Human Eye -- Peak or Average Response

[edavid]

I don't know why people are messing with you.   The human eye has very close to average response, over a wide range of conditions.  So don't worry about DC vs. pulsed drive, it's not going to make a significant difference.

[Someone]

I've got a white landing light of about twenty watts (DC power).  I shine the light at the ground when I'm about 350' AGL.  It lights a rough ellipse on the ground and lets me see whether or not there are animals on the runway before I collide with them.

I run the light on DC and get a particular illumination.  If I run the same exact device on AC of any frequency you want above the flicker level and I adjust the current so that the pulse width times the current gives me the same power into the light, will the light appear brighter to me as reflected from the SAME ground at the SAME altitude at the SAME attitude, at the SAME ....   In other words, the only differences is pulsed power versus DC.  Does the pulse width affect the result (right up to the maximum pulse current of the device)?  Is it any brighter at any combination of pulse width and current than at DC?

THat should take care of any test conditions, I hope.

It really doesn't provide enough information. The "enhancement" reported by various groups that you are chasing is highly dependent on the conditions under which the observer is viewing the test, and the absolute luminance observed and spectral characteristics etc.

Oh, dear God.

I  can buy 50 watt LEDs if I can figure out how to heat sink them.

Your stuff about moving light sources and propeller sync have NOTHING to do with the question.  The LED is WELL outboard of the propeller.  PLEASE do NOT introduce extraneous problems as there are none.

You want a clean airframe?  Same problem; consider a glider.

You are missing the basic understanding of the problem to attempt chasing the last few percent. A pulsed source will cause artefacts to vision in that sort of application (looking through a propellor or not) unless you are pulsing way up into 10's of kHz or higher, I can not put a figure on the exact point because the information is missing. Trying to shoehorn LEDs into this application is almost certain to fail unless you have the background in optics which you haven't demonstrated.

[Someone]

I don't know why people are messing with you.   The human eye has very close to average response, over a wide range of conditions.  So don't worry about DC vs. pulsed drive, it's not going to make a significant difference.

It can make a big difference practically, or completely disorient the viewer if done poorly. But there are researchers who like to get publications by claiming significant improvements in apparent brightness (under very carefully controlled situations) and uneducated people try and extend that to making their lamps brighter (which will almost always never work and end up with a poorer result).

[Berni]

Sorry, I though it obvious that I would have to get over the 24 Hz. flicker frequency to avoid the persistence of vision obstacle.  Let's say that you have your choice of any frequency from 25 Hz. to daylight.

Viewing angle is dead on boresight.

Empirical testing to follow.  It is a trivial experiment and one that I probably should have done instead of asking the question if anybody has done the experiment and how did it turn out?

Let me pose the question for what I **REALLY** want to know.

I've got a white landing light of about twenty watts (DC power).  I shine the light at the ground when I'm about 350' AGL.  It lights a rough ellipse on the ground and lets me see whether or not there are animals on the runway before I collide with them.

I run the light on DC and get a particular illumination.  If I run the same exact device on AC of any frequency you want above the flicker level and I adjust the current so that the pulse width times the current gives me the same power into the light, will the light appear brighter to me as reflected from the SAME ground at the SAME altitude at the SAME attitude, at the SAME ....   In other words, the only differences is pulsed power versus DC.  Does the pulse width affect the result (right up to the maximum pulse current of the device)?  Is it any brighter at any combination of pulse width and current than at DC?

THat should take care of any test conditions, I hope.

Jim

If your aim is to have a brighter landing light then the answer is no. Any potential difference is going to be tiny while causing more problems.

At a low frequency of 24 Hz the only thing that will be significantly noticeable is that everything will have that 24fps movie look to it. As in any fast moving object not appearing to move completely smoothly but being able to discern the "frames". There is nothing bad in making your landing look like a film movie, but the low "framerate" could impact your ability to reliably track moving objects and increase your reaction time. When playing video games having the game run at lower than 30 fps has a very noticeable effect in that you don't see the result of your control input quite as quickly and leads to the game feeling less responsive and clumsier.

While true that 24fps is the framerate where our eyes start seeing a sequence of images as continuous motion, but 24fps is not as fast as our eyes can see. We can typically see up to about 50Hz and this is why CRT monitors always scan at 50Hz or more.

You can get around that by PWMing it at 100Hz, but you still wouldn't get much of a brightness difference, so the most noticeable effect would just be flickering trails behind very fast moving objects, or perhaps a bit of that "DLP projector" feeling if you move your head quickly.

