UKC

Hi,
A quick series of questions
Am I correct that LED head torches flicker the light?
If so, is this flicker simply to do with the way an LED functions, or is it there to save energy?
Do mains LEDs found in the home do the same?

I'm trying to find the efficiency of a modern head torch and of common home LEDs. I know the max will be 683 lm/W, and in practice 30%, but I can't find out if this is assuming no flicker. If the lamp was off half the time, but the eye didn't spot it, might this give an equivalent of 60%. If so, is there a word for luminous efficiency that takes this into account.

Thanks.

In reply to David Coley:

I don't know about headtorches specifically, but this is not a feature of LEDs, they emit light at a constant brightness so long as a voltage is applied. To 'flicker' the light would entail more circuitry, and potentially shorten the life of the LED through wear and tear so it seems unlikely.

There has been some work done on LEDs that operate on AC without needing a DC converter, but even here only part of the LED is dark during the off-cycle, so it still emits continuous light.

In reply to David Coley:

Yes they're using Pulse Width Modulation (PWM) to dim the LEDs. This isn't to save energy it's so that they can have different intensity settings.

House LEDs won't do that unless they're dimmable ones.

You get the same effect on some cars rear lights. Passat I think uses the same LEDs for rear lights and brakes. When you look across them and away you get a trail in your eyes. It's a weird and quite distracting effect.

In reply to David Coley:

If the LED's are powered by a DC supply like a battery then there shouldn't be any flicker.
Its not a physics phenomenon that I know of.

The circuit could be set up so that they are on a duty cycle in order to increase battery life being as the human eye cannot detect refresh rates of over roughly 60 fps (very roughly).

No sure if this is useful



TS

In reply to planetmarshall:


I don't know owt about LEDs, or physics generally, but my anecdotal experience is that my LED headtorches do seem to have a slight strobe effect compared to my old halogen one - waving your hand side to side in front of your face, or shining the beam on running water should demonstrate the effect. Something to do with wavelengths, or the eye's effective 'shutter speed', I assumed

In reply to DancingOnRock:

Does your answer mean that with some headtorches, that reducing the intensity is not going to extend battery life?

In reply to nufkin:


Yes, but this is a stroboscopic effect caused by Pulse Width Modulation, as mentioned above. It's not intrinsic to the way LEDs operate, but a side effect of the fact that the LED controller is digital ( it has two states, on or off ), so the only way to 'dim' the LED is to oscillate the power signal causing the LED to flicker. Normally this would not be perceptible to the human eye, but if you wave your hand in front of the light you cause an aliasing effect which makes the flicker visible.

In reply to veteye:


No. It will extend battery life.

If the LED is off there's less current flowing, the current isn't diverted into another component.

Theoretically 50% intensity should double the battery life.

Usually there's an info graph on the packet. Showing intensity v battery life.

In reply to planetmarshall:

This depends on your persistence of vision. I find some headtorches completely unusable because of it. Same for cheap home LED lighting which tends to pulse at 50Hz (because of the cheap rectifiers they use). And I find the 50Hz pulsed car tail lights very unpleasant to look at - I think this is a potential safety issue and there should be a legal minimum frequency of at least 100Hz.

In reply to David Coley:

By flicker, I take it you mean what is termed 'pulse width modulation' [PWM], rather than the strobe function?

PWM is used to lower the light output of an LED so that the amount of light emitted, is lower than the light emitted when the LED is run at the cutoff voltage on DC.

According to this forum post - http://www.candlepowerforums.com/vb/showthread.php?380962-New-Petzl-Tikkina...

- Petzl headtorches do use some PWM in at least some of their headtorches. PWM is not necessary as a design decision - you could have say 10 LEDs, and have either 1, 2, 3, ..., 8, 9 of them on to have 9 power levels.
e.g. theoretically this one
http://www.petzl.com/en/Sport/SPECIALIZED-headlamps/ULTRA-RUSH?l=INT#.Vo5Qs...

Not sure what you mean by mains LEDs?

If you were running a cycle whereby 50% of the time, the LED is off, 50% it's on - then the power used is approximately 50% of the power used compared to running the LED on DC. The light output is also approximately 50% as well - thus your ratio of lm/W is approximately the same, no matter the amount of light a PWM'd LED is emitting.
It's not _quite_ this simple, as PWM incurs a "switching loss" that is proportional to the switching frequency, and related to the duty cycle. Also many headtorches have "constant lighting" technology, which would also probably incur a slight loss. As for the numbers of these - I don't know, it depends is really the answer - but it's probably quite small, as headtorches do not heat up during operation, the sign of appreciable losses.

In reply to David Coley:
Ok, so a bit of my old university knowledge here...

I don't know how LED headtorches function in terms of flickering...

