LED stage lighting flickers in video

JackW wrote on 12/9/2008, 11:47 AM

An interesting problem. After reading your post I just got off the phone with one of the technicians at Hollywood Lights who said that you can rule out the AC dimmer idea: LED lights use DC for dimming and cannot operate on AC dimmers.

His experience is that there is no consistency relative to flicker with LED lights. He points out that television studios routinely use LED lights with no adverse effect.

Hollywood Lights has installed LED lights in several large venues in the Seattle area and has had several reports of flicker. However, when a second video company comes in to shoot in the same venue there are no problems. One church they installed lights in reports that the house video equipment picks up flicker while a staff member who shot the same service experienced no flicker.

The technician suggests that there might be a difference in camera circuitry (I'm not sure what this would be,) in frame rate or in shutter speed.

So far our company has not experienced flicker and I hope that won't change. I'm interested to see what others bring to this discussion.

Jack

Lander. wrote on 12/9/2008, 11:48 AM

I think it may be your shutter speed. Try shooting with a slower shutter. You may need a ND filter if the stage lighting is really bright.

Lander. wrote on 12/9/2008, 11:51 AM

A quick test if you have the cam handy to test shutter speed is point it at a LCD monitor and adjust the shutter speed to see it it yields the same results as your footage

blink3times wrote on 12/9/2008, 12:40 PM

LED lights don't flicker (I should say PROPERLY installed led's won't flicker). They operate on a direct current which does not fluctuate like ac lighting. You CAN however get LED's at different wavelengths....one of the reasons why you can get different colors. Given that... I would imagine that you can get clashes with shutter speed and LED wavelength.

farss wrote on 12/9/2008, 12:49 PM

LED lights can and do operate at high frequencies. One dimming mechanism for LEDs uses PWM. The wavelength of the light is quite irrelevant to flicker problems.

Bob.

musicvid10 wrote on 12/9/2008, 12:50 PM

Blink,

DC comes in two flavors -- pulsed (rectified) and filtered (smoothed).

If the packs output filtered DC, there shouldn't be any flicker at any shutter speed.

If the packs are not filtered, the effect is the same as square-wave AC, only with a DC bias referenced to a 0 volt (relative) baseline -- IOW, a strobe-light effect at 120hz, presuming full-wave bridge rectification on a 60hz circuit.

We used to build IR LED audio transmitters/receivers for fun. Got 'em working up to 30 ft. in a couple of instances in low ambient light. Running on 9v batteries, they worked great, but on DC adapters the buzz was so bad even heavy capacitive filtering didn't remove it completely. It's the rapid slew rate in LEDs that is the problem.

Edited for spelling.

farss wrote on 12/9/2008, 12:53 PM

It's likely the problem is dues to interaction between the frequency of the LED lights and a CMOS imager.
Specifically which camera was used?

I've had this problem with fluro lights on dimmers. No way to get rid of it however it didn't show up us flicker rather as rolling banding.

Bob.

blink3times wrote on 12/9/2008, 1:06 PM

"DC comes in two flavors -- pulsed (rectified) and filtered (smoothed)."

Well.... yes... and no

Pulsed supplies are smoothed in the end. A computer or a car power amplifier uses a pulsed supply but they can't operate on a pulsed dc current.. so it doesn't mean the output is pulsed... Pulsing is simply the method used in the power supply. It keeps one from having to use a massive transformer like they did in early high powered home stereo systems but it's not pulsed in the end because smoothing capacitors and filters are added.

If however you use a cheap power supply with something like a half wave rectifier and very small caps on the output, then you surely will get flicker.

Christian de Godzinsky wrote on 12/9/2008, 1:15 PM

Hi,

LED lights DO flicker - on purpose!! Most (if not even all) professional LED driver circuits use PWM, to control the average optical output power. This is essential especially driving white LED's, to keep the color temperature constant, independent of the overall brighthess.

White LEDs (that are actually blue ones with a yellow scintillating phosphor on top) will certaily vary the color temperature (detectable by human eye) if you dim them linearly with adjustable current.

Even if it might sound surprising, primary color leds (R;G;B) also suffer from wavelength modulation versus operating current, even if the mechanism causing it is different.

