Put out the lights on the age of reason

[venta]

Help me, LJ, you (and your knowledge of physics) are my only hope.

Why do LEDs bounce up and down in car mirrors?

I noticed ages ago that, when driving down the motorway, the LED variable-speed-limit signs bounce around crazily when seen in the rear-view mirror. Obviously in a mirror you're looking at the other carriageway's signs, and the road, cars, streetlights, gantry etc are all more or less stationary but the LED signs? Well, they're joggling around like wild things.

Trundling back on Monday from a trip to Scotland, I spotted that some cars (mostly newish Mercs, I think) now have a strip of white LEDs under their normal headlights. When viewed through the rear mirror the car and the headlights look perfectly sensible, while the LEDs are - you'll have guessed - madly dancing around. I don't know if our car has an unusually vibration-prone mirror, but in general the image in the mirror seems stable.

I haven't noticed this driving at non-motorway speeds - though I'm also not sure if I've had the requisite LEDs-behind-me in any other situation.

I just asked Physics!colleague about it, and his first theory ran thusly: LEDs have quite a narrow field of view (compared to other lights), so if the object in the mirror is offset (as a sign on the other carriageway would be) you might be on the edge of the field of view, and the effect is that the light pops in and out of vision. This feels inherently wrong to me: I'd expect the sign to look more flickery than it does. Also, a Merc driving in the lane behind me is not offset.

His second theory (bolstered by a quick google) was that there is no such effect, I am insane, and ChrisC (who claims also to have observed it) is just humouring me.

So... has anyone else seen this happen? Anyone know why it happens?

Comments

[zotz]

I've seen similar illusions, and I've seen it mentioned in print, but I don't know what it's called.

[venta.livejournal.com]

I notice you stop short of affirming that I am not insane :)

[zotz]

There is surely no room for doubt concerning your sanity.

It sounds like it should be related to the illusory movement described here, which even mentions car mirrors, but annoyingly doesn't give it a name. All the examples I can think of do involve vibration, but I don't know why it would be more obvious with some images, such as LEDs, than with others.

[venta.livejournal.com]

That's interesting. Although, as you say, it doesn't really explain LEDs vs. everything else, I didn't know you could produce stroboscopic conditions just by vibration. I may develop an interest in eating carrots, now...

[zotz]

I don't see how that would work if they weren't flickering, but maybe they are? While the engine's running there could be a significant flicker from the alternator . . . unless it's properly smoothed, of course.

[venta.livejournal.com]

I was briefly confused there, and wondered why carrots would flicker.

I have just realised that I have remarkably little idea of how LEDs actually work, and whether they do emit continuous light or flicker very quickly. The Wikipedia page seems quite long, so my education may have to wait while I finish what I'm actually meant to be doing this morning.

I think the most noticeable effect is from overhead LED signs, where presumably alternators aren't relevant. Though I guess they may be AC powered, if that's in the least relevant. Good grief, I have absolutely no idea how all this stuff works :(

[zotz]

I can't really vouch for the publication - maybe someone else knows more about it - but this article says:

All of these LED driver solutions, including some of the more sophisticated LED drivers exhibit some degree of this line frequency sine wave artifact. Many LED drivers introduce low-frequency flicker at two times the line frequency: 100Hz flicker for a 50Hz line frequency or 120Hz for a 60Hz line frequency.

This little Q&A agrees. Hardly conclusive, though.

[shermarama]

LEDs emit continuous light, as far as I know (they're just converting electrons to photons), are usually run from DC because they need a minimum voltage in a particular direction to work, and there'd probably be a fairly well-smoothed supply going to high-power ones in a road sign. The more powerful they are, the fussier they are about their power supply, or rather it's worth being fussy about the power supply to get the most out of your expensive high-intensity LEDs.

My totally off-the-cuff theory would be something to do with point light sources, and the fact that you can see much more clearly where all the individual points are in the LED strip. They might be moving no greater distance than the headlight above them, but perceived movement of a few cm in a thing that's clearly a small point might trigger different 'moving thing!' brain reactions to the same perceived movement in the relatively much bigger headlight.

[shermarama]

(Never noticed it myself, though, and can't keep an eye out for it because I don't have a car right now. Perhaps in another era you would have made a particularly sharp-eyed hunter?)

[venta.livejournal.com]

I've only started noticing it fairly recently, so I'm guessing it might be that LEDs are just becoming a lot more prevalent than they were a few years ago.

I suspect hunting wouldn't be my forte. At least, not unless I was allowed to claim deer, boar etc that died laughing as I crashed noisily through the undergrowth towards them.

[shermarama]

Oh, yeah. LEDs have a lot to recommend them as a light source but it's only recently they've been usable for actual lighting applications. The difference between what was called an ultrabright LED ten years ago compared to now is almost shocking.

[huskyteer.livejournal.com]

I've noticed something similar with LED cats-eyes; I find it really distracting. I've always assumed it was because they're very small and very bright.

[venta.livejournal.com]

Actually, now you mention it, so have I. When they first installed lit cats-eyes on the A4074 between Reading and Oxford a few years ago, I noticed that they were very distracting in the wing mirror.

I'm not sure whether they were actually bouncing about, or just unexpectly present.

