The ESI shutter was custom-made by the UCO/Lick Observatory shops. It is a dual-curtain design similar to what is used in single-lens reflex (SLR) cameras. Both curtains or blades start on one side of the aperture. An exposure is started by one blade moving towards the other side of the aperture. One integration time later the other curtain or blade starts to also move to the other side. This assures that all parts of the field will get equal exposure, and that the actual integration time is very close to that requested.
The shutter blades take 150 msec to cross the aperture. The effective integration time is in principle independent of this speed. If you wanted a 50 msec exposure you would trigger the second blade 50 msec after the first. As long as both blades have similar acceleration and velocity profiles across the aperture, each field point will get a 50 msec exposure. The second blade will obviously start crossing the aperture before the first blade has reached the other side. Unfortunately, the Leach-2 firmware prevents exposures of fractional seconds, so 1 sec is the shortest shuttered exposure you can take.
In the original measurements in the Lick shops this was repeatable to +/- 10 msec. Since 1 sec is the shortest exposure which can be obtained, this implies an accuracy of at least 1%.
However, we have had problems with the shutter sticking. Work done in January 2000 hopefully has fixed the problem, but the potential for wear and tear on the mechanism increasing the 10 msec accuracy exists. When in doubt, try to obtain calibration data which will limit the effect of a slightly variable exposure time. For instance, take relatively longer (10 sec or more) exposures rather than using 1 sec.
There is a separate Web page for troubleshooting a faulty shutter.