Static, the opposite of harmony. A difficult week, partly because teaching and adminis-trivia flourished, and partly because efforts to quiet the noise didn’t work. But I do have a new hypothesis to worry like a loose tooth. Readers of the previous post will know that in attempting to extend the work on the stability of the root’s velocity profile, I found that there is an unacceptable amount of noise in the imaging system, somewhere. This appears as intensity levels fluctuating from image to image, even when nothing is moving. Based on comparing short movies made with light supplied from an IR diode array vs a flashlight, I had zeroed in on the mains power and this past week I set out to test this explanation.

I found a power conditioner, a gray box the size of a cinderblock and just as heavy, thanks to the electrical storage room of George Drake, Morrill’s IT master. George also had a spare uninterruptable power supply (UPS) that he thought would be worth trying too. Neither seemed to reduce the noise. This was more or less expected with the UPS but a serious disappointment for the line conditioner. Perhaps it is not the mains power after all?

Going back to square one as it were, I set up in the darkroom and tried a variety of light sources. While the IR diode source continued to emulate a candle, the room’s overhead fluorescent lights were rock solid. Not quite trusting to my eyes, I started measuring the intensity of an area in the images and calculating the average and standard deviation. This revealed that contrary to my impression, the flashlight LED was noisy (the narrow spot-like image just made the jumps a bit harder to see by eye). Nevertheless, the movies made with room fluorescent light proved stable. This is heartening because it would seem to rule out the camera or the computer as the source of noise. I wonder if both the IR diode and the microscope light source are intrinsically noisy?

I tried a compact fluorescent bulb in a fixture plugged into the wall and this was as noisy as the IR diode, though in retrospect I did not let it warm up very long. However, a regular incandescent bulb replacing the compact fluoro in the same fixture was stable. Since I expect that an incandescent bulb would respond to voltage noise in the mains, this could be further evidence that the light sources are the culprits.

To get information about the frequency of the noise, I got the IR diode array nice and close so the exposure was 2 msec, and then I took sequences with 10, 100, 1000, or 10,000 msec between frames. There was no difference in the SD for these sequences. This implies that the noise is faster than 100 Hz. I have no idea what this means but someone might.

As soon as I started to make measurements, I realized that I needed more frames. While 5 frames is fine to “see” the flickering, to get a good handle on the variation (e.g., by calculating a mean and standard deviation), I need a larger sample size. I think ten frames would be good and perhaps a few replications. I will compare the IR diode array to the microscope’s light source by measuring some long-ish sequences to see if the standard deviation is the same – if yes then back to the mains power—if no, then good evidence that the noise is from the lights, and just bad luck that two different sources are each noisy. I also want to confirm that the incandescent light is stable but the compact fluorescent is not with some longer runs, and with the latter sufficiently warm. Will harmony descend?