[bateleur.livejournal.com]

Well, I don't know what the actual answer is, but here's a completely mad theory: is it possible that you're seeing interference patterns due to the fact that all of these sources use multiple small LEDs and the light from LEDs is usually a pure, single frequency? The vibration could be moving your direction of vision rapidly enough relative to a fine-grained interference pattern that it produces a sense of flickering even though you would not normally be able to perceive the interference.

</implausible_handwaving>

[venta.livejournal.com]

I'd say it very much doesn't produce a sense of flickering - it's definitely a sense of movement without flickering.

But I'd have bought the argument if only you'd ended with more plausible handwaving!

[bateleur.livejournal.com]

Just as well I didn't then!

(According to zandev I have a particular talent for persuading people of things that aren't true. Normally in the context of mathematics, but it's possible it works for physics too.)

[venta.livejournal.com]

I can't judge your skills in that direction, because I'm extraordinarily easily persuadable of almost anything!

[ar-gemlad.livejournal.com]

I blame polarisation. I have nothing else to add, because I haven't had enough coffee for science, and I've had too much coffee for batshit handwaving.

[venta.livejournal.com]

Polarity has a lot to answer for (http://www.thedailymash.co.uk/news/society/osborne-reverses-polarity-of-pound-2013022761155).

Surely lots of coffee makes for excellent batshit hand-waving?

[ar-gemlad.livejournal.com]

I shall have to drink more coffee. For science.

[lanfykins.livejournal.com]

Here comes the war...

...or to put it another way, I have rear-view mirrors? :)

[venta.livejournal.com]

Your kudo is temporarily suspended until you assure me you're joking about the mirrors ;)

[lanfykins.livejournal.com]

I am joking about the mirrors. I actually rely on them quite significantly.

And now, I'm going to be glued to em watching for the phenomenon you mention...

[venta.livejournal.com]

Phew, I shall release the kudo :)

Please do also remember to look forwards. I hear it's qutie important, too.

(Best phenomenon-spotting time is, I think, on the motorway at dusk.)

[bateleur.livejournal.com]

watching for the phenomenon

Do doo da do do...

[venta.livejournal.com]

I was relying on someone to say that :)

[feanelwa.livejournal.com]

*checks using handy LED night light*
Oh yes, so they do. And it's not in rhythm with my heartbeat either.

I wonder if it's to do with saccades, which are about tens of ms long. Something to do with LED lighting being an array of small point sources of light and your eyes don't quite get back to the same position each time they look at the array, so your brain gets lots of overlapping images displaced by ~100ms of the the same array at slightly different places on the retina? That sounds vaguely plausible. I wonder how you could test it. There's got to be a trick with video somewhere.

[venta.livejournal.com]

*checks using handy LED night light*
Oh yes, so they do.

Hang on, do you mean you reproduced it while you were stationary? Not bouncing about in a car? If so, that's interesting.

There's got to be a trick with video somewhere.

I have been wondering today whether it'd be possible to catch the effect on film, or if (for example) the frame rate would be too low and would make it look different. Only right now I'm a bit hung up on how to film something in the rear view mirror without endangering life and limb :)

[feanelwa.livejournal.com]

I don't know if it was the same amplitude and frequency as what you saw.

How to film something in the rearview mirror without endangering life and limb: be the passenger. But you'd have to consider different variables, like do you want the camera to be fixed relative to the mirror, the rest of the car, or the human? If you want to stick it to the human of course it's not that hard, you can get cameras that stick to helmets or boats so you could presumably stick one to your head with a suitable hat.

[venta.livejournal.com]

Yes, I took it as read I'd be the passenger! Just not sure how you'd achieve it without getting in the way of the driver's view. Actually, it might be possible to film from the back seat (behind the driver) without causing trouble.

I assume I'd want the camera stuck to the human; at least for a first attempt, I guess I'd want it to be as near as possible what I see.

[feanelwa.livejournal.com]

Relatedly, this video is *awesome*, especially near the end where they slow it right down loads so you can actually see them. http://www.youtube.com/watch?NR=1&feature=fvwp&v=ySMtB5nWxPs

[venta.livejournal.com]

Wow, that is cool, thanks! I kind of know rationally that eyes are full of liquid/jellylike stuff, but seeing an eye like that actually makes it real.

[zotz]

Hang on, do you mean you reproduced it while you were stationary? Not bouncing about in a car?

Sorry - yes. Very much so. I - and via the magic of the interwebs I can also say various other people - have noticed this with LED clock-radios, particularly while eating, humming or using electric toothbrushes.

[venta.livejournal.com]

I realise cars aren't particularly required - just that Jo didn't specify any means by which she was vibrating...

[zotz]

It's in a book too, with someone humming while watching television. It might be something by Neal Stephenson.

[fractalgeek.livejournal.com]

LEDs have almost no lag between power and light, and no smoothing. So if scanned, modulated or powered by any sort of AC or chopped supply, and there is untracked movement, you will see the modulation. An extreme example is the in-tunnel adverts you see from trains (eg in Japan) where a column of LEDs stays still, but projects a full image, due to the train's movement.

OK - now to freak you further.

http://www.newscientist.com/blogs/nstv/2013/02/flash-lag-illusion.html