The Starry Night, 46
:: home ::
04/20/2011. I want to produce flats as easily as darks. I realized that I was losing a lot of opportunities to try different compositions and optical configurations because I could only take evening and morning sky and t-shirt flats.
(Note for those of you with actual night lives: "Darks" are images captured with the shutter closed which match the duration and chip-temperature of an astronomical image; you subtract them from astronomical images to remove electronic noise in the form of bright pixels. "Flats," shorthand for "flatfield images," are photographs of an evenly illuminated surface. "Sky flats" are flats that rely on the starless twilight sky to provide that featureless subject. "T-shirt flats" are flats made by stretching t-shirts [or something] over the objective and shooting in dim light. However they're lit, flats reveal the shadows of dust particles within the optical path, vignetting, uneven illumination — any nonuniformity arising within the imaging system. Darks record only features of the CCD detector; the rest of the optical system is not important to darks. You can make them anytime by simply closing the shutter and cooling the chip to the proper temperature. Flats must be captured with the sensor and the optics in exactly the same configuration as when taking the astronomical images they are meant to correct. The data in flatfield images are divided into astronomical data so faults arising from uneven illumination cancel out. After applying darks and flats, more aggressive histogram stretches and deconvolution algorithms can be used because [
Having to rely on flats shot at the beginning and end of night meant I could not rotate the camera or change effective focal lengths or swap out, add, or remove 2-inch threaded filters without committing to staying up (or getting up) to shoot dawn flats. At best, I could try two configurations per night provided I could take twilight flats to bookend them. A light box to take the place of the twilit sky is an ideal solution; electroluminescent panels provide extremely even light across relatively large areas. With one of those, I could change whatever I wanted then casually take a new flat to match the new configuration. Before EL panels, flat field boxes were huge, relying on multiple diffusers and elaborately distributed light sources. You can buy a very nice EL flat box if you have $300 and a USB port to spare; or you can make one and save upwards of $250 and a port.
I ordered a small (A5) electroluminescent panel from the Glow Hut in Chino, CA, and a transformer to power it from 12v sources. $38 all up. I'll need a cigarette-plug adapter. And an appropriate length of wire, and some acrylic to sandwich it, and some bolts, and... that's about it.
04/23/2011. That was fast. The Glow Hut's package arrived today (I'd opted for the free "slow" 1st class shipping). The panel is, as promised, 5.8 x 8 inches. The panel arrived with a thin static-cling film for protection which I only noticed because it was not applied perfectly and where it was loose, the panel's light was slightly different. Once seen, it peeled right off.
While brainstorming and messing with the pieces, I noticed that the transformer was "rated" for 12v so I wondered if it would be happy with less. Sure enough, a 9v battery drives the transformer and the EL panel just fine. The light is dimmer than with 12v, but it's plenty bright enough. Maybe too bright. This simplifies much: forget the cigarette lighter adapter and long cord because a 9v battery is small and light enough to ride along. I cut out two 8x10-inch pieces of white acrylic from the sheet that used to comprise the top of that huge lightbox I built for slide sorting (remember "slides"?) back in the 1970's and sandwiched the EL panel between them.
The sandwich is held together by 4 bolts. I used silicon glue to attach the transformer to the acrylic behind the EL and soldered a 9v battery connector to its input leads. I made a battery holder from a short piece of aluminum channel sold to edge 3/4-inch plywood. A generous dose of black tape holds the wires down and keeps everything tidy. Pictures show the back uncovered (haven't found or made a suitable cover yet) and the electroluminscent panel electroluminescing out on the porch. Voila! A self-contained light souce for flats. For ordinary use, either before or after a series of astronomical image captures, slew to the zenith, snap the 9v battery to the connector, lay the EL-panel-on-acrylic sandwich over the objective, take a flat; you're done.
I let the silicon glue set for a couple of days while we visited the Tiptons in JC. By the time we got back, it was fully cured and as secure as it's ever going to be. The weather forecast says I might get to try it all out later this week. I thought that the device might need to be mounted in some kind of case for durability, hence the generous 8x10 dimension: most any picture frame will serve as a robust case if one is needed.
