A wonky LED question for manufacturers

JChenault

Well-Known Member
One of the most irksome issues ( at least to my eye) is the bump in led fixtures at the start of the fade. Most fixtures that I have been able to play with have a noticeable bump as they move from 'Off' to 'Barely On'.

Many years ago at ( I believe) a USITT session, I saw a simple set up which consisted of an LED bulb, battery, and potentiometer. When the potentiometer was turned so there was no voltage the LED was off. As you turned it up, the LED ramped up with no bump at the low level.

My understanding is dimming for theatrical fixtures is done by a switching power supply that switches the full voltage driving the LED on and off very quickly. For a low level you leave in 'on' for a much shorter time slice than a higher level.

My question. If varying the voltage ( not slicing it up ) does give you a smoother curve - why don't the manufacturers build power supplies and electronics that vary the voltage, instead of switching it on and off very quickly?

I suspect there is a good reason for this design approach ( or my understanding is flawed) - I'm just curious for the rationale.

Thanks
 
One of the most irksome issues ( at least to my eye) is the bump in led fixtures at the start of the fade. Most fixtures that I have been able to play with have a noticeable bump as they move from 'Off' to 'Barely On'.

Many years ago at ( I believe) a USITT session, I saw a simple set up which consisted of an LED bulb, battery, and potentiometer. When the potentiometer was turned so there was no voltage the LED was off. As you turned it up, the LED ramped up with no bump at the low level.

My understanding is dimming for theatrical fixtures is done by a switching power supply that switches the full voltage driving the LED on and off very quickly. For a low level you leave it 'on' for a much shorter time slice than a higher level.

My question: If varying the voltage ( not slicing it up ) does give you a smoother curve - why don't the manufacturers build power supplies and electronics that vary the voltage, instead of switching it on and off very quickly?

I suspect there is a good reason for this design approach ( or my understanding is flawed) - I'm just curious for the rationale.

Thanks
@JChenault From sort of the same mindset, this now blind dude has a little off the shelf / mass produced photocell activated LED nightlight plugged in a standard duplex receptacle in one corner of the seniors' facility in which he now resides. During daylight hours sufficient light enters via my loan window from the north to keep my LED extinguished. Being the curious geezer that I've always been, I periodically amuse myself by bringing the shadow of my palm slowly closer and closer to the nightlight's photocell and ALWAYS being amazed by how smoothly my little Canadian $7.00 LED nightlight manages to gently ramp up from totally extinguished to full brilliance. What does the red Chinese army know about mass producing LED nightlights at bargain prices that apparently eludes the rest of us? Granted we're not talking about anything equivalent to a 1K 3200 Kelvin FEL but then I keep reading on Control Booth about the exact problem you're posting about. This still doesn't create the glorious amber shift in color temperature we grew up with using our beloved incandescents but, like you're pointing out, it certainly doesn't appear too difficult or too spendy to achieve a smooth transition from totally extinguished to full output AND back. I too agree and am curious to learn the rationale.
Toodleoo!
Ron Hebbard
 
Many years ago at ( I believe) a USITT session, I saw a simple set up which consisted of an LED bulb, battery, and potentiometer. When the potentiometer was turned so there was no voltage the LED was off. As you turned it up, the LED ramped up with no bump at the low level.

You may be right but i thought diodes were digital devices and you describe an analog function. Since even tungsten dimmers are switching on snd off, I guess your suggesting a sine wave dimmer.

I do find some LED units do pretty well at low end but not many.

Good topic and I look forward to more knowledgeable people commenting.
 
Problem is, there is very little difference between the point at which the diode turns on and the point it is at full brightness. (about 1/2 of a volt) As such, what works on a small scale with a low power diode does not work too well when the scale is increased. Chopping is the best way, but to get a really fine chop, you need a very high frequency waveform. Let's say you are driving the diode at 40Khz and you want to be able to allow only 1/100,000th of the waveform through- Although the distance between chops may still be at 40Khz, the actual duration of the chop gives it a MUCH higher frequency characteristic. Add in the fact that it is a square wave chop (Square wave = sine wave with every possible harmonic added), you suddenly have a fixture that is very hard to quiet down to the point where the FCC is happy. Because of this, we end up with design compromises that leave the area under 1% a mess. One thought would be to switch from chop to chop and voltage in the 0 to 1% region.
 
Every design is born of a series of engineering trade offs. Apparently, smooth dimming at the low end is one of the compromises they have to make to balance all of the other factors. LED lights are built that way for the same reasons switching power supplies are widely used in practically all electronics now: weight, size, heat, cost, and efficiency. A supply that can vary the current smoothly down to zero tends toward being big, hot, and heavy. An expensive LED light with a huge heat sink or noisy fan isn't going to sell.
 
Part of it is design cost and efficiency.

That $7 light is directly using the photoresistor to limit the amount of current flowing (perhaps coupled with a small transistor)

The thratrical fixture on the other hand, is likely to be using Pulse width modulation, which makes the lower end of dimming a bit more difficult as far as I'm aware.

