This page will assume you are an amateur astronomer with some experience taking and processing CCD images. The procedures for doing this are described in the AAVSO CCD Observing Manual, except that it is not essential that you be able to perform transformations. Also, see our Bibliography, and Software Resources pages.
Equipment
For observing EBs, the equipment doesn't differ from other CCD imaging. You will be taking images of the same star field continuously through the night, so autoguiding serves not only make for better images, but for keeping the field centred through the night. Differential photometry through a single filter is the most common, and an effective observing method, so a filter wheel is not essential. For equipment, you need a telescope (for DSLR, this could be your telephoto lens), and a mount capable of following a point in the sky (though some DSLR photometry of bright stars can be done with the camera stationary). The mounts most commonly used are equatorial, though alt-az mounts driven on both axes can be used. Fork mounts or German equatorial designs are the most common. With the latter, a mount that can continue tracking through the meridian is best. With a German mount that cannot track through the meridan, imaging must be halted at the meridian and the mount 'flipped'. Very often the resulting photometry shows a discontinuity at this point, creating difficulties in using the data. This discontinuity shows there a problem with the flat-fielding of the images, sometimes caused by 'flop' of the optical system, or possibly a stray-light problem. Fork mounts, or German mounts that can track through the meridian, hide this problem.
There is one additional item that will greatly increase your observing: an observatory. This does not have to be an elaborate dome! It could be as simple as a tripod or pier firmly set in place that your mount can be put back on, retaining polar alignment, perhaps with some of the cables left in place. Or a shed on wheels that can roll away from a tripod or post that carries the mount and telescope. Any provisions that reduce the nightly setup and teardown time will result in more observing.
Method, and how long to observe
The main method is differential time series, preferably through a standard photometric filter (typcically a "V" filter). Shoot the same field continuously, typically with exposures of 30 seconds to 5 minutes, back to back, throughout the duration of the eclipse. The exposure length is chosen to get high "counts" for your target star and the comparison and check stars, without taking your camera into its non-linear response region. The images must be "reduced" or processed through standard dark subtraction and flat fielding. Good flatfielding technique is often the most difficult step. Some sort of observatory helps with flatfielding: if you can leave the camera and filters attached to the telescope tube, with no mechanical disturbance (dis-assembly and re-assembly), then you can probably get away with not taking flat field images every night.
The length of the observing run is important. The information about the time-of-minimum comes from the sloping parts of the eclipse light curve, i.e. the ingress and egress. If the eclipse is total (flat-bottomed lightcurve), that segment contributes nothing to determining the ToM. An observing run that spans an hour of ingress and one hour of egress (not counting totality) is probably an absolute minimum, and in fact is probably too short. A longer time series is definitely recommended. The goal is accurate data that future researchers will find trustworthy, so don't shortchange your legacy by generating lots of questionable data. It would be better to observe each eclipse from when the brightness starts to fall from its out-of-eclipse level, right down through the eclipse and back up to close to out-of-eclipse level again. That is a subjective statement, and highly problematic for stars with eclipse durations that exceed the length of night! The main point is that you need to record an amount of variation during ingress and egress that is much much greater than the noise in your data.
If you are observing the "Legacy EBs", Gerry Samolyk provides a recommended minimum length of observing run in the ephemeris. For other stars, there are several resources providing information about the duration of eclipses. They all ultimately draw on the same original sources, so it's a matter of what is the most convenient data source for the observer. The GCVS lists eclipse duration for some EBs. VSX imported the GCVS data, and has more up-to-date information in many cases. Finally, the data compilation by Avvakumova et al., 2013 contains that information, as well as length of totality for both primary and secondary eclipses, when the information is available. The Avvakumova et al. catalogue is available via VizieR as J/AN/334/860, or search on "Avvakumova". Unfortunately, in all of these sources, the eclipse duration information is not present for many of the stars.
Timekeeping, and Record Keeping
You must have a method of accurately setting your computer's clock, as it is used to "timestamp" all the images you take, and will ultimately determine the ToMs you produce. Most observers set their computer's clock by way of the internet. You should strive for an accuracy of one second or better, and independently test that. E.g. set the computer clock via the internet, and assess the accuracy of that by way of radio time signals. Although it will be too late to save your data, a check of the computer clock at the end of an observing run will confirm the clock has been okay through the run. Sometimes computers running older versions of Windows fail to "roll over" the date at midnight, especially when downloading data from a CCD camera, so it is useful to check the date has rolled over properly (if necessary you can correct the problem while reducing your images, as the time can still be correct even if date rollover fails).
You almost certainly want to use FITS format for your images, as this is what most photometry packages require. Whatever software you use, it should record the time the exposure was taken, and the length of the exposure in the header portion of the image. It is very important that you determine whether your software is recording the start time, mid time, or end time of your images. Ultimately, what you want to use are mid-times of images. If you get this wrong, your reported times will be significantly in error, because on a good time series, it is possible to determine a ToM with a precision substantially better than an exposure length. A very good ToM has a precision and accuracy on the order of 0.0001 days, which is 8.6 seconds.
The images you are taking are scientific data, that others will rely on in their study of these stars. It is important that you know how the data was derived. Keep a log of each night's observing session, relating your target field to the filenames of the images you store on your computer. Develop a file-naming convention that makes sense to you, and is compatible with your image processing software. Archive your images for future access and re-analysis. You may discover you've been making a systematic error that you want to correct at a future date, or your results may show an interesting effect that someone wishes to investigate in detail.
Next: photometry.