Glow Hut sells a 9v transformer with an on/off hi/lo switch, but it's listed for A6 and smaller panels. I went with the 12v transformer recommended for A5 panels, and I am underfeeding it. Wise? I dunno. Better choice? I dunno. Does this work reliably and well over the long haul? I'll let you know.
04/28/2011. I decided to go after something faint to try out the lightbox. It seems to work great, except that the intensity of the light fades rapidly after it's turned on. Maybe the battery is too old; maybe it just can't supply the current. Anyway, behold: Leo-1 in the glare of Regulus. The big, sharp circle is obviously a reflection of Regulus, probably from the front and back surfaces of the Baader fringe-killer I left in place after some experiments with a 6-inch achromat. More about that by and by. Leo-1 is a dwarf galaxy, a satellite of the Milky Way 800,000 - 900,000 light-years out, discovered in 1950 using one of the world's great survey telescopes. Here it is from my backyard:
In the uncalibrated image, I've marked dust motes and circled an area on the right that is significantly brighter than an area on the left a similar distance from Regulus. Note the peculiar "laddering" at lower left. And note especially that all these things are missing from the image calibrated with the flat. The bright white line at upper right is a column defect in the CCD; a dark frame would remove that. Next is the flat field itself showing only the defects. An ideal optical system, of which there just aren't any, would yield a uniform gray rectangle:
Here's something a little brighter, without the glare of a first magnitude star to make life more interesting than it needed to be. I'm particularly happy to catch the slight twist at the extremities of the galaxy's disk; more light should make it clearer. A brief series of RGB hurt more than it helped. Eventually. In the original, I suspect a swarm of very faint galaxies beyond the upper tip:
There are faint spiral arms reaching off around the top of the galaxy but they are easily lost in removing the gradient which runs from lower left to upper right. Surprisingly enough, some of that appears to be instrumental because the flat reduces it (or else removes the visual cues that make it obnoxious), but neither I nor Gradient Xterminator has managed to completely remove the gradient. The gradient is much worse in later, low-elevation, beginning-of-twilight images. I shot 60 five-minute subs; the last few are useless, but tossing the last 14-15 is worthwhile. Some shadows of optical nonuniformities remain, so the flat is not as good as it might be. Still, not bad for the first time out for the flatbox. I wonder to what degree the tiny galaxy to the right of M101 mirrors the relationship of Leo-1 and the Milky Way?
04/29/2011. The telecompressed F4.2 version of the A-P is great for big nebulae and wide starfields, but when imaging galaxies, scale matters more. It seems clear from yesterday's photos of Leo-1 that the fringe-killer is introducing some terrible internal reflections with a 1st magnitude star sharing the field. Tonight I removed the filter and the telecompressor. To make a flat, I slewed to the zenith and laid the flatfield box on the dewshield as originally conceived. I shot flats in LRGB just for drill, but I used only one of the L flats. I took a single round of RGB to put color into the stars and I was surprised to see the dwarf in each subframe. Put it all together and Leo-1 resolves nicely:
Leo-1 and Regulus
It is hard to imagine just how hard I am pushing 100 minutes of data to show Leo-1 that well and to pull stars out of it, too. The middling bright field stars bloated up some and act strange, but I am astonished that Regulus behaved as well as it did. Note that the field is very clean with no significant gradients. That's the way you do it. If I'd suspected that Regulus would be so clean, I'd've gone for a vertical composition with the entire luminous ball of Regulus's glare. Next time.
From Leo-1, I went back to M101. I ran the camera into the pier (twice) trying to pick it up around midnight. Ten minutes later, all was well. I cooled the chip another five degrees to -30C, made a new dark frame, took 10 minutes of RGB, and then left the telescope to collect luminance data until dawn. Note that more data is not always better; sometimes better is better.
:: top ::