LED Camera flash?

OK, I have a requirement to take pictures with a flash at the rate of 3 per second for 10 seconds. The range of the flash needs to be about 50' and cover an area about 15' square at distances from 20' to 50'.

I am wondering if anyone has used LEDs to build a camera flash and what challenges exist when using LEDs in this type of application. I have searched but all the LED flash info I have found so far is about camera phones.

It doesn't really matter if I have to use multiple LEDs and drivers or whatever. (Budget shouldn't be an obsticle.)

My thought was to leave the light on until the photos are taken but the requriement is for it to work like a flash.


Some questions that come to mind.

1. Are LEDs fast enough?
2. Can I eliminate the charging circuit that standard flashes need?
(Assuming yes, which is really the point of using LEDs.)
3. Will using LEDs like a flash, quick frequent cycling, cause degredation?
4. How many lumens are produced by standard 400 Watt flashes? (What we currently use. These illuminate a much larger area than required though.)

Thanks for your help.

Pertinax

Designing a LED based flash to replace an (incredibly powerful in comparison) 400 W/S monolight or flash head is waaaaaaaaaaaaaay beyond the scope of anything I've imagined, but I'll try and help...

1. Yes. "strike" time to come up to full power is in the order of nanoseconds, according to LumiLEDs. Able to beat Xenon flashtubes

2. If you can somehow get a circuit that is able to turn on and off a sufficient amount of power to the assembly, that would be possible. The problem is 'how'.

3. No, not that I know of. Many variable-power LED lights are controlled by Pulse Width Modulation (PWM), which essentially cycles them on and off many hundreds of times a second. PWM is the recommended form of dimming for Luxeon LEDs according to their own datasheets.

4. This is the tough one - I don't think there's a direct comparison but my rough guesstimate is that a small on-camera hotshoe flash already produces several thousand lumens of light for a tiny fraction of a second. At full power it can totally flood a small room to the point of whiteout (nearly total overexposure) whereas even our brightest lights barely come close to that.


Just for fun I took out my camera and Canon 430EX Speedlite. Set camera to manual, 1/200 sec, f/4.0, ISO 400. Set Speedlite to manual power control, 1/16 (no E-TTL2). Took a couple of frames, the room comes out nicely illuminated.

Turned off 430EX, used known 1150 lumen light source to illuminate the room. Took another couple of frames - completely unuseable. The frame is almost completely evenly black!!

So even at 1/16 power, the 430EX kicks the pants off the 1150 lumen source.... and as I've heard, the small hotshoe flashes like it are less than 100 w/s at full power!

Lux I's or III's is what you'll want to use. LEDs turn on within a few nanoseconds, but the driver should be powered up and standing by for ignition. Switching drivers sometimes take hundreds of milliseconds to power up. Pulsing an LED won't hurt it at all, in fact they can handle much higher current when they're pulsed. You can wire then in series and drive them with high voltage and reduce the complexity of the power supply (assuming you're using 110 AC). A common R-bin Luxeon I will produce 45 lumens at around 1 watt. Do you know how many lumens you need and what dispersion you want? The dispersion of a Lambertian Lux is 180 degrees semi-spherical without a reflector.

Not sure about using an LED for a flash. If you get it to work, make sure you do a write up on here.

I know that my video camcorder uses 3 led's (probably 5mm's) for some area lighting, but they don't throw at all. I really is only useful for close up.

For your application, I'm thinking multiple luxIII's maybe? or maybe an array of luxIII and luxV's like this:
0 0 0
.*.*.
where 0=lux 3
and *=lux 5

If it does work, you probably still need to charge a capacitor, just not to a very high voltage. I'm don't think (but not sure if) it is a good idea to just cycle the circuit using a bare battery as the source (rather than a capacitor-charge circuit).

Hopefully someone else will be able to give you a better answer.

1 - yes
2 - sort of, you would need the appropriate driver circuitry for the LEDs instead of the appropriate circuitry for a xenon flash bulb
3 - not really AFAIK
4 - a LOT. Think of it this way, your 400 watt-second flash head is dumping out 400,000 watts of power in its 1/1000 second flash duration.

