Brightest LEDs on the planet?

[Alex Tsekenis]

Daylight bulbs are amazing, a pure bliss to the eye in an electronics lab. The only issue is that the eye is not so sensitive to that light so maybe you need a few more tubes. If I have daylight tubes on and then change to warm white my eyes hurt. I think they are great, you should try them maybe on a bigger scale. I even convinced Intel Corp. to change the ones in my area.   

[Hero999]

Colour temperature is the equivalent to the light emitted by a black body at that temperature. The filament of a typical incandescent bulb is around about 2700K, a halogen filament is hotter about 3000K.

It's a matter of opinion, I hate daylight bulbs, they don't emit daylight but a horrible bluish grey coloured light, give me 2700K any day.

[Um...Me123]

I know this is getting a little off topic but while I have you all here...
An led I'm considering is the Luminus Devices CST-90.  It seems to be the best balance between $s and lumens.  $45@2700lm with a perk of having a thermistor.
Question:  It runs at 3.2-3.7V @ 1-13A
With a car as the source what is the best method of power supply which has the following qualities:  cheap, easy, full dimming, and doesn't destroy led in 2 days?

[Alex Tsekenis]

You could use a normal constant current source which will be quite inefficient, you could use a hysteretic driver much more efficient or you could use very specialised ICs based on hysteresis again like the new LT3743 or the MAX16818. These are in order of increased complexity.

But I am sure there must be a ready product that meets these specs. That is where I would start.

[Hero999]

An SMPs can be made using fairly common components.

The first shows how it can be done using diecrete components and the second shows a comparator. If a comparator with a push-pull output is used Q1, D1 R4 and R5 can be eliminated.

[Um...Me123]

I'm new to SMPs but see that is definitely my way to go.  For benefit of learning, what would be wrong with just PWMing a mosfet and putting a cap across the output?  I can see that there could be fluctuations in output when the input changes but besides that is there any reason one couldn't do such a thing?
Thanks,
A

[Alex Tsekenis]

No, no reason why you couldn't do that. But do you want to do that?
Once the transistor is ON, your output will shoot to 12 V (assume car battery) and there is nothing you can do about it by changing the duty cycle. When the transistor is OFF your output volatge will fall depending on component values. You can't regulate this system.

Anyway, constant voltage drive is not suitable for power LEDs. You need constant current drive (see suggestions below).

Attached is a waveform of the output voltage of your proposed system to illustrate my point.

Also, see what I came across:

http://www.amazon.co.uk/gp/product/0750683414/ref=sib_rdr_dp

Prob. an overkill.

[Hero999]

I'm new to SMPs but see that is definitely my way to go.  For benefit of learning, what would be wrong with just PWMing a mosfet and putting a cap across the output?  I can see that there could be fluctuations in output when the input changes but besides that is there any reason one couldn't do such a thing?
Thanks,
A

You're half way there.

You need an inductor, not a capacitor if you're going to do that. The value will depend on the switching frequency, input voltage, LED voltage, current and maximum acceptable current ripple.

It's true that the current will vary depending on the input voltage but that might not be a problem. This circuit is the switch mode equivalent of just using a resistor but it's well worth it as it's much more efficient.

[Alex Tsekenis]

And just to expand on Hero's post below, this is the topology of the ICs suggested earlier in this topic. You can google hysteretic LED driver for more info, Elektor Electronics (the magazine) published an article this year too.

By adding a resistor in the LED's path you can sense it's current and modily your duty cycle.

[Hero999]

The unregulated circuit in my previous post is based on the same topology as the circuits in the post before it.

The two regulated circuits I posted both work on the same principle: the output transistor turns on, as the current in the coil builds the voltage across the sense resistor exceeds a certain threshold, set by either a comparator or the conduction voltage of a base-emitter transistor junction, the output transistor turns off for a time determined by an RC time constant and the cycle repeats.

Here's an article which describes how the comparator circuit I posted works, see page 38.
http://ww1.microchip.com/downloads/en/DeviceDoc/41215C.pdf

You could use a discrete comparator if you like but it needs to be push-pull, if it's not i.e. it's LM311, LM393, use the circuit with a comparator above.

Note that the circuit contains an error: the +/- inputs are the wrong way round.