Re: first images

From: Marco Langbroek (
Date: Tue May 19 2009 - 11:57:37 UTC

  • Next message: Scott: "Satobs 19 May 2009"

    Greg Roberts schreef:
    > I do NOT suggest you use the 28mm lens - the field is far too wide and 
    > this reduces the positional accuracy 
    Greg, Gerhard,
    A 28 mm lens in principle will do fine for satellite work. With the correct 
    astrometric software, the resolution is sub-pixel. Using a 50 mm lens and the 12 
    megapixel sensor of my 450D I get 5" to 7" (arcsecond!) astrometric accuracies 
    on stellar objects (tested it on asteroids). That is much better than needed for 
    good satellite positions. My off-track residues on satelllite trails are seldom 
    larger than 0.01 degree, or 0.5' (arcminute). Sometimes I use a 17 mm lens (see 
    below) and the accuracy is then still good.
    I mostly use the 50 mm and sometimes a 17mm (VLEO or objects with large time 
    uncertainty). For HEO objects I use a 100 mm but solely because that lens goes 
    fainter in objects.
    The best argument to go to a slightly higher focal length is actually not 
    resolution, but that you will catch fainter objects. It is a trade-off however 
    with smaller FOV, which means your aiming has to be better (I use a 8 Mw green 
    laser parallel to the camera optic axis to aim. It gives a visible beam at 
    night, showing you where the FOV is pointing. It is a simple green laserpointer 
    that you can buy for a handful of dollars nowadays)
    > With LEO sats you will have problems re the time of the image as ideally to 0.1 seconds is required - Marco achieves this with his digital camera so is the man to approach re this aspect.
    Indeed, it is the time resolution that sets the limits to your accuracy, more so 
    than the astrometric accuracy of the camera + lens.
    You need to calibrate the timing of your camera, plus a good time source.
    In my case, I use a radio-controlled clock which receives and synchronizes to 
    the DCF77 signal of Mainflingen (Frankfurt). I am not sure whether something 
    similar exists for South Africa. Before each observing run, I forcefully have 
    the clock re-synchronize by shutting it's power off (taking the battery out and 
    then putting it in again).
    What you do, is that you attempt to press the shutter button of your camera 
    (wire release!) on a round time, e.g. 20:12:20.00, as the second digits change 
    on the display. I have found that making a series of "mock" presses during the 
    last few seconds of the countdown helps you attain the rhytm of changing seconds 
    and improves accuracy.
    Of course, you have to calibrate the camera timing first. There is a small lag, 
    due to the camera processing electronics, between you pressing the button and 
    the camera actually opening. In my case (EOS 450D), that is approximately 0.30 
    seconds. Besides of that, for shutter speeds > 1 s the accuracy of the indicated 
    exposure times is not quite accurate. For example, the "10 second" shuter speed 
    of my 450D is 10.05 seconds in reality, and the "15 seconds" is 16 seconds (!) 
    in reality. This matters if you measure both start and end of the trail.
    Calibration is best done on satellites with tightly fixed orbits for which 
    accurate tle's are available. The Iridium satellites are very suitable. Just 
    spend a few days catching and measuring as much of those as you can, and looking 
    at your residues in delta T (e.g. using Scott Campbell's SatFit software) and 
    you'll see the pattern appear of your equipment-induced delta T's, which yields 
    your time lag corrections.
    I second Greg's remark (and already sent you a note about that) of changing your 
    F-settings. I use F2.8, not F5, combined with a 10 second (actually 10.05s) 
    exposure and either 800 ISO or 1600 ISO (the latter for very faint HEO).
    - Marco
    Dr Marco Langbroek  -  SatTrackCam Leiden, the Netherlands.
    Cospar 4353 (Leiden):   52.15412 N, 4.49081 E (WGS84), +0 m ASL
    Cospar 4354 (De Wilck): 52.11685 N, 4.56016 E (WGS84), -2 m ASL
    Station (b)log:
    Subscribe/Unsubscribe info, Frequently Asked Questions, SeeSat-L archive:

    This archive was generated by hypermail 2b29 : Tue May 19 2009 - 11:58:34 UTC