Why are you trying to use LEDs for this? Just get the appropriate strobe(s) and be done.

Let's see ... 50 foot range, pick a moderate f-stop like f/5.6, that means you need a minimum guide number of 280 at whatever ISO you are running. Since you are currently using 400 W-s heads, that tells me you are running a slightly smaller f-stop closer to approx f/8.

Again, why are you trying to use LEDs for this?

Thinking of incans, I'd SWAG that your exposure (not complicating things with the strobing for right now) is using 10,000+ lumens at the light heads. That + might be a factor of 2, 4, or more. So you'd be looking at 100+ premium Lux-V LEDs plus the appropriate driver and strobe circuitry. At ballpark $30 per LED plus $20 per driver plus assembly plus reflectors plus housings plus-plus-plus, I'd guess that your cost for this would be $10,000 give-or-take. Seriously, $10,000+.

Just get the appropriate flash heads. To get your strobing, you could either get multiple 400 W-s units and sync them to get the strobe or get higher power units and run them at reduced power to quicken the cycle times.

I really don't think LEDs are the way to go for what you say you want to do.

MoonRise said:

Thinking of incans, I'd SWAG that your exposure (not complicating things with the strobing for right now) is using 10,000+ lumens at the light heads. That + might be a factor of 2, 4, or more. So you'd be looking at 100+ premium Lux-V LEDs plus the appropriate driver and strobe circuitry. At ballpark $30 per LED plus $20 per driver plus assembly plus reflectors plus housings plus-plus-plus, I'd guess that your cost for this would be $10,000 give-or-take. Seriously, $10,000+.

Click to expand...

Maybe much less than that. You have to consider buying bulk and using the most efficient Lux's which aren't Lux V's or Lux III's. 10000 lumens/45 lumens per Lux I = 222 Luxeon I's. At Future Electronics they are $2.99 each X 222 = $663 The power supply would just be one unit and I could design one for $200 or less.

As a photographer, I would be more concered about the tint of the LEDs. Flash lighting by itself causes a certain starkness in a picture, I would want to make sure that the LEDs were as warm in color as possible.

Dadof6 said:

As a photographer, I would be more concered about the tint of the LEDs. Flash lighting by itself causes a certain starkness in a picture, I would want to make sure that the LEDs were as warm in color as possible.

Click to expand...

I was about to say the same thing. Your color rendition may be a problem as well.

Unless you are doing b+w and/or going for some high contrast effects, you will probably have color and tint issues.

millermods,

A 45 lumen spec Lux-I R-bin will get you ~30 lumens real world, after temperature and optical losses.

So he'd need 350+ Lux-I LEDs, plus circuitry, plus opticals, plus-plus-plus, for 10,000 lumens.

Spaced on 1-inch centers and using 20mm reflectors, he's looking at a lighting panel that's approximately 12x30 inches to get 10,000 lumens. And as KevinL proved, I think my initial guess of lumens needed was waaaaaaay low. My new guess is that KevinL's 1/16 power Speedlite pops being greater illumination than a known 1150 lumens source mean that his piddly little 100 W-s shoe-mount flash unit pumps out more than 10,000 equivalent lumens at full. So pert's 400 W-s studio flash heads are probably in the range of greater than 50,000 lumens equivalent.

So now he needs -5- lighting panels (or more!), with each one 12x30 inches and each panel containing 350 Lux-I LEDs. That's now 1750 LEDs, plus optics plus etc. Allow 10% "oops" in factory or assembly goofs, he needs almost 2000 LEDs at $3.00 each, or $6000 just in Lux-I LEDs.

Still looks like a $10,000+ project to me.

Just buy the right studio strobes and controllers and get the shots. Which BTW might still be a $10,000 cost if the current strobes can't be used in 'strobe' mode. Profoto D4 2400R kit is $5362.95 each at B&H Photo here http://www.bhphotovideo.com/bnh/con...195&is=REG&addedTroughType=categoryNavigation

Yikes! But could you image the glory of it all?! 2000 Lux I's with IMS 20mm reflectors all on at the same time focused on one area!

