Go Down

Topic: How to create a controllable and bright LED strobe light (Read 14156 times) previous topic - next topic

edmark

I have created a blooming zoetrope sculpture. I'm trying to turn this into a stand-alone display where the user can control the strobe. Rather than using a xenon flash lamp strobe, I would like to create an LED strobe light with a controllable frequency of between 10hz and 100hz. In order to avoid motion blur, the on time for each cycle needs to be less than 0.5milsec. Because the LEDs will be on less than 0.5% of the time, it's important that they be very bright (even though I only need to light a volume of approximately 1 cubic foot).

Does anybody have any idea whether this could be done with, say, an Uno? And if so, how? I'm thinking that I will need a total of at least 50watts of LEDs in order for it to be bright enough, but haven't been able to find any LED arrays along those lines. (On good thing about the on time being so short is that I shouldn't have trouble with the LEDs over-heating.) I've been told that the phosphers on some LEDS take some time to come up to full brightness, and also some time to go dark, so I'm not sure what kind of LEDS will work.

Any advice would be greatly appreciated.

Paul__B

Can't speak for the phosphors, and you clearly need a power MOSFET switched in turn by a transistor (to obtain the necessary gate turn-on voltage for the FET) to control 12 or 24V for the LEDS, but I might suggest using LED strips instead of spotlights.  For such short duty cycle you should be able to overdrive them by a factor of at least five (though it would be nice to know the actual specification of the individual LEDs in order to do this).

Helmuth

Cool project!

50W led light you could get from a RGB strip/matrix but: Check their internal PWM frequency, because that is the highest update rate they can deal with. It is arround 400 fps for the WS281x which is to slow for a 0.5% on cycle at 100 Hz.

LPD8806 Strips might be the better choice.

Probably cheaper would be a 50W led driven by a MOSFET.

The Uno is enough for the timing you need.

P.S. For 0.5 mS you can drive the most leds far beyond their specs...


edmark

Thank you Paul__B and Helmuth! Glad to know I should be able to overdrive the LEDs.

Helmuth, I'm a bit confused about your recommendation of the LPD8806. Given that I don't need to individually control the multiple LEDs in a strip or matrix, why would I want to get a strip/matrix that permits individual addressing? Wouldn't I be better off just getting a "dumb" set of LEDs, and controlling them as a group?

Helmuth

Quote
why would I want to get a strip/matrix that permits individual addressing?  
Only to avoid any extra work in case you are lazy.
You pack it out, connect it to a power supply and the Arduino, program it - done.

If you are willing to solder, dumb bright leds are better and cheaper, of course.

Headroom

I guess the first thing you have to decide is what form of light source you want. A spot light is  obviously difficult to realize with LED strips. However a light source with a larger surface is easy enough to make with LED strips. The individually controllable may simply be easier to drive, you use a 5V power supply with the need amperage and a given Arduino with one of the existing Libraries, e.g. FastLED2. Or you could use a Teensy 3.1 with the OCTOWS2011 library, in case you need several thousand LEDs ;-)
Instead of strips you can also use one of a variety of available LED panels. the one linked is from Adafruit and they usually have an Arduino library to go with their products.

Another option is to light your sculptures from the inside. If they are 3D printed I am assuming that the plastic is translucent enough to allow for that. Then you need either a single high power LED or at least a relatively small number (compared to an LED strip) of high power LEDs. In that case I personally would pick one or more of the ledengin LEDs. these are not cheap but available in high wattages and ledengin is a ver reputable manufacturers as opposed to a lot of the triple-c products on eBay (Cheap Chinese Crap) that have no spec sheet or really any documentation.
These I'd drive with one or more of my own High Power LED shields

Last but not least your sculptures and animations are a really novel idea and simply beautiful to look at.
http://trippylighting.com

http://ledshield.wordpress.com/

Paul__B

Given that I don't need to individually control the multiple LEDs in a strip or matrix, why would I want to get a strip/matrix that permits individual addressing?
You most certainly would not want individually addressable LEDs(, strips or panels).  That would be totally useless since you want a strobe.

(Guys!  Read the description before you make suggestions!)

Chagrin

50W of LED is quite a lot of light -- that's like 400W of incandescent light. If you looked at a single LED of that power you'd absolutely be seeing spots, or it would be much worse than watching someone welding. For your application I can't imagine that more than 10W would be necessary (and even 10W LEDs are too bright to look at).

DC-powered LED drivers shouldn't have any problem keeping up with the 1KHz rate you want. This one will do up to 20KHz. Don't select a driver that puts out more amperage than the LEDs you select and try to stay reasonably close to the total wattage spec.

Interfacing an Arduino to a driver like this is very straightforward. You'd just need a 10K potentiometer to connect to the Arduino to use as user input, and a single digital pin is all that you need to connect to the LED driver (aside from power, of course).