So PWM is the way to go, and is usually anyhow the simpler (and cheaper) way to implement digitally a brightness control.

High(er) quality LED lighting systems use typically high(er) PWM frequencies, in the range of 500Hz and up to 5KHz, and even higher. Those systems should not pose any problems at normal shutter speeds (less than 1/100th of a second).

The only thing you can do yourself to approach the problem, is to choose a slower shutter speed, if that is possible - and a viable option.

Some of the earlier LED lights did not take this in consideration, they used very low PWM frequencies. Anything above 100Hz is normally flicker free for the human eye. For moving spotlights this PWM frequency should be at least 200Hz or even higher. This can still be problematic for electronic "eyes".

Some LED lighting electronics I have designed had a PWM that was deliberately set at 500.00Hz. This seems to work OK with normal digital cameras, as well as PAL (50Hz) video cameras. This has to do with the very precise PWM frequency, so the beat frequency (compared to 50i) is negligible...thus minimal no flickering is guaranteed.

Hope this sheds some light (pun intended) on the subject...

Christian

blink3times wrote on 12/9/2008, 1:20 PM

"White LEDs (that are actually blue ones with a yellow scintillating phosphor on top)"

That's wavelength you're speaking of... not flicker. LED's are diodes built with a certain wavelength in mind to produce the different color of light. Every color has its own wavelength and if LED's were built with all the same wavelentgh then they would only emit one color.

Flicker on the other hand is accomplished by varying the voltage and or current. You can do this on purpose... or by accident with a cheap power supply

blink3times wrote on 12/9/2008, 1:29 PM

"Anything above 100Hz is normally flicker free for the human eye."

And this may or may not be true depending on the application. Home lighting operates at 60Hz but you can't see them flicker with the naked eye. It takes the filament longer than 1/60 second (frequency is the inverse of time) to react so the flicker is more or less absorbed.

farss wrote on 12/9/2008, 1:38 PM

Not all "white" LED lighting uses "white" LEDs. Quite a lot use R,G,B LEDs to produce white or any color you want.

The mains powered LED light I have use unfiltered full wave rectified mains. They will flicker very badly at 100 or 120Hz. The problem is probably worse than iron ballasted fluro light. With fluros the phoshors have some persistance so the light doesn't entirely turn off during the zero crossing. LED lighting could be much worse, especially the cheap lights that use large numbers of low power LEDs. The more expensive ones using LEDs by Cree or Lumileds use constant current drivers fed from well filtered supplies. They do not flicker.

Bob.

Christian de Godzinsky wrote on 12/9/2008, 2:16 PM

My point is here that some lighting armatures uses "white" LEDs (not RGB matrixes) that are blue leds with a yellow phosphor. These together produce a somehat strange spectrum, a broader peak around yellow, and a narrow blue. This is perceived by the human eye as white, since all receptors (R;G;B) in your retina gets exitation. The color rendering might not be perfect, but anyhow. So far we are not talking about flickering at all.

Now - the problem is that if you would like to vary the light output of such a white LED, you will have a noticeable shift in the color temperature - if you do it by just adjusting a DC current flowing into the LED. The mechanism causing it is outside the scope of this discussion.

Therefore such LED lighting armatures that uses white LEDs (blue with the phosphor) normally drive the LEDs with a constant high current, but PWM modulated. If the brightness is an adjustable property.

Now we come back to the flickering. The selected PWM must be high enough so that the human eye does not see any flicker. 100Hz is good enough under all circumstances, assuming that the light or viewer is NOT moving. 200Hz would be already better with moving objects, 500Hz close to pertfect. But these two frequencies are still too low to produce guaranteedly flicker-free imaging with electronic cameras using higher shutter speeds.

Any tungsten filament based lamps have typically less than 10% of light ripple, even if driven with an AC voltage. Fluorescent lights have a 100% ripple, but at DOUBLE the mains frequency, but stay on for more than 70% of the cycle. So this is not 50/60Hz flicker, it is 100/120Hz flicker. The decay in the phosphors is very short, so short that there is no decay to consider at 120Hz. Ok, there is a decay even at these frequencies, but it produces so little light so that is neglible compared to the peak light effect.