If you pulse an LED on then off, with equal time between peaks, you have essentially halved the power requirement, this is commonly used in power saving applications, but there is something more special you can do...

It is common practise to pulse an LED with a higher current than it's rated to achieve a brighter output. Driving an LED at a higher current continuosly (eg DC) will cause it to overheat, and eventually burn out. Pulsing it with a current at double the rated, will use the same power, however each time the LED is on, it will be brighter. If it is pulsed quick enough the human eye will not notice the gaps, and will be tricked into seeing a much brighter light for what seems like a continuous period. This is slightly simplified, but it is something that is done, and can work very well. Whether it is used in head torches, I have no idea, it could very well be used, and if done properly would not have any detrimental effects on the LED.

In reply to DancingOnRock:

Thank you. I wondered if there was a divert into a storage component, or if the "wastage" was just directed to heat, but with the PWM the current actually stops(or is it just drastically reduced?) so no not much inefficiency.

In reply to BStar:


I don't think it's quite that simple - otherwise you could not dim a PWM'd LED (same current, same voltage) slightly near the top end (which you can). The eye perceives the average luminosity, not the maximum.
Also LEDs are less efficient at higher current, so there is a lot more wasted power this way.

In reply to Neil Williams:

The recommended minimum frequency is 1kHz.

In reply to SenzuBean:

Like I said, I simplified my post...
Imagine if you will a LED pointing down a tap measure in a dark room. There would be a limit to the furthest point your eye could see.
Now increase the power of the LED, you would be able to see further.
PWM with double the current for on a 50/50 cycle at a few hundred Hz; for each time that the light is on, the new maximum distance on the tape measure would be visible to the human eye.

Like I said, I am not sure if this is used in headtorches, it may well be, it might not be.

In reply to SenzuBean:

Having done a bit of googling, I can see there is an open debate on this topic, some agreeing, some disagreeing. From my Uni days and the experiments I conducted with IR LEDs I would agree, but I can see both sides of this debate re Max vs Average

In reply to David Coley:

As an aside this use of a reduced "duty cycle" to power LEDs goes back a very long way. IIRC Sir Clive Sinclair was one of the first people to use this idea with LEDs in consumer products with his early calculators. Without this feature battery life would have been prohibitively short. It was used in digital watches as well - horrible things that required you to have to press a button to illuminate the display - another battery saving feature.

The technique has also been used extensively sice the early 70s in communications technology - walkie talkies were using logic circuitry to switch receiver circuitry on and off while the equipment was on listening watch, the user being unaware of this.

The other clue to more complex circuitry in the better headtorches is in the EMC statements, warning, for instance , of possible interference to avalanche transceivers when in close proximity.

Switching circuitry can also be used to extend battery life a little by converting the voltage of a battery to a useful level when the battery itself has not the direct voltage to run the LEDs, albeit at thee expense of greater current draw as battery voltage falls

In reply to BStar:


There's some interesting stuff I read while refreshing about current controlled LEDs being far more efficient (like 20-30% more) than PWMd ones in certain ranges of dimming. And even more interesting ideas about using a hybrid of current control and PWM. Not sure if they do it in practice though.

In reply to David Coley:

Thanks everyone. looks like we are half way there.

So head torches use Pulse Width Modulation to allow them to be dimmable. But the question now is to what degree using a 50/50 on/off time for example although dropping the lumens by 50% tricks the eye into seeing objects as though the light output was still continuous at 100%, or possibly some other level, but higher than 50%

In reply to DancingOnRock:
It seems clear that very few car tail lamps are using a frequency that high. The (expensive, branded Philips, £9 a go) LED GU10s I have in my house have barely any perceivable flicker at all, you can only see it if you move your hand very fast. I fear the car manufacturers are cutting corners, with a possible safety risk due to distraction - but one that not everyone can perceive.

I think this, for road use, needs to become mandatory as an MoT criterion, including retrospectively for any that have been fitted aftermarket. And soon, before any of the bad ones get too widespread as manufacturer fit.

FWIW Samsung phones also use too low a frequency - I cannot comfortably use one. Same with quite a few brands of laptop. Which doesn't bother me, as I can spend my money elsewhere (and I also dislike their UI "enhancements", preferring stock Android or iOS) but this shows just how widespread the issue is. But in none of those cases does your choice of phone or laptop cause me an issue - but another person's choice of car, or aftermarket tail lamp, may well cause me one.

Neil

In reply to Neil Williams:


Amen. I always assumed cheap LEDs used a full bridge rectifier and strobed at 100Hz. Will have to measure it.

Low frequency strobing car lights are one of my biggest peeves, although static but to bright LED brake lights are fast catching up (especially when sat behind a berk in traffic at night with their foot planted on the brake pedal) and over-bright dipped headlights.

