I have to admit that I do not know how to convert the power of the flash to lumens. To simplify the equation let's say we have I have all the led's "focusing" the same way as the flash to a distance of 5 m. How many leds would we need in theory to achieve the same peak power?
Suppose you have enough static lights to take a good picture at 1/100. That's usually a very well lit room, though not like broad daylight. While the shutter is closed, the lamps do absolutely nothing, photographically speaking. If you could synchronize the lights to the shutter, so they are only on while the shutter is open and exposing the "film" (whether actual film, or a CCD or other sensor), you could take exactly the same photos, with a fraction of the overall power used (because, unless you're shooting video, your shutter is closed most of the time).
Give or take things like automatic focus / exposure / F-stop / ISO, which you may or may not have set, which are calculated based on the static scene. Obviously, you have to have it set correctly for the scene with lighting.
Now, suppose you wanted to take photos at 1/1000 instead. You need ten times as many lamps, to provide ten times the illumination, for 1/10th the duration of the previous example. The energy is the same, but the cost just went up, because LEDs can't handle too much peak power.
Whereas, since the flash tube is already in the 1/1000 to 1/100,000 range (depending on design), with high peak power, so you don't have to do this sort of scaling with flashes when the exposure is longer.
Now... if you had a super-high speed shutter (in the micro to nanosecond range), the flash pulse would be many times longer than the shutter opening, and you would again have to use a great many lamps to achieve the same exposure. (On the plus side, this sort of photography was often used for
extremely luminous subjects, so lighting wasn't a big deal.)
Now let's assume that I do not want speed photography, I want some decent 1/500 exposure. Did not calculate, but from what I saw in terms of these leds, 100W power equivalent (10 pieces of those leds) would give me a good aperture of let's say F8 or? So if you say that led's are more powerfull, I could do the trick to load a capacitor that would discharge 100W in 1/500 sec, or?
As mentioned, LEDs unfortunately don't do pulses very well. There are a few reasons:
- Thermal -- chip and connections (bond wires)
- Saturation: at high current densities, the LED itself becomes less efficient, and you suffer diminishing returns. Effects from charge carriers, recombination and reabsorption (the LED acting as a photocell to its own internal light!), among other sundry physics, play a role here.
- Phosphor: the LED itself is blue; some of this is converted to a broad yellow, giving an overall white. But it can only do so much, for similar physics as the LED itself. (I don't actually remember offhand what all mechanisms limit a phosphor, but they should be similar, at least.) It's also slower, so whereas the blue light might be controllable in the 10MHz range, the yellow part won't -- something to keep in mind if you're transmitting data over ambient light.
This is easy to imagine if you remember those "nite glo" phosphors, that can be "charged" with a bright light, then keep on glowing for hours. Every phosphor has a different decay curve, some fast (~femtoseconds), some slow. What happens is, a given "center" in the crystal (which might be a rare earth dopant) gets excited by some energy, then after a while, decays to the ground state, releasing light of some color (with less energy than the original stimulus). While an atom is stuck in the excited state, it's *not* giving off light, so the power transfer of a phosphor has to be inversely proportional to the time constant. LEDs are pretty bright, so their phosphors have to be fairly fast to keep up (without needing a massive chunk of phosphor, that is). Trying to pump more energy into the phosphor will either accomplish nothing (it goes right through, not being absorbed), or pushes atoms to even higher energy levels, causing damage (ionization, permanent chemical change, dislocations..). Which I think is a major player in the life of a phosphor as well.
- CRI: the efficiency of the phosphor varies at different intensities, too. Unless the device is characterized at different currents, it's probably only correct at the rated current. (This could be one good reason to PWM LEDs to control brightness, rather than regulating DC current.)
For the same reason as flash, I cannot keep a powerful LED long as it just kills the eyes...
And certainly not a hundred at once!
Regarding existing hardware, an LED lamp may take anywhere from ms to 200ms to turn on, and may have undesirable ripple in the output (anywhere from line frequency modulation to 100s of kHz).
Now this is a bit contradicting for me, against what you said at the beginning. Are they fast enough to turn on? You said "LEDs actually light up faster than flash tubes do: less than a microsecond" now: is it it less than one millisecond or is it 200ms... .Or I did not get something right?
Yes.. well, unless the application requires it, the driver can be pretty much whatever speed the designer wishes. Typical figures might be:
- < 2us: linear current regulators, PWM (unregulated), very fast switchers, special purpose (optical data, etc.)
- 10-100us: medium to fast switchers, regulated PWM (e.g., backlight driver chips, which can be on-off PWM'd, or analog throttled); the scope shot above is a perfect example of this sort of design and performance
- 10-50ms: soft start circuits; line powered switchers; passive LED strings (e.g., Xmas lights -- usually pulsing at line frequency)
- 100ms+: Active PFC switchers (the long time constant is necessary to filter out line ripple), intentionally long time constants (to 'soften' transitions?)
So if you're looking to explore this practically, be mindful that probably no one really cares about how fast their light is, so if you want pulsed drivers, you'll need one that's documented as being fast enough. Nothing that needs, like, custom design and certification, but you may have to test a few products (like those eBay drivers) and see which ones work to be sure.
Tim
Home
Help
Search
Login
Register