LED lights that are driven straight from mains (a long string of LEDs with a series resistor) are the worst lighting armature you can think of. The leds turns completey off twice each mains cycle, and the dutycyle is very low.

There is an old saying; if you want a cheap and a good one, you have to buy two, one cheap one, and one good one... Its true even for LED lighting...

Christian

farss wrote on 12/9/2008, 2:44 PM

One though that did occur to me is that the light output of the cheap mains powered LEDs could well have a duty cycle of around 50%. LEDs have a signifcant forward voltage drop.
I have both iron ballasted fluro lights and cheap mains powered LEDs. I know the fluros produce really bad problems with CMOS cameras at 24p in 50Hz land. I haven't tried the LEDs as yet.

Bob.

blink3times wrote on 12/9/2008, 2:57 PM

"I haven't tried the LEDs as yet."

It's great so long as you use a reasonable power supply. And if you can.... use batteries... nothing produces a smoother dc current than a chemically generated one.

A simple car battery will run a rather HUGE number of led's for a surprising amount of time.

blink3times wrote on 12/9/2008, 3:15 PM

Christian:
You have a valid point when using PWM to control led's... but then other the other hand, controlling led's with PWM is not really necessary. Pulse width modulation is mostly a tactic used for varying the voltage in AC lighting. Granted it can be used in conjunction with trick led lighting.

Because led's can be (and mostly are) DC operated then by far the best (and cheapest) way to vary the brightness is by controlling the voltage/current. A simple led circuit with a battery and a varible resistor will emit a steady and variable brightness without involving the pulsing effect of a PWM

You CAN operate as you say a led from a AC source and a simple current limiting resistor... but ewhat you get is a whoppong pulsation as the led turns on/off, so it is not the norm... well... maybe if used as an indicator light or something of that nature.

farss wrote on 12/9/2008, 3:29 PM

What I was referring to was a cheap 225 LED light I bought off eBay. It became very popular with some at DVInfo as it was very cheap. It's actually an Aquarium Light. With some rework a set of them could make for a very cheap functional light for video. Color rendition would be no worse than any of the other cheap units around i.e. pretty bad but good enough for video. You might want to add a diffuser as the beam angle is very narrow.

If anyone wants a serious LED light then the Zylight is the ducks guts. These are now selling well despite the big price hike due to the falling AUD.

Bob.

PeterWright wrote on 12/9/2008, 3:31 PM

This may be totally different, but I was getting occasional "flicker" - more like rolling waves - on my EX1, even in daylight, as were several others, and this was eventually tracked down to using the camera's Flicker Reduction function when it wasn't necessary.

farss wrote on 12/9/2008, 4:29 PM

Make certain to switch it to Off.
If you reset the camera it goes back to Auto.

Bob.

earthrisers wrote on 12/9/2008, 5:57 PM

WOW -- lots of info!

Looks like a Brave New World of possible problems or not-problems...

Tomorrow evening we expect to have an opportunity to shoot during a rehearsal at another theater that has recently gone to LED backdrops.

We'll run whatever experiments we can (shutter speed, different cameras, etc.) at that time, and if there are any interesting results, I'll post them here.

johnmeyer wrote on 12/9/2008, 6:44 PM

You have a valid point when using PWM to control led's... but then other the other hand, controlling led's with PWM is not really necessary. Pulse width modulation is mostly a tactic used for varying the voltage in AC lighting. Granted it can be used in conjunction with trick led lighting.I may be misunderstanding what you are trying to say, but I can state with certainty that you CANNOT dim an LED by varying voltage, at least not to any great degree. Once the forward voltage drop has been exceeded (about two diode drops, depending on the LED, or about 1.4 volts), the darn thing turns on. Some sort of current limit (a resistor) keeps it from burning up as the voltage is increased (the current will otherwise go to "infinity" just by increasing the voltage a few tenths of a volt).

The ONLY way to make the eye see something other than full on or full off is to pulse the LED. Since an LED, unlike incandescent bulbs, can turn on and turn off in microseconds, it will definitely produce beat patterns that a camera might pick up, depending on the shutter speed of the camera.

