Affiliation
American Association of Variable Star Observers (AAVSO)
Fri, 10/19/2012 - 21:22
I'm doing I band photometry and am seeing etaloning in the images--I believe it's being caused by constructive and destructive interference due to the scraping of the chip and the gap between the chip and its mounting in the camera.
It's exactly the same pattern in each image, so I'm wondering if there's a way to compensate for it using a special type of flat (a normal flat won't get rid of it). Does anyone know of an etaloning technique--a special type of flat or other solution--that will work?
Thanks.
Steve Smith (SSTB)
Images taken at
Hi Steve,
Images taken at longer optical,near-IR wavelengths tend to show this most often due to OH- emission bands.
Although airglow is not constant, a propotionially adjusted "fringe" correction should
be possible.
Cheers,
Doug
I assume that you are using a thinned back illuminated chip to do photometry. Otherwise etaloning would not occur. In most broad and medium bandwidth filters, etaloning isn't a problem since the fringing averages out as explained in the excerpt from a Princeton Instruments white paper below. I have attached the complete white paper for reference and it URL is http://www.princetoninstruments.com/Uploads/Princeton/Documents/Whitepapers/etaloning.pdf
"In most imaging applications with back-illuminated CCDs, spatial etaloning is not evident because the applications are at shorter wavelengths, where the silicon absorption damps out the etalon effect. In addition, many applications use light that is spectrally broad enough to span (and average out) several etalon-fringe cycles. The latter requires only a spectral bandwidth of a few nanometers. In a spectrometer, by comparison, the light on any one column of pixels is very narrow spectrally, typically less than 0.1 nm. Thus, this spectral bandwidth is much less than the period of etalon cycles (~5 nm). As a result, spatial etaloning is quite evident when viewing an image of a uniform spectrum (e.g., tungsten bulb) in the NIR (see Figure 5)."
The exception, as WEI mentions, would be where strong emission lines are present since they are monocrhomatic sources and ar bright enough to be noticeable.
In the photometry I have done with back illuminated chips I have not run into the problem. I guess I was luck not to have objects with bright emmission lines. It seems to me that you should be able to spread out and average the fingeing from a stellar object by slightly defocusing. that would create a range of path lengths through the chip that could average out the fringing. that seems to be the basis behind reducing the etaloning in spectroscopy by spreading the spectrum out along the height of the chip so that spatial and spectral etalonging tend to cancel each other. Perhaps Arne Steve Howell, or someone else with a lot of experience with etaloning in back illuminated chips could comment on whether this would work.
Sorry failed to attach the white paper
This happens primarily with thinned, backside illuminated chips where the thickness of the silicon is equivalent to the mean free path of the photon. Therefore, it is more prevalent with longer wavelength photons, such as at I-band and z-band. Our Sloan z images at NOFS were always fringed.
As Brad suggests, this is just like a sky background (actually, it is EXACTLY like sky, just a modulated version), so you can easily subtract it out. What I typically would do is take a few long exposures of dithered blank sky, remove the stars, and form a master fringe image. You then just scale that to match the sky background in your image, subtract, and voila! The fringing goes away.
There are a couple of gotchas. The pattern depends on the wavelength, and the night sky emission can vary during a night. So you need a fringe map that corresponds with the monochromatic illumination source present at the time of the science image. It also is a weak function of temperature, so will differ at different operating temperatures. Finally, since you are subtracting this, you want good signal/noise in the master fringe image, not always easy to obtain.
Arne