# BASIC LED DRIVER FOR 10 WATT CREE LED

#### teliocide

Joined Sep 26, 2013
89
I need a LED Driver circuit to power a 10 Watt Cree LED
The Led will be used for outside lighting so slight fluctuations in brightness can be tolerated.
As I have to make a number of these cost comes into the equation.
I intend to run the LED at a lower current than the rated max 3000mA to give them some longevity.
I am yet to determine the "sweet spot".... brightness vs current so I dont know exactly the running current but anticipate around 2000-2500 mA
The Supply voltage will be 12 volt from SLA batteries

LED SPECS:
Specifications
Forward voltage: 2.4-2.8V
Cont Forward current: 3000mA max
Viewing angle: 140°

Thanks Greg

#### LowQCab

Joined Nov 6, 2012
1,096
How many LEDs per SLA-Battery ?

#### DickCappels

Joined Aug 21, 2008
7,703
The “sweet spot” in the illuminance vs current curve will depend on how good your heat sink is.

#### teliocide

Joined Sep 26, 2013
89
The “sweet spot” in the illuminance vs current curve will depend on how good your heat sink is.
Very good actually - All from PC CPUs - I never throw them away

#### teliocide

Joined Sep 26, 2013
89
How many LEDs per SLA-Battery ?
It will depend on which SLA I use
Probably only one LED on the 20AH SLA and if possible 5 on the 120 AH SLA

#### LowQCab

Joined Nov 6, 2012
1,096
Battery Voltage should be selected based on how much Voltage You have to
get rid of in the form of HEAT.
.
.
.

#### Ian0

Joined Aug 7, 2020
3,232
Diodes’ AL8843 is probably the lowest component count solution.
IC, schottky diode, inductor, sense resistor and decoupling cap.

(by the way, there is no “sweet spot” - the efficiency falls as the current increases throughout the entire operating range)

#### Bordodynov

Joined May 20, 2015
2,906
See

#### teliocide

Joined Sep 26, 2013
89
Diodes’ AL8843 is probably the lowest component count solution.
IC, schottky diode, inductor, sense resistor and decoupling cap.

(by the way, there is no “sweet spot” - the efficiency falls as the current increases throughout the entire operating range)
Actually there is a sweet spot ..... it is not quantifiable in your language
Brightness is a human perception and is very non linear
There is a point where the brightness is more than acceptable. At this point the trade of begins ie a barely perceivable increase in brightness for a significant increase in current.

#### DickCappels

Joined Aug 21, 2008
7,703
The perceived brightness is relative to the illumination in the environment. If you run with insufficient heatsinking the LED can become too hot and its efficiency decreases.

Increasing the current and therefore the temperature can result in a dimming of the LED.

#### Bordodynov

Joined May 20, 2015
2,906
See

Last edited:

#### teliocide

Joined Sep 26, 2013
89
Battery Voltage should be selected based on how much Voltage You have to
get rid of in the form of HEAT.
.
.
.
View attachment 244512
Thanks for the circuit
I realize its more efficient to operate LEDs with a power supply voltage as close as possible to the LED operating voltage.
It always puzzled me (and annoyed me) that to operate LEDs and things such as Voltage regulators "excess" electricity has to be dumped as heat.
I vaguely remember making my first voltage regulator in the early 70's.
It was to allow the battery to be removed (permanently) from my Suzuki TS 250
It was just a 2N3055, a resistor and a zener and a bloody great chunk of aluminium

#### crutschow

Joined Mar 14, 2008
27,714
I would suggest a constant-current switching regulator for best efficiency and battery life as Bordodynov showed.

With this the LEDs will also not change brightness as the battery voltage drops.

#### crutschow

Joined Mar 14, 2008
27,714
If you want to roll your own switching regulator with common ICs, below is the LTspice simulation of a simple hysteretic type (bang-bang) constant-current switching regulator that uses a LM339/393 comparator as the controller.
The hysteretic control has the advantages of being very simple and inherently stable, thus requiring no feedback compensation (and thus has no startup overshoot).

The six CD4050 buffers (one IC package) are used in parallel as a poor-man's MOSFET gate driver.
The TL431 provides a stable 2.5V reference voltage.

The efficiency should be in the neighborhood of 80-90%, much higher than a linear regulator or series resistor, thus greatly reducing heat-sink requirements.

There is some ripple in the LED current, but that is not noticeable to the eye at that high a switching frequency.

Edit: Cautionary Note -- The circuit has no short-circuit protection so a prolonged short from the L1 output to ground will likely blow the MOSFET.

Last edited:

#### Bordodynov

Joined May 20, 2015
2,906
Hello crutschow.
Could you calculate the energy efficiency of your electronic circuit. Or upload your electronic circuit in LTspice format.

#### teliocide

Joined Sep 26, 2013
89
Very nice Mr Crutschow

#### Ian0

Joined Aug 7, 2020
3,232
The six CD4050 buffers (one IC package) are used in parallel as a poor-man's MOSFET gate driver.
RS prices:
CD4050 £0.361
MCP1402 gate driver: £0.358

#### crutschow

Joined Mar 14, 2008
27,714
Hello crutschow.
Could you calculate the energy efficiency of your electronic circuit. Or upload your electronic circuit in LTspice format.
Based upon a rough calculation I came up with about 90%, but the actual will likely be less.

LTspice file below.

#### Attachments

• 4.4 KB Views: 5

#### crutschow

Joined Mar 14, 2008
27,714
RS prices:
CD4050 £0.361
MCP1402 gate driver: £0.358
I stand corrected.

#### Ian0

Joined Aug 7, 2020
3,232
I stand corrected.
Odd isn't - we still think that CMOS is as cheap-as-chips and MOSFET drivers as being exotic. But CMOS is getting expensive as it is going slowly out of fashion, and everyone's using MOSFET drivers these days.
Some of them have hysteresis on the inputs with relatively accurate thresholds and with a bit of ingenuity you can implement most of the usual 555 circuits with them, but with massive amounts of output drive.