So, after a couple of years of working with untracked astro photo techniques, and a fair amount of success, I finally gave in and got a tracking system. I went with the Astrotrac because of the advertised light weight, easy setup, and long, accurate tracking capability (2 hours, advertised.)

I’d seen some pretty spectacular astro photos, considerably better than mine. So I bit.

Well… as always seems to be the case with marketing… it turns out that practical tracking is limited to about three minutes. This is because it is almost impossible to align the Astrotrac, even with the specialized included polar scope, perfectly with the axis of the earth’s alignment. You can get close – as soon as you figure out that the polar scope, unlike the camera, inverts left with right and top with bottom, and that the reference constellations are not rendered with any degree of angular accuracy, and that figuring out how to match what you see with the markings in the polar scope isn’t assisted in any way by the markings themselves… It took me several days to come to terms with what stars go where in the polar scope; easily the poorest documented, and poorest performing, part of the system.

Once that hump was scaled, though, I began to be able to set the system up in about five minutes. The process is: Open the tripod and level it. Mount the alt/azimuth head, zero it (I leave the altitude at the proper degree setting… once set, no need to ever change it unless you change locations), and point it north. Mount the Astrotrac. Power it up and initialize it. Mount the ball head, the polar scope and the camera. Start the Astrotrac actually tracking. Align using the polar scope as best you can. Attach the timer to the camera, set the camera to C2 (which has my long term astro settings memorized [ISO 1600, f/4, bulb]), aim it at some fabulous DSO, focus, and let fly. Repeat.

The trick with the polar scope I will be glad to pass on. Perhaps I can save someone a few cold nights of cursing into the darkness.

There’s the north star, which is almost, but not quite, at the right place. There’s a second star which fits elsewhere in a slot on the polar scope. You use the third marking (for some reason I fail to understand, the polar scope is calibrated for 1990, 2000, and 2010 epochs. Not that you could possibly use the 1990 or 2000 epochs. Or that the Astrotrac existed at those times. Duh.) So use the third mark for the second star and put Polaris at the edge of the mark near the center. This adjusts the positioning for the drift the stars have made since 1990, when, apparently, the polar scope was designed.

Now, because the polar scope inverts everything, if the second star is up and left from Polaris, it will appear down and right in the polar scope. You have to invert the angle you see in the sky as if it was a rule placed across a clock; then imagine that line in the polar scope; then put the two stars in their respective slots. There is NOTHING in the polar scope that assists with this process. There are lines, all right, but not the one line that you need which would run between the two stars. Sigh. So you have to imagine it.

This will get you a good enough alignment — just barely — to track for three minutes. You can also use a process called “drift alignment” which is long, drawn out, and which will go to heck in a handbasket if you nudge your tripod, and so is not even worth trying, in my humble opinion.

So once you get here — again, about five minutes with some experience — you can shoot. Three minutes of exposure time is good enough to use Canon’s 200mm f/2.8L lens, which is easily the best lens I have for this application. Camera settings range from ISO 800 to ISO 3200, f/2.8 to f/5.6, and 45 to 180 seconds on the timer that drives the bulb (I use a Satechi timer, about $60 and well worth it.)

Optimum seems to be f/4, ISO 1600, and 180 seconds for most DSOs. This gets great detail on the Pleades nebula, which is generally pretty faint stuff, and is good enough to reveal Triangulum in considerable glory.

Sharp, but noisier, images may be had at f/5.6, ISO 3200, and 180 seconds. These, you’ll have to take a few (I suggest 9, 16 or 25… this will get you noise reductions and target smoothing of 3:1, 4:1 and 5:1 respectively.) and stack them. ISO 3200 is just too noisy to use a single frame for almost anything out there.

You can go the other way and minimize the noise with ISO 800, f/2.8, and 180 seconds, but you’ll get some chromatic aberration because of the low f-stop. You’re a lot less likely to have to stack with this setting, though; ISO 800 is a lot better, noise-wise. Which is not to say that you shouldn’t stack; not at all. Stacking makes the images better. Period.

And you can go for minimum time, thus reducing the requirement for accuracy of polar alignment (or contrari-wise, simply saving yourself some time) by setting ISO 3200, f/2.8, and 45 seconds. But again, this is noisy, brings the chromatic aberration, and you’ll have to stack.

Now, about stacking. Random noise reduces with the square root of the number of frames you stack. Presuming only they are properly aligned with one another. So if you shoot four frames, you get 2:1 noise reduction. Here’s a list of frames and the noise reduction you get (if you shoot some number between these points, you of course still get noise reduction, it’s just not a whole number.)

  • 4 frames = 2:1 noise reduction
  • 9 frames = 3:1 noise reduction
  • 16 frames = 4:1 noise reduction
  • 25 frames = 5:1 noise reduction
  • 36 frames = 6:1 noise reduction
  • 49 frames = 7:1 noise reduction
  • 64 frames = 8:1 noise reduction
  • 81 frames = 9:1 noise reduction
  • 100 frames = 10:1 noise reduction
  • When shooting 180 second images, shooting 9 frames means you’re out there for 9×3 minutes, almost half an hour. In the winter, that’s going to be annoying. Not just to you, but to the camera as well. So keep that in mind. Nine is what I go for when I shoot at 3200 ISO. I go for four, otherwise.

    Let’s consider this. I’m going from 3:1 to 2:1 noise reduction, or a drop of 1/3rd in noise reduction by shooting four frames instead of nine, but also having gone from ISO 3200 to ISO 800, I’ve picked up at least a 4:1 reduction in noise there, so I’m going to have a total NR of 8:1 (combining ISO 800 and a stack of four images.)

    Compare that to a total NR of 3:1 using ISO 3200 as a reference. Not bad, eh? Of course, I could still shoot nine frames, which will give 4:1 on 3:1, or about 12:1, which is fabulous, though it does cost more than twice as much time.

    One more factor. Battery life in bulb mode with the 50D is just about two hours; bring a spare if you’re going to be doing much of this. In fact, just bring a spare. You’ll be glad you did.

    Tip: Use C1 and C1 modes to memorize camera settings.

    I set C1 to be fast for seeing if I’ve found a DSO: ISO 12800, F/2.8, 2 seconds exposure, 2 seconds shutter delay drive. Sure, its noisy, but you can see if you have the camera pointed correctly in 2 seconds, which is extremely useful.

    I have C2 set to be normal for tracking: ISO 1600, F/4, Bulb, 10 second shutter delay drive, etc. This way, there’s no fumbling around, and consuming the camera battery, trying to get it set up right and remembering all the details. Get it right once, then don’t worry about it (or conveniently start from a known setting by whipping it to C1 and back to C2, then adjust f-stop or ISO in the dark, knowing exactly where the camera is set. Your night vision isn’t something to sacrifice lightly, and this little trick really helps.

    Some people tell you to set mirror lock up. I say don’t bother; the shutter is going to be open for three minutes. A one second vibration will result in less than a 1/300th instability if it was IN the shot — trust me, you don’t even have the focus set that accurately. But OTOH, if you want to, by all means. Won’t hurt a thing.