If you used the warm version Luxes the tint issue would perhaps be solved, or would it make it worst? But to get the same amount of lumens you'd need double the amount of luxeons.

MoonRise said:

Spaced on 1-inch centers and using 20mm reflectors, he's looking at a lighting panel that's approximately 12x30 inches to get 10,000 lumens. And as KevinL proved, I think my initial guess of lumens needed was waaaaaaay low.

Click to expand...

Yes, it is way low. I'm not sure where the figure came from. For a 400W strobe, that 400,000W figure was right for a 1/1000th second flash. Even at just 10 lumens/watt, you're looking at 4 million lumens.

For those keeping score, this is 100,000 45-lumen LEDs.

Lot's of answers but no one is defining the problem....

- What are you taking pictures of, and what do you think you need for a shutter speed in order to stop the action (if at all). The answer to this may either make your solution with LEDS easier, or virtually impossible.

- In terms of color, warm LEDS are not going to be any better than cool LEDS though Warm Luxeons may be better due to their smooth spectrum that is more like a halogen source as opposed to the unnatural spectrum of white LEDS. Are you shooting with film? If so, you will have a hard time getting good white balance. If you are shooting digital, will you have access to RAW sensor data, or only the processed data? If you have access to the raw data, you may be able to play with it to get better balance.

- If you do not need a really short shutter speed, maybe something will be possible. Most people are using continous operation assumptions for the Luxeon light output in this thread. Assuming you do not need it to last forever, you could run the LEDS really hard in order to maximize light output. At some point you get diminishing returns. If you get 80 lumens out of a LuxIII-T bin at 700mA, you can likely run it for 10milliseconds or so at 2-3A and get maybe 150lumens out.

- In terms of optics, you could put individual optics on a single luxeon, but you could group a ton into a circle and use a really large reflector.

- Assuming shorter duration, thermals are likely not much of an issue as duty cycle * power is likely to be low

Semiman

As already mentioned by other photographers, the colour temp of the leds is going to be a huge issue.

On another note, shutter speed is not really an issue, if you are relying solely on the flash and not exposing for ambient light at all, you can crank the shutter speed up as high as the sync speed will go on the camera. With flash only, it is the aperture that controls the exposure, shutter speed only comes into play when you want to use the flash as a fill, and are relying on the ambient light to do some/most of the work.

So it begs the question (unless I missed this), what are you photographing?

Oracle said:

Yes, it is way low. I'm not sure where the figure came from. For a 400W strobe, that 400,000W figure was right for a 1/1000th second flash. Even at just 10 lumens/watt, you're looking at 4 million lumens.

For those keeping score, this is 100,000 45-lumen LEDs.

Click to expand...

Holy cow....

I had no idea how to convert the two scales of measurement so I thought I would play with the camera a little (any excuse will do ) but now that we have a conversion, the numbers are mind-blowing.

I would think we're trying to solve the problem "as-posted" by the OP. Now that we realize it may be substantially more difficult than previously assumed, we may have to go back one step to the original requirements. Or, look at industry-specific solutions as has been proposed. I guess there IS a darn good reason why monolights cost as much as they do..

I was thinking that a rapid-fire substitute could be made with multiple banks of monolights and firing them in sequence, giving each individual head time to recharge while the others are firing. But I don't know anywhere near enough in this area to comment.

Well whether we call it shutter speed or flash duration, what are you taking a picture of and how fast it is moving?

While the color temp of the LEDS will be an issue, the unusual spectrum of standard cool white LEDs will be a bigger issue. Color balancing in digital cameras assumes a somewhat even spectrum (like a black body radiator). Even xenon has a pretty even spectrum, though the cameras will adjust for its unique spectrum. Unfortunately, LEDS are a completely different matter. You may have a hard time getting all colors to render properly.

Semiman

I think a very bright LED (Luxeon/K2/whatever) can be used as a focus assist in dark places. My camera (Canon 300D) strobes the flash to try to get lens focused and I'm not a big fan of this method.

