# High Power LED flash or strobe

#### BReeves

Joined Nov 24, 2012
410
I would like to build a high power LED flash with a really fast recycle time. ~15 flashes per second. The unit I would like it to be able to replace uses a flash tube and is spected at 3800 Lux (350 foot candles) @ 8 feet.

I'm not really up on high power LED's and not sure what or how many of what LED it would take to do this. Thinking of stealing the idea from flash tube strobes in that the power to fire the LEDs could be stored in big capacitors. Cycle time between 15 shot sequences can be minutes which is time enough to recharge the capacitors. This scheme wouldn't require a power supply rated at the power requirements of the LEDs.

First challenge is figuring out what type and how many LEDs will be needed. Once we have that down we can figure out the electronics that will be needed.

Forgot to add this will be powered from the AC line and could use 220 if necessary.

Last edited:

#### MrChips

Joined Oct 2, 2009
24,619
If the circuit is to mains powered then there is no need to recharge the capacitors. Simply use a step down transformer and build an unregulated DC power supply. Then you can use a 555 timer circuit to drive a switching transistor such as a MOSFET to turn on the LEDs.

#### ronv

Joined Nov 12, 2008
3,770
What is the time between each one of the 15 flashes? About how long for each flash?
This will tell us how high we can over drive them.
Any special color of white?

#### Alec_t

Joined Sep 17, 2013
12,312
Using this lamp as a baseline, and applying the inverse square law to its spec, I reckon you'd need the equivalent of ~ three times the number of LEDs similar to this to get 3800 lux at 8 ft if powered continuously. For pulsed use, over-driving the LEDs while keeping to the same average power-level should be possible. No spec for lamp or individual LED power requirements given in the link.

Last edited:

#### Bernard

Joined Aug 7, 2008
5,788
How large is the area to be illuminated?
Just finished an 18 LED blinking light for halloween costume, 10 ms on 20 ms off. 555 & FDS9435A P ch FET, 3 AAA's.

#### BReeves

Joined Nov 24, 2012
410
Thanks we are making a little headway..
The light will be triggered by an external 5 volt input signal, or better yet have the light supply the 5 volts and the external trigger can be a contact closure or transistor. It needs to be able to fire with the external trigger ~15 times in a second once a sequence has been shot it might set idle for several minutes then be able to fire another 15 shot sequence. Each shot will be fired from the external trigger, no need for a 555, it may only need to fire 5 times in 1/2 second but having it able to do 15 per second would cover most situtations.

I have a drawing of the light pattern and maybe flash duration but it's at another location, I will get more info on that aspect tomorrow. (Drawing is showing a 3 foot circle at 15 feet with 710 Lux at the outer edge)
I was figuring I would need some sort of reflector or lens to concentrate the light in the proper direction.
Duration wouldn't be really critical, the camera has a shutter and it will usually be set around 1000 of a second or faster. Something like 500 ms should be long enough. (Spec on the factory flash is 100 Microseconds)

Thinking once we get the type and number of LEDs figured out we can work on the details.

Last edited:

#### BReeves

Joined Nov 24, 2012
410
Back at it... Found a 100 watt white LED array on eBay for \$6.8o and ordered 4 thinking I could test 2 or 3 to destruction and still have 1 or 2 working. The specs are 10,000 Lumen at 3500 ma with a 35 volt supply.
Found a little info on the net about over driving high power LED's resulting in figuring if I over-drive it by a factor of 8 it will emit about 4 times the light or 40,000 lumen. The info I found says I need to keep the Duty cycle at 5% which is pretty easy in my application with an on time of 500 microseconds.

8 X 3500 ma is about 28 amps. Thinking rather than using a 28 amp power supply I could charge an electrolytic during the off time and let the cap provide the high current needed for the LED. Worst case off time will need to be 10 milliseconds but in reality it will be more like 50 to 100 milliseconds.

With this information in hand I drew up a rough schematic for a test circuit using a 555 to provide the 500 microsecond pulse and a mosfet switch for the LED. Items I'm not real sure how to come up with are.....
1.What voltage will the LED power supply need to provide.
2. How large Electrolytic caps will I need.
3. The values of R1 and R2 which will depend on the voltage(#1) and size of the cap(s).
4. What would be a good Mosfet for this application, have no experience with mosfets at all.

Any other suggestions or corrections to my thinking are more than welcome.

#### Attachments

• 37.7 KB Views: 114
Last edited:

#### MrChips

Joined Oct 2, 2009
24,619
You will have to do some trial and error experimentation.
Here are some seat of the pants calculations.

Firstly, assume the LED operates at 35V@3.5A
That represents a load of 10Ω.

