Circuit for reducing LED current

Thread Starter

kuera

Joined Aug 17, 2012
39
Good evening..
I'm not sure if the title is misleading or not but I'm not sure exactly how to explain in the title what I'm looking for.
I was thinking about your typical "Joule Thief". I've used them before but was wondering about something. Considering the LED is actually blinking at a very high frequency does it actually run using less current in the long run compared to the same LED running on a direct circuit?
My whole thought comes from the fact that I was messing around with building little garden lamps using 1watt LEDs and a battery pack (they're for carrying around) and I wanted to use the idea that some LED drivers reduce power consumption by blinking the LED. Now the only drivers I've found are for a higher power input but I just want to run the LED off a 3.7V power supply (same as the LED rating) which is why I was wondering if the Joule Thief idea was viable.
Can anyone recommend an idea for what I'm trying to get to?
So it's basically just a Li-ion power supply (3.7V) -> Driver (or whatever you want to call it) -> 1Watt LED.

I'm not too worried if the driver does affect the brightness of the LED too much as I'm expecting it, just as long as it's not by a stupidly high amount. there are also chargers and such involved but that I'll sort out.

Many thanks in advance.
Gordon
 

Thread Starter

kuera

Joined Aug 17, 2012
39
You need a constant current led driver, that will give out the required level for your led.
From what I can tell these drivers don't really try and conserve power as much as just supply what the LED needs. I'm more trying to find a way to run the LED at a very high frequency to try reduce power consumption without dimming it too much.
 

AlbertHall

Joined Jun 4, 2014
12,345
Whatever frequency you drive it with will not reduce power consumption for the same LED brightness. Reducing the average LED current will reduce brightness and power consumption.
 

ian field

Joined Oct 27, 2012
6,536
You need a constant current led driver, that will give out the required level for your led.
Not on a Joule thief - they invariably deliver narrow pulses that can by definition only put a small amount of energy into the LED. Its a form of pulse drive that permits the pulse current to be somewhat higher than the LED's rated continuous current.

Its not impossible to blow a LED with a Joule thief, but there's a simple version that's popular with beginners (and lazy constructors).

With the blocking oscillator style joule thief; its easy to put a current sensing resistor in series with the LED - the developed voltage switches on a transistor that shorts out the base drive to the "chopper" transistor.

Obviously the current pulse has to have already happened to sense it - but the net effect of the shunt transistor is to reduce the duty cycle, you can at least adjust it to keep the LED within its power rating.
 

Thread Starter

kuera

Joined Aug 17, 2012
39
Since looking around I think what I want can be more simply explained.
I'm just looking for a way to pulse a 3.7V x 1Watt LED by something like 50 mhz with basic components (no ICs, mainly just resistors, caps and transistors) so that the LED looks like its on in a solid state but the blinking is actually reducing power consumption. It's a method used by some bike lights or head lights so that instead of a battery lasting only something like 2 - 3 hours it last way longer.
Any ideas?

PS. The LED I'm using is just your average super bright 3.7V 1 Watt
The power supply is a Li-Ion 3.7v battery.
 
Last edited:

AlbertHall

Joined Jun 4, 2014
12,345
'Flashing' the LED at 50MHz will not gain you anything except splattering radio interference all over the place. Usually they are flashed more slowly - visibly flashing - so when they on they are bright but current is reduced because they not lit all the time. This has the additional benefit that flashing lights are much more noticeable than steady lights. Circuits like this are used:
 

Thread Starter

kuera

Joined Aug 17, 2012
39
'Flashing' the LED at 50MHz will not gain you anything except splattering radio interference all over the place. Usually they are flashed more slowly - visibly flashing - so when they on they are bright but current is reduced because they not lit all the time. This has the additional benefit that flashing lights are much more noticeable than steady lights. Circuits like this are used:
The idea is not to be noticed. Its to be used in lanterns so im not worried about inteference at all. Im actually going for the idea so you dont notice the flashing.
 

RichardO

Joined May 4, 2013
2,270
I am with AlbertHall on this one. I doubt very much that any of the circuits you mention flash the LED at over 1 KHz. Even that is much higher than is needed to keep the human eye from seeing flicker. Even 50 Hz (not 50 MHz) is high enough to not cause noticeable flicker.
 

Ylli

Joined Nov 13, 2015
1,086
The efficiency of an LED is higher at higher currents. For example, if you provide a common LED with 20 mA, the light output may be 1000mcd. If you provide 200 mA pulses at a 10% duty cycle, the LED gets the same average current, but may put out 1200 mcd. Typically, you don't need switching speeds more than 120 hz or so.
 

Thread Starter

kuera

Joined Aug 17, 2012
39
I am with AlbertHall on this one. I doubt very much that any of the circuits you mention flash the LED at over 1 KHz. Even that is much higher than is needed to keep the human eye from seeing flicker. Even 50 Hz (not 50 MHz) is high enough to not cause noticeable flicker.
Oooh, ever closer to the answer heh. Would anyone be able to supply a schematic for a circuit that would allow me to make the led I want to use to flicker at lets say a 100 Hz pulse.
The led is one of those hex shape 3.7V x 1 Watt and the supply is also 3.7V.
I just want to make it as simple as possible.
Much thanks again in advance.
 

ian field

Joined Oct 27, 2012
6,536
The idea is not to be noticed. Its to be used in lanterns so im not worried about inteference at all. Im actually going for the idea so you dont notice the flashing.
The circuit posted by alberthall produces very narrow spike pulses - the ones in the old roadworks flashing beacons had a resistor in series with the timing cap to stretch the pulse width. The old ones used bulbs and the pulses would've been too fast for the filament thermal inertia.

22uF looks too big for continuous light - you might have to go as low as 0.22uF or even lower. The capacitor and its charging resistor determine the pulse repetition rate - they have much less effect on pulse duration. So you set the power to the LED by that capacitor and resistor. Making the resistor too low can latch it up, making the capacitor smaller increases the repetition rate that conspires to become a greater PWM ratio.
 

dannyf

Joined Sep 13, 2015
2,197
Can anyone recommend an idea for what I'm trying to get to?
it is not clear what you are looking for. Two possibilities:

1) an indicator of the led is on: obviously, lighting up the led at very low duty cycle (=blinking) saves energy than lighting up the led at 100% duty cycle.
2) blinking the led at very high frequency but low duty cycle can reduce current consumption while maintaining equal perception of brightness, relying on the eyes' persistence of vision.

1) can be made to consume very little current. 2) can be made to significantly reduce current consumption, vs. a constantly lit led.
 

ian field

Joined Oct 27, 2012
6,536
it is not clear what you are looking for. Two possibilities:

.
The impression I got was; The TS wants to make a lantern with regulated LED current - presumably, without wasting power in a resistor.

The circuit posted by Alberthall is one way of doing that - but with the large timing capacitor, the result would be more like a flashing beacon.

The CR values can be juggled with to effect pulsed power control - you can increase the PRF by reducing R in the CR, which increases the pulse to dwell proportion. Or you can get a limited pulse width adjustment range by putting a pot in series with the capacitor.
 

Thread Starter

kuera

Joined Aug 17, 2012
39
This is pretty damn close to what I'm looking for. or well from the description it seems so.
Thanks for the link, if nothing else I've been reading through it just because it's informative =)
From what I can tell that circuit can be used to amplify the current through the LED as well as pulse it? Not sure if I'm understanding it correctly but that's what I can tell from what he says about it.
 
Top