[Nusa]

If you are actually talking about LED landing lights (presumably aircraft certified), and they accept both AC and DC, that just means there is rectification going on inside the lighting unit to convert any AC sine wave to DC before it gets to the actual LED(s). I assure you that LED's are not AC devices. Light Emitting Diode.

[tooki]

24 Hz may give you persistence of vision, but flicker is easy perceived by just about everyone. Still, even that was pretty amazing to people early on in the moving pictures industry. These days it's upsampled for display, unless you want the old-time effect on purpose.

The 24fps film “look” is from being able to use shutter speeds as slow as 1/24 sec, allowing a small amount of motion blur.

But indeed, 24fps is flickery, which is why cinema film projectors actually project each frame twice or three times before advancing the film, producing an effective “display” frame rate of 48 or 72fps. (Digital cinema projectors use higher frame rates still, AFAIK.)

At a low frequency of 24 Hz the only thing that will be significantly noticeable is that everything will have that 24fps movie look to it. As in any fast moving object not appearing to move completely smoothly but being able to discern the "frames". There is nothing bad in making your landing look like a film movie, but the low "framerate" could impact your ability to reliably track moving objects and increase your reaction time. When playing video games having the game run at lower than 30 fps has a very noticeable effect in that you don't see the result of your control input quite as quickly and leads to the game feeling less responsive and clumsier.

While true that 24fps is the framerate where our eyes start seeing a sequence of images as continuous motion, but 24fps is not as fast as our eyes can see. We can typically see up to about 50Hz and this is why CRT monitors always scan at 50Hz or more.

You can get around that by PWMing it at 100Hz, but you still wouldn't get much of a brightness difference, so the most noticeable effect would just be flickering trails behind very fast moving objects, or perhaps a bit of that "DLP projector" feeling if you move your head quickly.

You wouldn’t get the “24fps look”, you’d get annoying flicker instead.

As someone who is sensitive to flicker, the 100Hz flicker of LED lighting dimmed from 50Hz mains is super annoying and distracting. For aviation this would be outright dangerous for me.

As I’ve discussed in other threads on the forum, to fully eliminate ALL flicker artifacts in ALL conditions requires a PWM frequency of about 30KHz, at which point a non-PWM constant-current LED driver makes more sense, generally. (The “flicker fusion threshold” of 60Hz that’s usually accepted... well, it is true, but only for the comparatively slow, high-resolution center of the eye, and only when everything — head, eyes, and object being viewed — are perfectly still. Our low-resolution peripheral vision is much more sensitive to flicker, and as soon as anything is in motion, we need much higher PWM frequencies to prevent it breaking up into dots of light, known as the “phantom array” effect.)

[jweir43]

If you are actually talking about LED landing lights (presumably aircraft certified), and they accept both AC and DC, that just means there is rectification going on inside the lighting unit to convert any AC sine wave to DC before it gets to the actual LED(s). I assure you that LED's are not AC devices. Light Emitting Diode.

You are kidding, right?

Jim

[Nusa]

If you are actually talking about LED landing lights (presumably aircraft certified), and they accept both AC and DC, that just means there is rectification going on inside the lighting unit to convert any AC sine wave to DC before it gets to the actual LED(s). I assure you that LED's are not AC devices. Light Emitting Diode.

You are kidding, right?

Jim

Which part do you think I'm kidding about?
Certification requirements for aircraft parts? http://www.faa-aircraft-certification.com/certification-of-aircraft-components.html  (And virtually everything requires a logbook entry!)
That LED's are diodes that emit light? I'll let you google that one.

[Zero999]

I was taught that a fast pulse or pulses of light for durations less than 30ms appear to be dimmed. I am not talking about a slow incandescent light bulb.

You were taught wrong.

Jim

No, he is right. The human eye acts like a low pass filter. Short pulses will appear dimmer, than long ones. Awhile ago, I did an experiment  which involved putting short pulses, of hundreds of Amps through high powered LEDs. The flashes appeared to be dim, but would have been bright enough to cause eye damage, had they been continuous, assuming the LED would have survived, which of course it wouldn't..

[tooki]

Indeed, if our persistence of vision didn't perform a LPF effect, PWM dimming wouldn't work.

[Benta]

Sorry, I though it obvious that I would have to get over the 24 Hz. flicker frequency to avoid the persistence of vision obstacle.  Let's say that you have your choice of any frequency from 25 Hz. to daylight.

Forget the 24 fps when talking lighting. The 100/120 Hz flicker from fluorescent lamps is absolutely perceptible in your peripheral vision, and there have been threads here on this forum with users who have problems with this in the direct vision field.
Absolute minimum is 200 Hz, and for a design I'd go for 500 Hz and above. Especially for the application you've mentioned.