If we use a time-constant tau of 1ms (as a starting point)
C = tau/R = 0.001/10 =100μF

Hence you want R2 = 0Ω

The pulse duration will depend on the value of C. I think you can safely make C = 1000μF
The intensity of the LED is dependent on the supply voltage. You might experiment with increasing the supply voltage to 70-100V for double the current.
The value of R1 will control the recharge time. Try R1 = 100-500Ω.
If you choose a 100VDC supply, you will need about 100VDC@0.5A

#### BReeves

Joined Nov 24, 2012
410
Great, gives me a good starting point for the tests.
Going to be a while before the LEDs get here (coming from China) in the interim I will need to get a few mosfets ordered. Looks like I will need one that will handle ~30 amps at ~100 volts. In what little research I have done it seems like there are a few different types of mosfets? Anything special I need to look for?

#### BReeves

Joined Nov 24, 2012
410
One of these days I really need to download and learn how to use LTspice
Thanks for the tips, looks like the mosfet you suggested will work, just ordered 5 from Digi-Key. For testing, the power supply isn't going to be an issue, I have a 0-40 volt 0-40 amp bench supply but from my calculations with a 150 ohm resistor I should need less than 1 amp at 35 volts. This is where I will start, have all the parts ordered now off to eBay to see if I can grab a flash meter for a reasonable price.

#### Attachments

• 32.9 KB Views: 116

#### ronv

Joined Nov 12, 2008
3,770
If you are going to run it from the big supply you don't need to worry about the capacitors. You can just limit the current with the resistors in series with the LEDs.
Here is the spice file that you can play with. It does not have your exact components, but pretty close.
Take note that there is a N=10 after the LED part #. This makes spice think there are 10 in series.

#### Attachments

• 1.9 KB Views: 46

#### BReeves

Joined Nov 24, 2012
410
Great, thanks allot.

#### BReeves

Joined Nov 24, 2012
410
Been playing with LTspice, thanks to Ron that got me started I think I have a better handle on what I will need. Using the LEDs available in the default library I came up with a diode array that closely simulates the current and voltage of the 100 watt LED I have on order. To get close to the 28 amp over drive current I originally calculated it looks like I will need an 80 volt power supply.

What I am trying to determine now is what current I will need from the supply. If I use ohms law and assume the caps are completely discharged at start up I come up with 8 amps through the 10 ohm resistor. However LTspice is showing a much lower current in the simulation. Wondering if I am missing something in what LTspice is telling me.

This is what I have right now, the current through R2 is the off yellow trace.

#### Attachments

• 147.8 KB Views: 69

#### ronv

Joined Nov 12, 2008
3,770
The 8 amps is just to initially charge the caps. After that most of the current will come from the caps.
But if you make the timing like your original spec. the voltage will sag as the caps discharge. So make sure you have the right time on and the repetition rate you need.
I think I would be surprised if your LED liked 28 amps - but maybe.

#### BReeves

Joined Nov 24, 2012
410
It works..

Finally managed to receive everything I ordered and put a prototype together. First I tested with a safe 35 volt power supply. This is whiten the LED's continuous ratings so I was pretty sure I wouldn't blow anything up. Set up my flash meter and fired away.. Only had the flash meter about a foot away from the LED and it read F5.6 at 400 ASA. Then I replaced the 35 volt supply with Big Bertha, an 80 volt 10 amp supply I built inside the case of a 1000 watt 220 to 110 step down transformer. Fired it several times and nothing smoked.

The flash meter now read F11 at 400 ASA the same distance as in in the first tests. Gained two F stops which if I'm right is 4 times the amount of light which is exactly what I was expecting. If the specs for the LED are accurate this should be giving me about 40,000 lumens. Next step is to add a lens and reflector to direct all 40,000 lumens in one direction and compare the light output with a studio strobe.

Discovered the hard way when I connected the trigger input to a pulse generator that the transformer I was using for my 80 volt supply isn't isolated. The pulse generator is actually the parallel port on a computer running software I wrote that simulates the triggers from a multiple seat roller coaster. I smoked the 555, mosfet and burnt a ground trace on the board. Have it all fixed and now running off the 35 volt isolated supply from the original test. The burnt ground trace ended up being the fuse that prevented a real catastrophe. Using an inexpensive 220 to 110 step down transformer for a power supply isn't a very good idea as I have since discovered most are auto-transformers and one of reasons they are cheap.

I found and ordered a 600 watt DC to DC boost supply off eBay that hopefully will take the 35 volt output of the existing supply and boost it to 80 volts. At least the specs say it will. Then I can start experimenting with resistors to see how fast I can make it fire at full output. Right now I have a 10 ohm resistor between the power supply and the caps and a 1/2 ohm resistor between the caps and the LED.

I did find out when the mosfet shorted the LED will take 80 volts through a 10 1/2 ohm resistor for at least a couple of seconds, boy was it bright but it still works. This has me thinking I can hit it harder than I first thought if I keep the duty cycle whiten reason. This was taken in the shop before I smoked it, was using a push button switch to trigger it which wasn't tied to AC ground.

#### Attachments

• 159.1 KB Views: 94

#### ronv

Joined Nov 12, 2008
3,770
Cool.