DMK21 camera review - still under construction
DMK21AU04.AS camera with 1.25" nosepiece attached
In November 2007 I acquired an Imaging Source DMK21AU04 camera, mainly for the purposes of planetary imaging, and I thought I would write a review of my experiences with it for the benefit of anyone else considering getting one. The camera is available in a number of variants with CCD sizes 1/4", 1/3" or 1/2" giving resolutions of 640x480, 1024x768 and 1280x960 pixel resolutions respectively. Further, it is available in colour or monochrome versions and now USB2 as well as the original Firewire versions. I selected the 640x480 mono model (as pictured above) for the reasons that (a) the mono CCD is more sensitive, especially for near IR/UV imaging, (b) I was keen to start RGB composite imaging and (c) there seems little point in getting the larger CCD, except maybe for Lunar imaging, as even the larger planets imaged at long effective focal lengths fit comfortably onto 640x480 pixels. The smaller CCD versions also have the advantage of higher frame rates (up to 60fps uncompressed).
First impressions:
My camera was purchased from Modern Astronomy and delivered efficiently two days after I had placed the order. The first thing that strikes you about the camera is it feels quite heavy and solid in keeping with its industrial heritage. It is a no-nonsense technical object and has none of the flimsiness of a webcam. All parts of the casing are metal and the base is finished with a heavy 52x30x8mm metal plate containing 3 1/4" UNC threaded holes for tripod etc mounting. The camera body has a c-mount thread into which the supplied 1.25" nosepiece screws. Installation of the (supplied) drivers and control software on a Dell laptop running Windows XP was entirely straightforward and took no more than a minute or two.
Supplied capture software:
The camera came with IC Capture v2.0 which seems to be a capable camera control and image/video capture application. All camera settings may be controlled using this interface (see screenshot below) which is quite intuitive and well-supported by on-screen help. No written manuals etc are supplied but these are not really needed.
For planetary imaging the normal mode of operation would be to record video streams as avi files at the chosen frame rate although it is also possible to save individual frames as bitmap files if longer exposures (up to 60 mins) are used, in which case the fps setting is overridden.
Thermal noise characteristics:
Something that is quickly apparent when the camera is used is that it does warm up a bit (as with webcams etc). In a test at 20.5 deg C ambient room temperature with the camera on a table top, the camera casing heated up by about 6 deg C reaching thermal equilibrium after an hour or so (see graph). Operation outdoors with lower ambient temperature and a breeze might help to keep the camera cooler.
The effect of temperature rises of this order does seem to be relevant to the noise characteristics of the camera. In another experiement I captured bmp images of a 66 second exposure with the camera at different temperatures (left to equlibrate in a cool room, warm room, refrigerator etc before power up). The result shown below clearly indicates an increase in thermal noise between 18 and 25 degrees. It was also interesting to see that in this admitedly rather crude test, cooling the camera below 12 degrees did not appear to have much further effect on reducing noise.
| 66 second dark exposure with maximum gain; 65pixel square from lower left corner of frame magnified x3 | |||
 |
 |
 |
 |
| 2 deg C | 12 deg C | 18 deg C | 25 deg C |
Obviously random noise and hot pixel intensity is a function of exposure duration as well. The dark frames below indicate the appearance of some very hot pixels, marked background noise and what I assume is 'amp glow' at the upper left corner of the frame over a 30 sec exposure at maximum gain.
1sec dark exposure at max gain (@ 25deg C ambient temp) - just a couple of hot pixels
30 sec dark exposure at max gain (@25 deg C ambient temp) - several hot pixels appear plus noticeable random background noise and 'amp glow'
First light experiments:
Naturally I was desperate to try out this new camera when it arrived but unfortunately this had coincided with a spell of particularly unsettled weather with high winds and much rain. I managed a couple of quick sessions dodging the clouds, firstly with the cam on my C9.25 with a Barlow giving f48, and then on my 5" Maksutov with a 0.5 focal reducer.
First target was Mars which was imaged through a Baader IR pass filter. Even though this filter stops a lot of light, near-full gain gave very bright images even with a 4x Barlow and f48. Capturing at 30fps is very satisfying after being used to 5 or 10fps max with a webcam; this will make large frame numbers easy to acheive without any planet rotation effects becoming evident. I tried a number of the video codec settings and I'm not sure I have this quite right yet. It was striking that I could see a fair bit of detail in the preview screen with much less noise than I have either with my colour (unmodded) or mono-chipped ToUcams.
Mars 2007-12-05 C9.25 SCT @ f48; Baader 670nm+ filter; 1000 from 2000 frames in Registax4; very poor seeing
Next target a couple of nights later was M42. I soon discovered that it is a breeze to set up the capture software to take image sequences as long as you like. I kept the subs to 5sec only as I had this 5" scope on my old EQ3 mount and the tracking was fairly terrible. The onset of rain brought the session to an abrupt end. The result below is 17x 5sec exposures stacked in DSS and with levels adjusted a little in PS6 afterwards. This was a very half-hearted attempt at imaging a deep sky object, which is not really the primary purpose I bought this cam for, but I was quite impressed by how sensitive it was. I am looking forward to trying this again at a shorter FL, on a better mount with more time.
M42 - Trapezium 2007-12-07 5"Maksutov with 0.5FR giving c.f6; Orion skyglow filter; 85sec exposure
Update: 2007-12-10
A first attempt at LRGB imaging with this camera. I don't have a proper RGB filter set yet and the filters I used were quite bad quality; the result is interesting nevertheless. Red was used as the luminance channel. Subs were all 4 secs with dark frames subtracted.
| Blue - 15 x 4sec subs | Green - 14 x 4sec subs | Red - 19 x 4sec subs |
 |
 |
 |
M42 2007-12-10 120mm refactor at f5
Update: 2008-05-07
It's been a while since I did any more work with the camera but here are a few more images from a session on 2008-05-06. The M5 and M57 images were made using the C9.25 at f5; the M3 image had the C9.25 at f10. The Saturn image had the C9.25 at about f40. Deepsky images were acquired as bmp files and combined and dark-subtracted in DeepSkyStacker. The Saturn image was made from an avi file of around 3000 frames which were processed in Registax. The only filter used for all images was an IR/UV block filter.
Saturn - 3000 frames
M5 - 71x6sec subs with dark frames subtracted
M57 - 17x6sec subs with dark frames subtracted
M3 - 54x6sec subs with dark frames subtracted
Update: 2008-12-06
What kind of deep sky images can you make using just a DMK camera and a c-mount lens? This image of the North America nebula (NGC7000) in Cygnus is not very good but I have put it here to show how sensitive the camera is and how an image can be acquired in the most unpromising conditions. The imaging setup was camera+12mm c-mount (CCTV) lens fitted with LPR filter on a photo tripod (i.e. no telescope and completely static). To make matters worse, the target was at the time in the most heavily light polluted part of my sky. Despite all of this I was surprised to see how much of this faint H-alpha emission nebula could be resolved; it was even obvious on the monitor image!
NGC7000- 23x14sec subs with dark frames subtracted
To be continued....