[bsdphk]

There is some text about this related to the the LED display in the HP35 somewhere, I remember the text being "non-stiff" so it is probably in one of the oral histories, but check also the HPJ issue.

The gist of it is that they saved battery-life by driving the LEDs with microsecond-width high-current pulses, based on the observation that it gave same visual response but used a lot less power.

That must presume some kind of peak-ish response-function to make sense.

[TheUnnamedNewbie]

There is some text about this related to the the LED display in the HP35 somewhere, I remember the text being "non-stiff" so it is probably in one of the oral histories, but check also the HPJ issue.

The gist of it is that they saved battery-life by driving the LEDs with microsecond-width high-current pulses, based on the observation that it gave same visual response but used a lot less power.

That must presume some kind of peak-ish response-function to make sense.

Likely also just the fact that driving with hard switching can be more efficient, both space-wise (on a chip) and energy-wise. Linear dimming of an LED requires you (unless you do fancy DC/DC converter stuff) to waste the excess power somewhere else (be it a resistor or a transistor). Switching an LED hard allows you to use more current (smaller resistor, if any resistor is used at all), for a shorter time, resulting is less wasted heat.

Another strange thing with PWM at low-ish frequencies is that you might not notice when it is still, but really can notice when it moves. This is, after all, what those rotating-stick-of-LED displays are based on. Easy way to notice: Just have an LED PWM at a few hundred Hz and then shake it violently, and now notice how you see dots instead of a line of light.



I have also been told that you often want to drive an LED for lighting (totally different field than what is discussed here) in PWM instead of constant-current for CRI reasons. The phosphors are non-linear in their response, and so will change color index slightly as you reduce current. In order to avoid this you want to drive the LED hard always, and just change dutycycle.

I was taught that a fast pulse or pulses of light for durations less than 30ms appear to be dimmed. I am not talking about a slow incandescent light bulb.

You were taught wrong.

Jim

No, he is right. The human eye acts like a low pass filter. Short pulses will appear dimmer, than long ones. Awhile ago, I did an experiment  which involved putting short pulses, of hundreds of Amps through high powered LEDs. The flashes appeared to be dim, but would have been bright enough to cause eye damage, had they been continuous, assuming the LED would have survived, which of course it wouldn't..

How much of them being less dim was simply because of the inductance in the leads and the capacitance in the LED, causing the actual current through the LED to be substantially lower? Not to mention, if you really start putting that much current in the LED you are going to significantly alter the bandgap, which will change color and efficiency both.

[Zero999]

There is some text about this related to the the LED display in the HP35 somewhere, I remember the text being "non-stiff" so it is probably in one of the oral histories, but check also the HPJ issue.

The gist of it is that they saved battery-life by driving the LEDs with microsecond-width high-current pulses, based on the observation that it gave same visual response but used a lot less power.

That must presume some kind of peak-ish response-function to make sense.

Likely also just the fact that driving with hard switching can be more efficient, both space-wise (on a chip) and energy-wise. Linear dimming of an LED requires you (unless you do fancy DC/DC converter stuff) to waste the excess power somewhere else (be it a resistor or a transistor). Switching an LED hard allows you to use more current (smaller resistor, if any resistor is used at all), for a shorter time, resulting is less wasted heat.

If the LED is being powered from a resistor or linear regulator, then PWM makes absolutely no difference to the efficiency. The main reason for PWM in LED displays is it's easier to implement with digital outputs, than a linear solution.

Another strange thing with PWM at low-ish frequencies is that you might not notice when it is still, but really can notice when it moves. This is, after all, what those rotating-stick-of-LED displays are based on. Easy way to notice: Just have an LED PWM at a few hundred Hz and then shake it violently, and now notice how you see dots instead of a line of light.

Yes, that's another reason to keep the PWM frequency high.

I have also been told that you often want to drive an LED for lighting (totally different field than what is discussed here) in PWM instead of constant-current for CRI reasons. The phosphors are non-linear in their response, and so will change color index slightly as you reduce current. In order to avoid this you want to drive the LED hard always, and just change dutycycle.

I think it depends on the LED. Green LEDs definitely do shift towards the blue end of the spectrum, especially when overdriven. I agree, it's better to keep the forward current to that recommended by the manufacturer to keep the colour and wavelength within specification.

I was taught that a fast pulse or pulses of light for durations less than 30ms appear to be dimmed. I am not talking about a slow incandescent light bulb.

