Camera settings for time accuracy
Our aim here is to reduce shutter lag to a minimum. Ideally, the shutter lag that remains should be consistent and repeatable.
To achieve this, we need to reduce the amount of work the camera does by:
Use mirror lock (if available)
Don't use auto focus.
Maximum aperture requires no mechanical movement so using a 50 mm lens at F 1.8 may be faster than at F 8
Ensure the battery is fully charged.
Use Bulb or Manual exposure so that the camera does not have to calculate aperture and shutter speed.
DSLR Trigger Intervalometer settings
Do calibrate the shutter open and close lag. See Calibrate shutter open lag and Calibrate shutter close lag
Use Bulb mode in preference to Manual exposure mode. In most cameras, their exposure end time becomes increasingly inaccurate as the exposure length increases. In Bulb mode, the shutter close lag can be calibrated out and the trained PC clock provides the accuracy.
Ensure that the time interval between shots is long enough so that the camera never has to buffer images. On some cameras, the shutter lag increases once the camera starts buffering. Make sure that the camera 'Write' LED goes out before the next shot starts. Using a fast memory card can help, but only up to the camera's maximum write speed.
The camera may fail to fire if the buffer runs out completely. This will cause the shutter button and mirror lock button presses to get out of sync. Avoid this at all costs!
Use RAW only (no jpg). To create a jpg image, the camera has to process the RAW which is CPU intensive. Saving both RAW and jpg would increase the time required to write the data to the memory card.
Computer setup
Train the PC clock, ideally via a NTP time server such as Meinberg NTP and NTP Time Server Monitor (which uses multiple NTP time servers for excellent accuracy) or if an internet connection is not available, by using a GPS dongle and software, such as Ublox-7 GPS module and NMEATime2 software.
If you are using a GPS dongle with software, do calibrate the GPS dongle time delay. The software should provide a way of doing this. It is likely to be around 70 ms. When calibrating, use an NTP time server that is within a couple of hundred miles of your location.
Once the PC clock has been trained, don't allow the PC to go to sleep. The time errors introduced during a PC sleep can be dramatic.
If possible, try not to use multiple USB hubs between the PC and the USB switch.
Photographing a satellite
These hints assume that the DSLR camera is either tracking the stars, or on a stationary tripod.
The exposure needs to be long enough to record plenty of stars. One second usually works well. The stars are needed in order to plate solve the image (calculating the image scale, orientation and the RA and DEC of the image center.)
Use maximum aperture (e.g. 50 mm at F1.8). This will produce a 'softer' image, but this has little affect on the astrometry accuracy. The extra aperture will record fainter satellites and more stars.
The exposure should not be too long. Since the satellite is moving, a longer exposure creates a longer satellite trail, but does not increase its brightness. But it does increase the sky background brightness. Hence shorter exposures provide a better signal to noise ratio.
Each photograph provides two data points at the start and end of the satellite trail. Ideally take as many photos as possible while the satellite is within the field of view to maximise the number of data points. Short focal length lenses (e.g. 50 mm) provide the necessary field of view.
Determine where to point the camera ahead of time. The Heavens Above database provides star charts and maximum altitude.
Set the ISO to a high value. Although this does not affect sensitivity, it can significantly improve signal to noise ratio, but at a cost of reduced dynamic range. Since most satellite trail photos are likely to be around 1 second, the reduced dynamic range is unlikely to be a problem.
Analysing the images
This stage is optional.
Plate solve the image. For example, this can be done within software such as PixInsight, or via http://astrometry.net/use.html
Convert the file to FIT format and use the free software SAOImageDS9 Check that the FITS format uses the correct coordinate origin. SAOImageDS9 expects the origin to be the lower left. PixInsight defaults to upper left, but this can be changed within its Format Explorer (double click on FITS to display the FITS Format Preferences).
Use SAOImageDS9 to measure the RA & DEC of the trail start and end points. Note that the trail's ends are not square, but blurred semicircles with a diameter equal to the trail thickness. This is due to the seeing and the optical system. It is necessary to judge were the center of these circles are in order to measure the real trail start and end points. Use SAOImageDS9's magnified window, and move the pointer with the cursor keys for sub pixel accuracy.
Wide angle lenses (50 mm and below) can suffer from significant distortion at the edge of the field of view, so it may be worth ignoring trail points that are too close to the edge.