Nightlight Theory
January 9, 2007 6:57 PM   Subscribe

Lets say I have a room thats 10 ft by 10 ft. Each wall of this room has 4 outlets (this number can be increased or decreased) each outlet has a nightlight that is very sensitive and shuts off when the brightness reaches a certain level (all nightlights have the same sensitivity level.) What happens when the main room light goes off?
posted by JonnyRotten to Science & Nature (14 answers total)

This post was deleted for the following reason: doesn't sound like any problem to be solved, rather some idle 'what if' curiosity

to answer this we'd have to know at least the output and the ambient light threshold of the nightlights; possibly the positions of and distance between outlets.
posted by sonofslim at 7:04 PM on January 9, 2007

The answer probably depends on the wattage of the nightlight bulbs... But I'm guessing they will all flicker on and off annoyingly (like the automatic nightlight in my kitchen does at twilight).
posted by amyms at 7:04 PM on January 9, 2007

All the 'very sensitive' nightlights come on. Assuming they're sensitive enough (read: badly-designed enough) to consider the light emitted by four nightlights to be above their brightness threshold they simultaneously switch off. Rinse and repeat, ad infinitum.
posted by matthewr at 7:06 PM on January 9, 2007

Response by poster: Lets say one light is not strong enough to turn off another directly. But a second light being on produces enough ambient light to turn one off.
posted by JonnyRotten at 7:08 PM on January 9, 2007

There are all sorts of potential modes, and there's no way to predict it a priori (probably even if you know the thresholds and positions).

Assuming the thresholds are such that the lights trigger each other (a very questionable assumption, as the photodetectors have to be positioned to the light from that nightlight doesn't hit it), you might get some fraction of the lights on and some fraction off, or some sort of flickering pattern. This will be dependent on geometry of the situation (distances/placements), the power and direction of the light, the sensitivity and directivity of the detector, whether the detectors have the same on/off threshold (some things turn on at a different value from where they turn off), and the timescale of the electronics within the nightlights.

In general, physical systems with mutual feedback will either assymptope to some value (all-on or all-off, for example), or oscillate. The oscillations might settle into a specific pattern or might be chaotic.
posted by JMOZ at 7:10 PM on January 9, 2007

FWIW, I have a friend who has a night light that is brighter than what the shutoff sensor allows for ambient light. The stupid thing flickers all night.
posted by niles at 7:36 PM on January 9, 2007

I assume you are posing the question so that it is intereting, i.e. the nightlights will trigger each other. My intuition tells me that the result is chaotic. As in, they will flicker in a seemingly random pattern. I think the last sentance of JMOZ's answer is probably the best desription of the problem.

I don't think matthewr's answer (oscillation) is realistic. All on -> all off -> all on requirs exact synchronicity, and a small perturbation in either of the four lights - say one is slightly slower to respond and doesn't turn off when the other three do, and thus ends up staying on - would push the system into a chaotic state.
posted by PercussivePaul at 8:03 PM on January 9, 2007

I don't know the answer, but I think this is pretty neat idea.
posted by mrgreyisyelling at 8:43 PM on January 9, 2007

Assuming that the light from each night light is applied to all the other ones simultaneously, and all kinds of other weaselly conditions:

I think that as soon as the light level drops below the nominal ON setpoint, a race would occur due to the impossibility of exactly synchronizing setpoints and response times. So, the fastest/highest setpoint night light would come on, adding enough light to hold the other 3 off.

Later, if/when the (ambient light + (1 x night light)) fell below the setpoint, same thing again.

In real life, they would either not affect each other, due to distance and sensing angle, or they would in unpredictable ways. You would then decide if maybe a better way might be to have them all controlled from one light-sensing switch, and the setpoint adjusted to not trip on their own light level.
posted by ctmf at 9:31 PM on January 9, 2007

Rereading the question, my answer doesn't really apply, since it assumed ambient light went down gradually.

If you step-changed the light level to below all night-light setpoints, then the response times of the night light circuitry would be the important part. You couldn't really predict what would happen unless you designed a particular behavior on purpose. For example, you could put delays in the ON switch of 0,1,2,3 seconds so each light had a different delay. Then they SHOULD come on one at a time until the number of night lights was enough to supplement ambient to above the setpoint.

Hm, is there such a thing as a light-level regulator, that could control a dimmer on a night light to maintain the light level exactly at some value? One of those controlling all the night lights together would be neat.
posted by ctmf at 9:46 PM on January 9, 2007

You've basically sketched out a cellular automata. The nightlights are cells and neighbors of each other and they have associated thresholds for transitioning into different states.

As with most CAs, such as Conway's Life, you just have to model and then run it, as a program, to find out what the final state will be. To model imperfect physical behavior you should build in switch delays as ctmf suggests. The result might be stable but more likely short-lived (because this is a small discrete system) chaotic behavior followed by an oscillation end-state.
posted by vacapinta at 11:26 PM on January 9, 2007

What happens is unpredictable based on the limited info in your scenario.

The only thing that is certain is that the illumination from the main lamp will disappear. Beyond that is entirely speculation, but vacapinta's suggestion of chaotic behavior seems closest to the target. I suggest he's being charitable by considering the spirit of your thought experiment and assuming that you can get "equal" sensitivity units, etc.

Data missing includes night light sensitivity, hysteresis, mutual illumination, night light output and light distribution, response times, proximity to the main lamp, room characteristics, just to name a few.

The question I would have is "What do you want to happen?" If you want a specific response, such as oscillating illumination, you could probably derive a set of specs to make the nightlights behave that way. For certain, you'd have to get a handle on hysteresis and sensitivity, and intentionally design some sort of mutual illumination rules and I think it would be progressively more difficult as the number of lamps increases.

That's just a guess, though.
posted by FauxScot at 1:58 AM on January 10, 2007

I agree, too little information.

But I have this thought: Light intensity diminishes with distance. If a nightlight's own bulb can't trigger its sensor, why would its neighbor's?
posted by ardgedee at 5:14 AM on January 10, 2007

Good observation, ardgedee.

OP's thought experiment seems based on a whole bunch of the coming on at once, though. Actually, reading between his widely spaced lines, he seems to suspect that they'd ALL come on at once, and the net illumination would cause them all to then go off. He's pondering if an oscillating system would result, I suspect.

You could easily MAKE one, but in his box, there isn't enough info for analysis, only speculation. His room size speaks against it for exactly the reason you mentioned plus a few others.

A normal given night lamp CAN extinguish itself, if the light is reflectively coupled back to the sensor, incidentally. The designs of most I have seen have simple mechanisms for preventing it, but I have seen many oscillate transiently. Cup your hands around one and see.
posted by FauxScot at 5:58 AM on January 10, 2007

« Older How much to pay kids to tag photos?   |   Drunk sex = regret Newer »
This thread is closed to new comments.