You were taught wrong.

Jim

No, he is right. The human eye acts like a low pass filter. Short pulses will appear dimmer, than long ones. Awhile ago, I did an experiment  which involved putting short pulses, of hundreds of Amps through high powered LEDs. The flashes appeared to be dim, but would have been bright enough to cause eye damage, had they been continuous, assuming the LED would have survived, which of course it wouldn't..

How much of them being less dim was simply because of the inductance in the leads and the capacitance in the LED, causing the actual current through the LED to be substantially lower? Not to mention, if you really start putting that much current in the LED you are going to significantly alter the bandgap, which will change color and efficiency both.

The capacitance and inductance of the LEDs and wires are not a problem, as long as the layout is good enough. The current was measured using a current transformer and forward voltage monitored.

LED efficiency does drop, at higher currents, but this was taken into account, as the light output from the LED was measured using a photodiode. The purpose of the experiment was to see how much an LED can be overdriven and the reduction in efficiency, rather than the response of the human eye, but it was noted that longer pulses appear proportionally brighter, than short ones. For short pulses, the brightness is dependant on the energy, rather than the peak power or length, so a 2µs pulse will a appear to be the same brightness as a 1µs pulse, of double the peak intensity.

[Siwastaja]

24Hz flickers massively, or almost "blinks". No one can stand it for long.

Cinematic projection has never worked at 24Hz. Depending on the actual projector shutter type used, they either doubled or tripled the film frame rate, so it's either 48Hz or 72Hz. TV was 50-60Hz depending on country, but CRT has some persistence (gradual fade during off-time) that LED doesn't have.

So now we are looking at a complex problem, which we can simplify:
1) After around 100Hz (there is some room for arguing), the eye integrates the light perfectly, and only the average matters,
2) Between about, say, 20-100Hz, it's a really fuzzy "depends who you ask" type of problem. Different people perceive quickly flickering lights in a different way.
3) Below about 20Hz, you have a clearly blinking light, and of course it appears brighter during the on-time, as your eyes recognize the on-time as a separate "event".

For 1), the only thing that matters is the light output vs. input current curve for the particular LED. Often, for modern LED types, this curve is quite linear for a big part. For a modern standard-power LED specified for 40mA absolute maximum, you would likely have the linear range somewhere between about 2-20 mA. If this is your case, then only the average current matters, and PWMing makes no difference. OTOH, if your PWM "on" current would be at the absolute maximum 40mA, the LED likely runs at lower efficiency, lowering the average light output for same average input current. But all this speculation is meaningless: you need to look at the curve of the actual LED you are using, and you have your answer!

[tooki]

Another strange thing with PWM at low-ish frequencies is that you might not notice when it is still, but really can notice when it moves. This is, after all, what those rotating-stick-of-LED displays are based on. Easy way to notice: Just have an LED PWM at a few hundred Hz and then shake it violently, and now notice how you see dots instead of a line of light.

That’s the phantom array effect I mentioned earlier in the thread.

[bsdphk]

Ok, found the HP35 reference, and it does not argue for peak sensitivity:  http://hpmemoryproject.org/timeline/dave_cochran/a_quarter_century_at_hp_00.htm

"We decided to use an inductive driver for the LEDs to store energy and to get the high currents. We had heard that super-linearity would get more light output than just proportional. I talked to the LED guys at HP Associates, "What happens if you pulse these? Can I pulse them with 1,000 amps?" "Oh, no, 1,000? No, no, no." No, we're talking about milliamps." "Oh, well, okay." The average current on the LED segments was something like three or four milliamps. And I said, "Well, suppose we hit them with 30 milliamps, but 10 percent duty cycle?" "Oh, well, that might work." We actually ended up with a one percent duty cycle. Got about three times the light output than we would have if we'd just applied the three milliamps. We put 300 milliamps through it, at one percent duty cycle. Got three times the light output; now that's a freebie."

"Somebody asked about reliability. I set up an array of constant current three milliamp displays and compared them to in a set at a 0.1 percent duty cycle, ten times more than our design. I ran those for six months. And then I looked at the comparable light output over time. They were the same. […]"

HP Application Brief D-004 about LEDs (5963-7073) explicitly say they eye averages when using multiplexing.

[Circlotron]

In the days before high power white LEDs I often wondered if you could use a xenon strobe lamp to make a decent spotlight and have the eye perceive the peak brightness. The average would have been nothing special.

[Dubbie]

I’ve studied this area quite a lot in relation to my career and from my point of view, Tooki and Siwastaja have explained everything spot on.