1

Sphere

This a fully enclosed sphere containing a web cam, white led and photo resistor in exactly the geometric arrangement shown. The led glows dimly (via fixed dropper resistor). It runs at about 9 mA. It's light reflects off the inside of the sphere and the web cam takes pictures of the inside of the sphere. The photo resistor is there to measure relative light levels. The system is theoretically in a steady state (including temperature which hovers around 29 deg.C). and there is no extraneous light. This is what the camera sees:-

web cam image

and this is that image's histogram:-

web cam histogram

The thing is, it's not steady state. These are the photo resistances and histogram means in two experiments each taking three weeks The initial number is at the start, and the second number is at the end of three weeks. They gradually change from one to the other as I measure them.

  1. 69.4k -> 136.5k, 33 RGB -> 1 RGB
  2. 28.7k -> 55.7k, 49 RGB -> 21 RGB

Both times the photo resistance and histogram mean drops very significantly, in experiment 1 almost to zero. The inference is that the inside of the sphere is getting darker as time goes by (measured in weeks). This question regarding photo resistor stability was inconclusive but cannot explain the shifting histogram.

It's difficult to judge by eye, but I think that the LED is getting slowly dimmer. Occam would suggest so, but I have used two brand new LEDs so they would have both behaved like this. The generic LEDs were both from a cheapo bulk eBay purchase.

Q1. Can this be explained from an electronics perspective?

Q2. Can anyone suggest any further investigation(s) I can undertake?

EDIT.

I came across something called current crowding which seems to fit my problem. I can't follow the link though due to a flight failure...

Paul Uszak
  • 7,327
  • 5
  • 37
  • 69
  • 1
    LEDs do degrade over time, though your time period seems short. However, degredation is increased if the LED is over-driven or over-heated. Are you over-driving the LED? Have you taken measures within your closed sphere to ensure the LED is not cooking? – Trevor_G Mar 22 '17 at 01:27
  • I'd also say a webcam is not a great "instrument" for this type of experiment, but since your LDR is confirming the results it points to the LED. And why is the image black..? – Trevor_G Mar 22 '17 at 01:42
  • @Trevor LED is running at 9 mA at 29 degrees C. The temperature stays around this point. The image isn't black. It's what the histogram reflects. It's just dark, but your monitor might not render it correctly. – Paul Uszak Mar 22 '17 at 01:51
  • You could try isolating it further, by using some other light source, maybe an external incandescent bulb fed in through a light pipe. Or measure the output from the led using some other form of light meter over time. You also need to be scrupulous to ensure the LED current and voltage are not changing. – Trevor_G Mar 22 '17 at 02:02
  • First maybe the resistor heat up and the ohm value change so the led current is not super steady. Second, the webcam surely have some image processing software that are really hard to disable (sometimes it is directly in the housing so you can't get the raw images). Those kinds of image processing are looking for contrast to enhance the image. By definition, an integrating sphere will have no contrast as the light intensity is the same in each steradian – A. Baril Mar 22 '17 at 02:49
  • @A.Baril What you say is true, but the IS achieves a virtually stable temperature within 24 hrs (it's measured). I'm expressly aiming for no contrast so that the images are of uniform intensity, excluding random fluctuations. The resistance and histogram change only in one direction over 3 weeks or so. It's not heating. The principal heat source is the web cam. – Paul Uszak Mar 22 '17 at 03:05
  • The image is dark because placing anything in the center of the sphere blocks most of the reflection paths in the sphere. All the reflections want to go through the center. When it is blocked, the sphere stays dark. Maybe this causing heating in the LED, if the LED is in the center. – JRE Mar 22 '17 at 05:57
  • You can check the bit about it being dark using any 3D rendering program. Place a light source and camera inside a reflecting sphere (but nothing in the center) and render the scene. Do the same with in object in the center of the sphere. The second scene will be very dark. Been there, done that, had the pictures to prove it. – JRE Mar 22 '17 at 06:00
  • 4
    Repeat with a primary colour LED - or 3 - R,G and B - from a reputable manufacturer - and see if the drift stops. White LEDs use phosphors to generate the yellow component, pumped by a blue LED. And phosphors do decay - especially if you bought the white LED from eBay... –  Mar 22 '17 at 11:07
  • 1
    @BrianDrummond Ah ha! It was extra cheap from eBay, bought in bulk. Is it possible that phosphor decay could be so apparent over the course of 3 weeks? Can you say more? Do you have any anecdotal evidence? Both LEDs were from the same bulk batch... – Paul Uszak Mar 22 '17 at 13:04
  • 1
    The LED and camera are constantly on, 24 /7. – Paul Uszak Mar 22 '17 at 13:20
  • @PaulUszak Did you paint the inside of the sphere a short time (days) before starting measurements? If so, maybe the paint is changing its reflectivity as it dries out. – Andrew Morton Mar 22 '17 at 13:39
  • 1
    @AndrewMorton It's probably been 6 months since I painted it. Nothing special, just ordinary brilliant white matt emulsion in several coats. I pondered mixing my own barium sulphate paint but opted for emulsion as the visible light reflectivity characteristics are very similar. – Paul Uszak Mar 22 '17 at 13:51
  • Fluorescent coating consists of different substances emitting different wavelengths. If the problem is degenerating of the coating of LED, then I think the ratio of color intensities will also change. Check the R,G,B channels invidually. – Ayhan Mar 22 '17 at 14:26

0 Answers0