Repeatability

The repeatability of the ESI shutter was tested in two ways; one by taking multiple 1 second exposures of a star field, and the other by taking multiple exposures of an arc lamp with various exposure times. The latter are the more extensive data, so we concentrate on them.

The intensity of one of the arc lines is shown below for a set of twenty 1 second exposures. The very first exposure is not shown, because the shutter stuck. (As is typical for ESI, the first time the shutter is triggered after a long rest, it sometimes does not work properly.) The second exposure probably suffers slightly from the same effect.

Note the odd/even behavior. Odd numbered exposures are brighter then even numbered exposures. The ESI shutter operates like the curtain shutter found in many cameras, except that it does not "rewind" and start always from the same side. Instead, its two blades travel left-to-right for one exposure, then right-to-left for the next. Apparently when the shutter blades move in one direction they create a longer effective exposure time. The difference is approximately 47 millisec as tested. If the difference is influenced by gravity, different behavior could be found at different telescope/rotator positions.

We can measure this small difference because the shutter repeatability is significantly better than this. Counting only the odd-numbered frames, we measure a repeatability of 7 millisec. For the even-numbered frames we measure 2 millisec. Notice, however, the occasional larger deviations. Keep in mind, too, that the measurements are susceptible to drifts in the arc line intensities due to current fluctuations and such.

Linearity

We expect a 2 second exposure to produce fluxes twice as high as a 1 second exposure. There is a slight discrepancy for the shutter moving one direction compared to moving the other direction, so already we know that a requested 1 second exposure will not always be exactly one second. Below is shown a linearity curve, with the expected slope of unity removed. This plot shows the fractional error for exposures of 10 seconds and less, assuming that exposures of 20 seconds are correct. The odd/even effect has been removed.

Note that even at 10 seconds the exposures are not completely linearly related. (In other words, a 20 sec. exposure does not contain twice the counts of a 10 second exposure.) Part of this could be due to nonlinear response of the detector. In this case a 20 second exposure would contain less than twice the counts of a 10 second exposure, as measured.