What would 'high-power' be considered? (Constant Current Source)

[Gryd3]

A friend of mine is wanting to build his own light-bar for his truck, and wants it 'daylight' bright.

Now... I have experience with the little 1W white LEDs, but the LEDs we will be using are 10x that..
The concern I have is from lack of experience. I have the theory behind setting up a lower voltage buck constant current supply and running the LEDs in parallel with a very low value resistor on each LED, or stringing them together in series and using a constant current boost converter.

The final power usage will be 10 of those LEDs which operate from 10-11V with a current draw of 900mA.
Considering I will need to supply either 10A.. or 100+V, which one should I pursue?

The title is a generic question that I'm not sure about. Is there a threshold when building a supply where it is considered 'high-power' or when special design considerations will need to be made? Or can I just up-scale a smaller constant current source and be fine?

 

[BobK]

There certainly is not a threshold that determines whether this is a high power circuit or low power, the designs need to change a bit for each step in power.

For your case, since the supply voltage is close to that of the forward voltage, I would tend toward using a simple constant current source for each LED. This would consist of a power transistor that would drop about 0.5V across an emitter resistor (0.56Ω) and a constant voltage on the base to raise Vbe to what is necessary for 900mA to flow. This would split the approximately 1V drop between the transistor and resistor.

Bob

 

[Arouse1973]

I agree with Bob constant current is the best option. Stay away from high voltage because moisture from outside can cause issues with circuitry that is not sealed.
Adam

 

[Bluejets]

Consideration would have to be given to the local traffic laws with regard to install of bright lights.
i.e. you cannot drive around with headlights on high beam blinding drivers coming from opposite direction.

 

[cjdelphi]

10watt chinese LEDs?

I've driven them purely from voltage, the diodes forward voltage drop depends on temperature, providing you supply a voltage which draws 60% of your total current, when summer arrives the foward voltage drop decreases and your current rises..

But, because there's some head room you'll find it might use 20% more current.


This method works, I've had no issues at all, but this only works with 3w and over.

A 5mm led the diode's vf change with heat would destroy it if you're supplying 20ma to a 20ma led, supplying 700ma to a 1amp led allows a 300ma swing which is fine if it's heatsinked nicely.

if not, you'd need to build a CC CIRCUIT

 

[(*steve*)]

A constant current circuit is a must.

I understand that some Chinese LEDs are manufactured to have a high forward resistance characteristic which flattens out the V vs I curve, but you shouldn't take the risk that one day you might accidentally use a genuine part by mistake!

There is an issue of the LED forward voltage is close to the battery voltage (e.g. Vf sums to 11.5V operating from a 12V source). If this is a truck with 24V electrics, and the LEDs have a Vf somewhere near 12V then you have plenty of overhead.

If you want many LEDs, the options are:

1. Use individual current sources for each
2. Use a high voltage source and drive them in series
3. Drive them in parallel from a single low voltage current source
   
4. As for (3) above but with additional series resistance.

(1) is obviously the best, but also the most expensive. If you go cheap by using a linear constant current source then you'll be dissipating as much heat in this as in the LEDs (assuming the rough 12/24 figures from above)

(2) has problems as noted previously, but in addition to issues relating to the high voltage, an open circuit failure of one unit will turn off all of the LEDs

(3) is just a bad idea. You leave yourself open for thermal runaway and a short circuit failure of any element will cause the entire string to drop out. Also open circuit failures will deliver more and more current to the remaining LEDs

(4) may be a useful compromise. If the series resistor is designed to drop maybe half of the excess voltage (6V in this case) then you are far less likely to get thermal runaway. The maximum current through any LED element is also effectively limited in the case of several LEDs failing open circuit. And finally any element failing short circuit would not cause all the other LEDs to go out (unless the resistor is shorted as well).

If you go with option (4), you would be wise to de-rate the LEDs by 50%. This is good practice in potentially hot environments anyway. The easiest way is to use 20W or 25W LED elements instead of 10W and drive them with half their nominal current.

If you're left with one (or maybe a few) constant current sources, then it makes sense to spend a few extra dollars design a switchmode device.

 

[(*steve*)]

Oh, another option that I didn't consider earlier is to have a single switchmode current driver with multiple sense resistors, each using a Schotky diode to effectively sense the one with the highest current.

In this case the voltage drop needs to be relatively large compared to Vf of the diode, but something between 0.5V and 1V would be heaps. Remember that the Vf falls with current, and for some Schottky diodes it approaches millivolts as the current tends to zero.

This could be combined with option (4) above to limit the maximum current even if most of the LEDs became disconnected.

 

[Gryd3]

The truck in question will be a 12V setup.
I do agree hands down that a const. current setup is required, and that's what I plan to build.
It was just building it for parallel or series operation for the LEDs, as the difference would be a supply that could provide in excess of 100V to allow for the constant current requirement of the series, or over 10A delivered to the LEDs in parallel (each with their own low value resistor to prevent thermal run-away of any individual light)

I do like the idea of using a linear constant current source for each LED in the array.
The truck voltage supply will vary from 11.5 to 14.4 depending on battery charge and if the alternator is running. I'm fine with dim LEDs while the truck is off, but need to make sure the power dissipated by the linear regulator while the truck is running is manageable.

The LED in question at the moment is http://www.dx.com/p/10w-900lm-3000k...yellow-silver-white-9-11v-192418#.VFRKf_nF-Qp
So at 14.4V, the regulator would need to dissipate anywhere from 3W to 5W. Considering I am looking to use 10 of these, I do feel I need to use switch mode supplies. I do not like the idea of wasting 30+ Watts as heat.

However, still from this conversation so far (thank you guys) I should probably segment the light bar and use smaller supplies rather than one big supply. This would most likely be well within my comfort zone.

 

[hevans1944]

You might want to consider a commercial light bar. Plug-N-Play.

Also, lower power LEDs for accent lighting come in strips that have built-in resistors every three LEDs or so. They can be driven with a low DC voltage applied between two parallel rails on the strips. Your choice of LED probably has a circuit board under the lens with five or so LED chips wired in series. It may also have a resistor included for connection to a voltage source, but you will have to discover that for yourself, perhaps by connecting one to a variable power supply to see if it behaves as a diode, or as a diode in series with a resistor.

 

[Gryd3]

You might want to consider a commercial light bar. Plug-N-Play.

Also, lower power LEDs for accent lighting come in strips that have built-in resistors every three LEDs or so. They can be driven with a low DC voltage applied between two parallel rails on the strips. Your choice of LED probably has a circuit board under the lens with five or so LED chips wired in series. It may also have a resistor included for connection to a voltage source, but you will have to discover that for yourself, perhaps by connecting one to a variable power supply to see if it behaves as a diode, or as a diode in series with a resistor.

Where is the fun in that?
And the LED I linked is the 'type' I am after, but is no means the final product choice at the moment.

 

[hevans1944]

Where is the fun in that?
And the LED I linked is the 'type' I am after, but is no means the final product choice at the moment.

Just saying, you should make sure it is just an LED (or several LED chips wired in series under the dome).

I agree, no fun installing what someone else thinks is cool.

 

[Gryd3]

Just saying, you should make sure it is just an LED (or several LED chips wired in series under the dome).

I agree, no fun installing what someone else thinks is cool.

I kind of figured it would be multiple chips wired together, but I guess they could always wire them differently. Thank you for bringin that up.