The entire issue of car lighting needs a fresh legislative perspective rooted in sensible units (output power) and perception (visible strobing)

In reply to DancingOnRock:

You can use PWM to save energy in LEDs. You need to apply a certain voltage for the LED to emitt light (cant remember why look on wiki). The voltage bettween this and max operation will give you the change in light levels. There will be an explantion somewhere on the internet but I cant be bothered typing it out. Its to do with the emmissions of electrons needing a certain voltage because your LED is a semiconductor.

To others talking about if this makes you see further.

If you drive it with a high frequency square wave at 20 % duty cycle you will only see the low frequency components of the square wave so it appears on all the time. You wont see twice as far unless its flashing slow enough for you to see the flashes.

In reply to David Coley:
On my headtorch, the strobing is imperceptible - apart from once, in light rain, where on full power the rain "spraydrops" were falling down from the sky and on low power it looked like they were going upwards. Very cool.

In reply to itsThere:


But might you see 1.5 times as far?

Am I right that the cells in the eye do not fire more aggressively as light level increases, but simply more often? This suggests they act as counting devices, so PWM would not allow one to see further and just the mean out would?

This is an interesting read: https://en.wikipedia.org/wiki/Photoreceptor_cell

In reply to Neil Williams:

I totally agree. It's a nightmare.

In reply to wintertree:


Might be that they are 100Hz and I'm more sensitive to it than I thought.


Agreed. There needs to be a standard, including a maximum, as overbright lights can hide the vehicle you really need to see because it's about to hit you - or worse the cyclist or pedestrian.

In reply to itsThere:




Yes, you need a forward voltage above the switching point. It's a diode.

In reply to David Coley:

I would need to look up about this as I cannot remember all the physiology, but Cones have a higher threshold than Rods, but they get bleached more readily. Rods responses are better if there is a period of dark adaptation. Hence the reason that you can see further in the gloaming without a head torch, compared to with a head torch albeit more vaguely(The concentration of Rods is less than the cones in centre of the retina and they tend to be distributed somewhat more peripherally. Plus in the dark the pupil is wider so there is less depth of field). Additionally Rods tend to be wired up in a multiple way to individual gangion cells for an additive effect, more than cones are(which tend to be more often 1:1)

In reply to David Coley:
I dont know, I dont think I am 100% correct into assuming the LED sees the square wave as a bunch of sine waves as then they would still need enough amplitude to make the LED emitt a photon.

There is prob an explanation in the book "the art of electonics".

In reply to David Coley:


This blog post doesn't address the flicker question, but it gives a really good account of the physics of where 683 lm/W comes from, how luminous efficiency works, and what can be achieved in practice. http://physics.ucsd.edu/do-the-math/2012/05/spectral-extravaganza-the-ultim...

In reply to itsThere:

They won't be square waves. It'll just be some approximation built from a couple of bipolar transistors. Good enough for a cheap head torch.

In reply to David Coley:



Just skimming this, but can't see any direct reply to this. My gut feeling is that it doesn't. Certainly that's not what I thought and I'm not convinced there's any physical or biological reason it could be true.

The point, as others have alluded to, is that [being a bit sloppy with my description] the drive circuits operate most efficiently when, at any moment in time, they are either fully-conducting or not conducting at all. Partially-conducting (as would be necessary for dimming) would lead to unnecessary heat loss.

In reply to jonny taylor:
It's more of a case that to dim in the case of a filament bulb, you reduce the current with a resistor. That resistor is a waste of energy as you're just using it to create heat. All be it a very small amount and probably not significant.

But also in the case of an LED it needs a particular voltage difference to switch on because it's a diode. If you use a resistor you also drop that voltage to the LED which is already down at 3v.

In reply to DancingOnRock:


It won't be approximating a sine wave for driving an LED; that's pointless. It will just be a simple on-off PWM. And most likely using a mosfet, rather than bipolar.

More sophisticated high brightness LED drivers regulate by controlling forward current, using a current-sensing high frequency switched-mode power supply (SMPSU). So they're sort of PWM, but using an energy storage device (inductor or capacitor) to provide a continuous drive current to the LED. These can be used to vary the output by changing the reference voltage, or the current sense resistance (or both), thus changing the forward current. There's probably a limit to this control range, so it may be supplemented by lower-frequency PWM, turning the SMPSU on and off with the PWM control signal.

Car brake lights are probably 'smart' these days, taking a signal from the CAN bus to tell them what to do; off, low PWM for rear light, high PWM for brake light, etc. So the PWM control is probably in the light unit itself (either the 'bulb', or the light cluster). At least that's probably how I'd design it...

The whole point of CAN is to reduce the size, weight, cost & complexity of the wiring loom, replacing unique cable runs with a power bus and a redundant CAN control bus. Smart actuators can also report their health back to the ECU.