The problem is that the shadow of the edges of the slit are not
sharp. The shadow of any object has two parts, the umbra and penumbra.
One can see this with ordinary sunlight where the shadow is not sharp
edged and it is very apparent during eclipses where the shadow of the
moon comes on gradually. It is due to diffraction at the edge. The
size of the penumbra of the edge of the slit is affected by 1, its
distace from the film, the longer the distance the greater the penumbra
and more gradual is the time of passage of the slit. In other words the
_effective_ slit width is greater than the physical width so the time of
travel is longer and the exposure greater.
The second factor, which also affects the size of the penumbra, is
the angle of the light from the lens. This is determined by the stop
size and the distance of the lens from the shutter. The smaller the
cone of light from the lens, and the greater its distance, the smaller
the penumbra and the closer the effective slit width is to its physical
size.
Since the size of the penumbra is fixed by the above factors its
effect on the effective exposure varies with the slit width, the wider
the slit the less effect it has.
So, at slow speeds with wide slits the effect is slight. Also with
small stops where the cone of light from the iris is small the effect is
reduced. Also as the lens is moved further away from the shutter and
film, as in focusing for a close object, the effect is reduced. It is
also, of course, less for a long focal length lens than a short one.
In the case of a Speed Graphic or Graflex camera (which has a very
similar shutter), the effect is enough so that at the highest marked
speeds, i.e. 1/1000 second, the actual speed may be closer to about
1/700th depending on the f/stop.
The effective speed can be measured in a couple of ways. One, the
best if its possible, is to use a very small photocell at the focal
plane and display its output on an oscilloscope. The effective speed is
the duration between the half voltage points. Something similar can be
obtained using a shutter speed meter like the Calumet meter. The meter
is placed behind the shutter at the film plane and the source of light
is adjusted with the lens at the desired f/stop setting to just
extinguish the reading. The distance is measured and the camera moved to
0.707 or the square root of two times that distance, at which point the
light striking the sensor will be double that at cut-off. A measurement
made under this condition will give the "effective" speed. Of course,
this is valid only if the source of light approximates a point source
and if the sensor of the meter or scope is small compared to the
dimensions of the shutter slit. In the case of the Calumet meter it
works well for large focal plane shutters such as the one in the Speed
Graphic but is not so accurate for small focal plane shutters such as
those in 35mm cameras. However, where the slit is large this meter works
very well.
Note that the Speed Graphic shutter has a series of slits of
different widths which, in combination with variable travel speed of the
curtain, yield a large number of speeds although only perhaps eight of
them are actually useful. In the Leica camera there are two curtains,
one released when the shutter release is pressed, and the other after a
delay. The delay is varied to obtain the various shutter speeds since
it is equivalent to varying the slit width. In fact, the second curtain
is allowed to accelerate slightly to widen the slit as it travels to
compensate for the overall acceleration of both shutters. A very clever
arrangement.
The old Contax shutter also varies the slit width but also the
curtain speed with a more complex mechanism than the Leica.
While the shutter in a 35mm camera is closer to the film than in a
Graphic, it also uses smaller slit widths so has the same problem with
the higher speeds being less efficient and having an efficiency which
varies with lens focal length and speed.
On 10/21/2015 11:03 PM, Martin magid wrote:
I've always believed that the focal plane shutter speed is based strictly on the time it takes for the entire slit to pass any point on the film. If you look at a table of shutter speed settings for, say, a Speed Graphic or a Graflex, you will see that for any tension setting the speeds get slower as the slit gets wider. That's because it takes longer for a wider slit to pass any point than it takes for a narrower slit to pass that point.
In the Graflex/Graphic table, the shutter speeds for any tension are almost, but not quite, inversely proportional to the width of the slits. I suspect the "almost but not quite" is due to (1) the movement of the curtain is not constant since it starts at zero and (2) some kind of averaging is used.
Marty