Headroom

You most certainly would not want individually addressable LEDs(, strips or panels).  That would be totally useless since you want a strobe.

(Guys!  Read the description before you make suggestions!)

What technical reasons speak against using individually addressable LED strips ?
Is there some aspect that I am overlooking that renders them unfit for a strobe light ?

http://trippylighting.com

http://ledshield.wordpress.com/

Paul__B

What technical reasons speak against using individually addressable LED strips ?
Is there some aspect that I am overlooking that renders them unfit for a strobe light ?
I rather thought that was obvious.  (But of course, people have suggested doing so, so perhaps it is not.)

You have to send a quite complex command sequence through the series chain.  What do you suppose that does to the response time?

Headroom

I rather thought that was obvious.  (But of course, people have suggested doing so, so perhaps it is not.)

You have to send a quite complex command sequence through the series chain.  What do you suppose that does to the response time?

Available libraries such as the ones I mentioned in my post make sending that complex command sequence quite fast and easy. I would not anticipate a lot of interference with the 10Hz - 100Hz pulse frequency the OP is looking for. In case of the WS2812b (and similar chips/LEDs) the complex command sequence is sent out by the micro controller with 400KHz. There are faster strips available.
Using WS2811 (or WS2812b etc) strips or panels the Teensy 3.1 (Available from OSH park for under $20) manages to refresh several thousand LEDs at still 60Hz. So a smaller panel with say 16 x 16 LEDs will easily go over 100Hz.

A panel or strip would open up very nice possibilities of fine tuning the lighting.
http://trippylighting.com

http://ledshield.wordpress.com/

edmark

Thanks for the helpful info (and kind words about my zoetrope)!

So, I now have a 100w 30v 2.4a LED chip strobing successfully, using this code on an uno (with a mosfet):

void setup() {
pinMode(2,OUTPUT);
}

void loop() {
digitalWrite(2,1);
delayMicroseconds(100);
digitalWrite(2, 0);
delay(20);
}

But it's not as bright at I need. I may have to resort to using several of these LEDs in parallel, but I would really prefer to over-driving this one LED,if possible. The problem is that this LED is 30V, and I don't have access to a bench-top power supply that goes higher than 30v, in order to test such a setup. I'm thinking of just going ahead and buying a constant current power supply rated at, say 120v, to get a 4-fold increase, and adding a capacitor so that I don't have to get a very powerful power supply (the average current when I'm running the strobe is less than .01 amps.)

But I wish I could try out various levels of over-driving before committing to a specific power supply. Does anyone have a suggestion for how I might do this? And, assuming I'm willing to destroy a few LEDs in the process, what kind of upper-bound should I be considering for over-driving this 100W 30V LED? Above, it was suggested that 5x is likely to work. Is 10x a possibility worth trying, or is that definitely too much?

(I can't seem to find any specs for the LED beyond the standard stuff: http://www.amazon.com/LOHAS%C2%AE-White-Power-Energy-Saving/dp/B00CZ75TWA/ref=pd_bxgy_hi_img_y)

Helmuth

Quote
And, assuming I'm willing to destroy a few LEDs in the process...
Have a look here for destructive testing of a LED: Link

Some datasheets tell you about the "absolute max" ratings for "peak pulsing current".

Headroom

Your duty cycle ratio is 20ms of off-time versus 100 microseconds (= 0.1ms) of on time so  meaning your LED is on 1/2000 or 0.05% You may want to increase that to at least 1  millisecond or more. I bet your cameras shutter-open time is more than 100 microseconds!

If the max forward current Vf for they LED array is 34V and it's a100W LED then the max forward current is about 3A.

In order to drive your power LED correctly to its brightest you need to supply it with a constant current near it's 3A rating. It is cheap enough so you can experiment with driving it to higher currents for pulsing with the risk of frying it.
The voltage you apply just needs to be above the forward voltage of the LED so if the forward voltage is 30V-34V  a 40V power supply is all you need. You will NOT make it 4 times brighter by going to 120V, you'll just  dissipate more energy elsewhere.


Also average current does not mean much other than that your duty cycle is way off! If your peak current is above the max current the LED spec allows then you'll damage your LED over time. As mentioned above, this LED is cheap enough so you can experiment in driving it to higher currents (NOT voltages!!!) for short pulses but risk frying it.


http://trippylighting.com

http://ledshield.wordpress.com/

Paul__B

The limitation here is that the LED may be considered as a voltage drop in series with a series resistance.

The brightness of the LED is proportional to the current as the quantum step (voltage) involved in the light generation is constant.  Much of the power dissipated is however in the ESR - the Equivalent Series Resistance, and that power is proportional to the square of the current.  So when you drive it at twice the current for half the time, that heat component nevertheless doubles in the process (that is, four times the power for half the time).

Go Up