The solution to all this is to find a lighting controller company that makes lights for studio use, and not stage use. On the stage, the audience will never see any of this. In a studio, it will be noticed right away. If you are stuck filming a stage event, then you''ll just have to experiment with shutter speed, as already noted. Usually the default slow shutter speed is what you want, but I suppose there could be exceptions.

blink3times wrote on 12/9/2008, 7:31 PM

You can vary the brightness of a LED if you change the current limit resistor.... although you are quite correct about the forward voltage. You will only get it so dim before it turns off completely since a led must be forward biased by at least 1.4 v. You also can not exceed the maximum current capability of the led.... otherwise you burn it up.

musicvid10 wrote on 12/9/2008, 7:39 PM

John is right.
Since the forward voltage drop in LEDs is low, using a potentiometer to vary current is not a very good way to control output (brightness). You can keep Vf high enough to open the diode, and low enough to prevent thermal runaway, but not much in between. Since current rises in an LED as the temperature rises, unlike conventional circuitry, fixed resistors are used. And, using potentiometers to control stage lighting is certainly not an option!

Now that I see Christian's post about PWM, I understand the flicker, but then I see LED stage lighting that is stated to run off of standard DMX dimmer packs, so I'm confused again about how AC is being handled at the can level. Do these lights have internal PWM conversion?

Of one thing I am certain -- I'm sorry I brought up the subject of output filtering affecting LEDs, since it seemed to at least one person here that I was making some sort of comparison between switching and conventional power supplies, which I was not; nor are such comparisons germaine to this thread, which is about commercial stage dimmers at the tens of Kw level (rather than wall warts or automobile electrical systems).

johnmeyer wrote on 12/9/2008, 10:54 PM

but then I see LED stage lighting that is stated to run off of standard DMX dimmer packs, so I'm confused again about how AC is being handled at the can level. Do these lights have internal PWM conversion?I just poked around a little, but couldn't find a definitive answer. Most sites recommend NOT using a conventional Triac/SCR light dimmer to control LED lights. However, in thinking about it, and in looking at a few sites that had technical information, I think this actually could work just fine. A typical triac dimmer varies the point in each 60 cycle (or 50 cycle) phase at which the the light bulb is turned on. The later in each half cycle, the dimmer the bulb becomes. Also, if you look at the waveform of the voltage across the bulb in a traditional dimmer circuit, it is in fact, pulse width modulation. There is no reason an LED couldn't work off this type of circuit. It wouldn't conduct on one half of each cycle (without rectification), but would be quite happy on the second half. As long as the PIV spec (peak inverse voltage) spec wasn't exceeded, it would be fine.

The problem with using a dimmer circuit, and why many LED sites recommend against it, appears to be the problem of having no current going through the LED (because of the LED forward voltage drop) at the point where the voltage crosses zero. The solution to most LED dimming problems encountered when trying to use traditional dimming circuits -- again this is from reading literature and not from practical experience -- seems to be to put a "dummy" incandescent load into the circuit. This can be any traditional small light and doesn't have to be very large.

I just purchased a bunch of LED lights to see if I could replace track lights and other high-wattage devices. I do not approve at all of compact fluorescents because they are another idiotic "solution" being forced down our throats by really stupid government officials who have not thought through the "end-to-end" environmental implications or the actual economies involved in normal duty cycle situations (they wear out fast when turned on and off a lot). LED lights, by contrast, are potentially much friendlier to the environment, not only because of the materials used, but also because they should have MUCH longer life than anything else now being used. Unfortunately, the light output of the samples I have is even less than most CF bulbs, although the light quality can be pretty decent.

Anyway, dimming in a consumer environment, where you cannot educate the consumer to put a "dummy" load on the circuit is out of the question, but for stage lighting, where professionals are in charge, I think this could easily be made to work.

P.S. I just was closing down some of the browser windows that I used for my quick research and it occured to me that if the stage LEDs are in fact being powered by regular dimmers, then since the DO increase and decrease in brightness (somewhat) as the voltage applied to the LED/resistor combo changes, then during each half-cycle (every 1/120 of a second) the LED is going to get brighter and then dimmer, and without rectification is going to be off completely during the second half of the power cycle. This could create a LOT of flicker which the naked eye would never see, but would be quite apparent to a camera. In fact, if you put the camcorder shutter way up to 1/250 or faster, you could end up with some fascinating effects.