Old info but still works with Star2000:  Dark frames are taken to correct for noise generated by the chip in the camera.  The Sony chip in the MX7C is extremely low in noise but during guiding with star2000 amplifier glow on the left side of the image can become prevelent at longer guiding lengths.  Dark frames easily correct for this.  Using Astroart dark frames are taken at the same exposure length as the image.  Since there is a time lag in moving the telescope during guiding, a slightly longer dark frame guiding time is necessary with Astroart.  In my case I usually guide with 1 sec guiding exposures.  My dark frame guiding exposures need to be 1.2 sec in order to correct for the amplifier glow.  If I use 1 sec exposures too much correction is made to the left side of the image.

    Its always better to average or median several dark frames together instead of using just one.  I usually take 16 darks and median combine.  In a perfect world you  would take darks at the beginning, middle and end of an imaging session.  Since I don't usually have the patience to do this, I've found that dark frames taken at a certain temperature can be used many times if the images were also taken at the same temperature.  I always log the current temp at the start and end of an imaging session.  I average the two and pick a set of darks taken at that averaged temp.  This technique seems to work perfectly fine.

 This is the way I did it with Star2000 and MX7C Several few months ago, another imager, Robert Vanderbei came up with a technique in which the imager takes master guided dark frames and bias frames.  The idea is to seperate the glow caused by star 2000 guiding from the normal image bias and hot pixels.  This technique works like a charm and I now routinely use it.  Follow this link to Bobs Site: http://www.princeton.edu/~rvdb/images/calib/dark.html   I spent an evening shooting master darks.  The more you use the better for reducing noise so I let the camera run all night and got (32) 900 sec darks (guide frame running at 1 sec).  I averaged these into a master dark frame and use these frames to scale to the image.  I also collected (32) bias frames (integration time set at 0) and use these to subract our the bias.  My goal was to have the most noise free dark as possible.