down to the load while maintaining a constant current through the load. Of course this presumes that the total Vf is lower than the Input voltage to the buck driver (so that it can step down).

There is also the 'headroom' that the buck driver needs (how much greater the Input voltage needs to be above the total Vf) to current regulate, this depends on the specific driver and can be current dependent - typically from 0.5 to 1.5V for most of the drivers you are likely to come across for LEDs.

cheers,
george.

 

[MrNaz]

The former, i.e. yes, the Current Regulated buck driver will step voltage down to the load while maintaining a constant current through the load. Of course this presumes that the total Vf is lower than the Input voltage to the buck driver (so that it can step down).

OK, and you get buck drivers in a variety of *currents*, and the buck driver will (in theory) ensure that that current is delivered, and adjust the voltage accordingly.

So, forex, this driver:
http://www.dealextreme.com/details.dx/sku.1886
Will adjust the voltage across its terminals to deliver 1.4A regardless of the load, so long as the Vf(total) is between 3.6V and 4.5V less the overhead. Correct?

On another note, I am planning on using that to overdrive a Cree Q5 in a Romisen RC-N3 to produce a short battery life but super bright small torch. Do I correctly understand what I'm doing or am I missing something fundamental? Will the Cree take 1.4A or am I better off using a 1A driver?

 

[georges80]

Those DX drivers are NOT buck converters. They are just linear regulators that will dissipate extra voltage x current as heat. The higher the input voltage versus output Vf the more heat will be dissipated.

1.4A through a Cree is quite a bit beyond spec, heatsinking is critical to keeping the junction (inside the LED) cool - or at least not boiling...

cheers,
george.

 

[MrNaz]

Those DX drivers are NOT buck converters. They are just linear regulators that will dissipate extra voltage x current as heat. The higher the input voltage versus output Vf the more heat will be dissipated.

Uh oh...
So if, as I planned to do, I put this one in a torch:
http://www.dealextreme.com/details.dx/sku.3160
I would not end up saving battery life, as it would still draw full power from the battery, only it would only deliver 350mA to the LED and the rest would be bled off as heat? So there's no battery performance benefit in using that one over this one:
http://www.dealextreme.com/details.dx/sku.1885 ?

1.4A through a Cree is quite a bit beyond spec, heatsinking is critical to keeping the junction (inside the LED) cool - or at least not boiling...

I'm fairly familiar with heat sinking. The Romisen has a decent mating to the outside housing, so running 1.4A through the Cree is likely to make the torch itself hot. It's mainly as an experiment to just make the brightest single 123A torch I can, a torch with a 30 minute or less run time is not of much practical value.

 

[Gunner12]

Those are fine for voltages around 3.7-4.5 but. The voltage drop isn't that much so not that much power will be lost. It would draw the stated current from the batty and bled off the extra voltage as heat, hence the 4.5v top to keep the heat reasonable.

If you want to drive 3 LEDs off for example an 18v source, wire the LEDs in series, ---(LED)---(LED)---(LED)---, with a buck driver and each LED should get the same current.

Are you going to be using a RCR123 or a primary CR123? If you are using a RCR123, look at the IMR(is that the right term?) RCR123s and make a light that direct drives a Seoul P7 or Cree MC-E off the IMR battery. An IMR battery can take much highr currents then a normal Li-ion battery so it can be used to direct drive a quad die LED even with a smaller capacity battery.