A fixture using transistors directly to control the light output at various gate levels effectively becomes that old Class B amplifier you've got lying around, which means you've just added a 50 pound hunk of metal to your design, and making the power delivery waste just as much, if not more, power than the LED itself.

It's worth noting that PWM also uses transistors, but these are optimized for high frequency switching from fully on to fully off, rather than a smooth transition between high and low, it then uses capacitors and inductors to smooth that ON / OFF pulse into a stable voltage.

As you scale the PWM down to very low duty cycle values, there's a couple things that start going wrong.

As far as i'm aware, as the pulses become shorter and further apart, you eventually have insufficient current to keep the inductor, well, inducting, and the smoothing is no longer able to hold a stable voltage.

There are ways around this, but they increase cost and / or decrease efficiency. A higher PWM frequency will enable much smoother and precise voltage control, at the expense of system efficiency.
 
The ones I have taken apart don't use much in the way of filtering between the PWM and the LED. What is generally there is more of an RF filter (small cap, small choke) so that the fixture doesn't broadcast. Hard to put any RF shielding in front of the LED ;)
As long as the frequency is high enough there is not much of a chance of flicker, even on film. (unlike domestic LED "replacement" lamps) Current is the key since junction drop voltage really can't vary. PWM enables the drive transistors to be very small and not dissipate much heat as they are not operating on the analog slope, just full on or full off. I played around driving one at 1 Mhz once and got fantastic low end. Didn't keep it on long as I feared the FCC would come knocking!
 
Is there a reason (beyond the obvious $$) that mechanical dimming shutters couldn't be employed on LED's?
 
Is there a reason (beyond the obvious $$) that mechanical dimming shutters couldn't be employed on LED's?
On top of @EdSavoie's remark, it's also mechanical. Part of the beauty of LED fixtures is their relatively low need for maintenance (just clean the lenses, and maybe clean the cooling system).
Shutters on a moving LED fixture make a little more sense compared to on a fixed unit, considering there's already plenty of moving parts in a mover, but it still adds room for mechanical failure.
From the cost stand point, you could have shutters with a relay to turn the LED off when the unit is closed, but there's a lot of parts. Cut shutters, at least two motors, more control circuitry, and increasing fixture size to accommodate the shutters.
 
I think manufacturers have to rethink what they are doing with dimming below 10%. Besides changing the nature of PWM, one real obvious fact come to mind:
LED fixtures (and all current sources in general) are "open loop" systems. Current is simply sent to the LED expecting it to act one certain way. In mechanics, we close the loop by using servo, a feedback system where the controller gets information and then corrects for error.
The same could be done with LEDs by mixing a photo-transistor or two in the emitter array. This would also have the benefit of making fixtures more consistent when LED batches are changed in production. The LED gets very argumentative at low levels. Perhaps a little internal guidance is needed.
 
Some manufacturer did that, ostensibly to match color. Prism is the name that comes to mind. Havent heard of them for a few years.
 
You may be right but i thought diodes were digital devices and you describe an analog function. Since even tungsten dimmers are switching on and off, I guess your suggesting a sine wave dimmer.

I do find some LED units do pretty well at low end but not many.

Good topic and I look forward to more knowledgeable people commenting.
"Digital" isn't the word you want - "non-linear" is. With a resistor, including tungsten, the current will vary proportionally with the voltage across that load (for a fixed load temperature).

With LEDs, there's a pretty sharp knee where the current will increase a greater and greater amount for a small increase in voltage. In other words, the range for an LED being on and full on, with a DC source, is very small.

So, by just varying the voltage you can't control the output - you need to also control the current. You need a resistor, but the smallest one you can get away with because that resistor going to generated wasted heat.
 
Adding a feedback system would be good, but using a linear element such as a phototransistor as a sense element brings problems due to the difference in response to phototransistors, even from the same batch, so some calibration needs to be done.

If you want smooth dimming, you need 16 bit dimming or two dmx channels so there are more steps available as the current pulses are closer together. Some manufacturers use a current coprocessor (my term) similar to the maths coprocessor in early computer days, to add extra steps at the bottom end of the dimming. If you look at the specs of some led fixtures, they state 64 bit dimming, but do not state 8 dmx channels for dimming.

This https://www.blizzardlighting.com/products/toughpar-quadra ToughPAR Quadra led has 32 bit dimming for 3 channel RGB dmx mode.
 
Marketing guy: Hey you, engineer - aren't you using a 32-bit timer inside the microprocessor for the LED dimming?
Engineer: Yeah, that's pretty much been common state of the art for the last decade.
Marketing guy: OK, so I can write 32 bit dimming in our advertising?
Engineer: Well... yeah, sure. Whatever.

In my experience it usually goes a bit differently ;)

If you look at the specs of some led fixtures, they state 64 bit dimming, but do not state 8 dmx channels for dimming.

It could be that they are interpolating dimming between 8 bit values with 32bit precision over some amount of time, which would explain why some LED fixtures can be configured in modes that don't snap immediately to off when you type "Fixture 10 @ 0 Time 0 Enter"
 

Users who are viewing this thread

Back