CCD Pixel Wars: Size Does Matter

Michael Purcell

As if choosing the right CCD camera for your telescope weren't confusing enough, there is an ongoing debate among astrophotographers about the "best" pixel size! Godzilla says, "Size does matter". But, is bigger also better? Perhaps. Although the theory behind the argument can be confusing, it does have a bearing on which camera to buy. Let me explain by using a mixture of theoretical and practical issues.

The pixels (quantum wells) on a CCD ship count the photons that strike them. Larger pixels collect photons faster than small pixels. This is the reason for the "buckets in a rainstorm" analogy: a bucket twice as large as another has four times the area and will collect raindrops four times faster than the smaller bucket. A practical consequence of this is that CCD chips with smaller pixels will take longer to build a usable image.

Another disadvantage of smaller pixels is that they make any imperfections in your telescope/CCD setup much more obvious. (If I had a dollar for every imperfect image I've ever taken then I would have already retired to my own private island in the Bahamas.) Prime examples are: bad focus, polar misalignment, periodic error, wind-induced vibration in the telescope, etc.

In addition, the "pixel war" is also about the practical limits of the optical configuration of your telescope. The key point is that ideally you want to match the pixel size to the smallest spot that your telescope can resolve, which is determined by the objective size and the focal ratio.. The size of this "spot" is the war.

So, how large (or small) are we talking about? The larger pixels will be about 25 microns square (25x25). For example, the pixels on the SBIG ST-6 are 23x27 microns. At the small end of the scale, the ST-7 and ST-8 have 9x9 micron pixels. These two cameras also have the ability to combine (bin) them into 18x18 and 27x27 microns.

Some literature (the FOSTERS series available through Software Bisque) makes a case for 25 microns as being the most effective size for amateur configurations. Yet, some advanced amateurs have achieved considerably better results. (See: http://voltaire.csun.edu/pixel.html)

Before I make any recommendations there is one more item to discuss: how "big" (pixels across) should be the smallest stars in an image? You might think that, ideally, the smallest stars should be a mere one pixel across. Wrong! This is called "undersampling" and will cause problems when processing your image. A better answer is three pixels. In my best images, when magnified, the smallest stars look like tiny "+"s, three pixels high and across, with the central pixel being the brightest.

Both documents mentioned earlier provide formulae to help determine what will be appropriate for you. After reviewing them, I now have a simple rule of thumb. Assuming three pixel sampling, perfect seeing, optics, focus, and tracking, the pixel size in microns should be about the same as the focal ratio of the telescope. Since we rarely have such perfect conditions, a more realistic size would be twice the focal ratio. Therefore an f/10 scope should be matched with a CCD camera using 20 micron pixels. The ST-6 fits this, as does the ST-7 using pixels binned to 18 microns. My f/6.3 LX200 is reasonably matched to my 9 micron ST-7. Other examples are the ST-5 and PixCel cameras. Both have 10 micron pixels and should be used at f/6.3 as well.

After having said all this, here are my recommendations.

  • If you already have an f/10 scope and want to buy an entry level CCD camera that has 10 micron pixels then you should also plan to buy a focal reducer.

  • You are much better off with an entry level CCD camera attached to a scope with excellent optics and a stable mount. Camera technology continues to improve. You can upgrade the camera later. Putting an expensive camera on a mediocre telescope is a mistake.

  • If your current telescope and mount are of an older design (like my previous Meade LX-5) then you should opt for a camera with pixels in the 25 micron range. It will be much more forgiving of tracking problems, vibration, OTA flexure, etc.

If you want to know more about this or how I derived my rule of thumb for pixel sizes, send me an e-mail at purcellmf@comcast.net