The flash is VERY bright, so I'm not sure when a LED or a collection of them would replace a camera flash.

There is no way to replicate the output of a camera flash with LEDs. Looking at the specs for an AlienBees B400 AC-powered stand-alone flash, the unit puts out 7,000 lumenseconds for 1/6400 of a second. Run the numbers and this is 44,800,000 lumens. Half a million Luxeon V, maybe?

Purpose-built Xenon strobes are available at this kind of power. They're used for things like industrial product QC and stroboscopic analysis for R&D. See this page for a list of suppliers.

pert said:

OK, I have a requirement to take pictures with a flash at the rate of 3 per second for 10 seconds. The range of the flash needs to be about 50' and cover an area about 15' square at distances from 20' to 50'.

I am wondering if anyone has used LEDs to build a camera flash and what challenges exist when using LEDs in this type of application. I have searched but all the LED flash info I have found so far is about camera phones.

It doesn't really matter if I have to use multiple LEDs and drivers or whatever. (Budget shouldn't be an obsticle.)

My thought was to leave the light on until the photos are taken but the requriement is for it to work like a flash.


Some questions that come to mind.

1. Are LEDs fast enough?
2. Can I eliminate the charging circuit that standard flashes need?
(Assuming yes, which is really the point of using LEDs.)
3. Will using LEDs like a flash, quick frequent cycling, cause degredation?
4. How many lumens are produced by standard 400 Watt flashes? (What we currently use. These illuminate a much larger area than required though.)

Thanks for your help.

Pertinax

Click to expand...

400w-s flash pushes 400 joules of energy per flash, probably over a duration of 1/100 second.

If you want three 400 joule discharge into the lamp per second, you're going to need to be able to supply 1,200 watt. Something like this takes a big, AC powered unit, probably on a 240v circuit or a 20A 120v circuit.

You can't dump 100s of amps into an LED for a very short duration and use it as a flash. You're going to have to use a strobe for that, or use LEDs and use longer exposure time (so you can't freeze motion)

aceo07: If you get the opportunity to play with any Canon "EX" series Speedlite or the ST-E2 transmitter, you'll find they replace the focus-assist system with a series of red LEDs projecting a 'grid' pattern. These are 5mm LEDs, not even the more powerful Luxeons. A lot more forgiving to your subject than a repeated strobing of the flash itself. I have the 350D and it does that too.


enlightenment: I've heard that the specs for the AB units may be a little overrated, but still, those numbers are daunting. So big monolights really are like holding the power of the sun here on earth at the ends of your PC sync cord..... :bow:

Great responses! Exactly what I was hoping I would get based on other posts I have seen here.

Here are some answers/thoughts about some of the points mentioned.

Requirements clarification.

Taking pictures of the rear of vehicles for plate identification. Think speed cameras or Speed Cameras.

1. The flashes in these systems have to work off of a 12V system so it's a challenge to cycle a flash quickly for any sustained duration.

2. The shutter speed needs to be high enough to take a clear picture of the back of a vehicle traveling at high speeds. The ISO can be turned up on the digital camera though which helps. The reduction in clarity that results is not an issue because the subject is close.

3. The color temp doesn't matter much. The most important thing is the ability to read the plate and the reflective paint reduces the light required to do this. Yes, you need to be able to recognize the car at night but an accurate representation of the color isn't a big issue.

4. The area that needs to be illuminated is much smaller than the area currently illuminated by a flash. While a flash produces much more light than a LED replacement would, much of that light is not useful. I just need to illuminate an area large enough to cover the back of a vehicle.


Note: Something has already been made that uses a bank of 5mm LEDs as a flash replacement but only works for toll booths due to its limited range.

I can justify the expense of custom electronics for a power supply, a delay circuit to keep the LEDs on for the required duration. (May not even need to turn it off between photos.)

I know the speed camera and red light camera thing is a touchy subject. Unfortunately they are here to stay and I think it would be a LED victory if a standard flash system could be replaced by LEDs.