Andy:
Since the sensors for digital cameras are linear in recording
information ( is this true ), then for a particular aperture, the
density of a white spot on a black background should be a function of
shutter speed alone. So . Take a series of photos of such an
object at set shutter speeds of 1/30th, 1/50th, etc., select the
density value for the shutter speed to be that which has been
measured by the spinning disk method ( that get's a baseline density
value for say 1/30th second ) and then calculate the other shutter
speed values from there.
This should get you values up to about 1/200th second or so.
Then reset the density value to the higher shutter speed and repeat
farther up the speed scale.
Repeat as often as necessary to complete the speed scale for the
particular camera's shutter speed range.
To check on linearity issues, a single shutter speed and varying
apertures should suffice to get the curve data.
To get the lighting value right, keep adjusting until the density
value ( 8 bit .JPG image ) of the white spot is something like 255
for each iteration of the process.
Cheers,
James
At 09:13 PM 11/17/2006 -0800, you wrote:
I was asked the following question and I replied below. I thought
some of you might be interested in the reply. I recall that Bob
Talbot was testing shutters in related ways.
> So, I am wondering how you would go about testing the shutter on
a digital camera. I am assuming they are plagued with the same
shutter issues that our film cameras are, but we only have pixel
values to deal with.
Good point. It is something that I thought about but not for long as
I must have gotten distracted by other things. Anyway, now that you ask ...
The shutter testers that depend on a clear light path through the
camera body obviously will not work as digital cameras do not
provide this capability. Sooooo ...
1. The simplest test would be to test as done in my high speed
class. Photograph a disc rotating at a known speed and measure that
resulting arc and divide that by the rotation rate in degrees per
second. We use a regular (LP vinyl) turntable and measure up to
about 1/100 second. But after about 1/30 the test becomes quite imprecise.
2. An alternative is to photograph a flashing light source such as
the display generated by some LED displays ... or even standard neon
signs ... by panning the camera while keeping the LED visible in the
finder. Counting the number of flashes of the image of the subject
by the rate the subject was flashing at (about 120 per second for
the neon tube and up to maybe 400 per second for the LED display
(set to the number 1 best) gives the exposure time.
3. An alternative to the above alternative is to use a calibrated
stroboscope to make an image of the slit (assuming a DSLR) of the
shutter ... without a lens on the body ... and at this point there
are two ways to go.
3a. Measure the distance one edge of the slit moves between
consecutive positions of the slit's image. Measure the size of the
slit. Divide the size of the slit by the rate the slit's edge moves
per second and that gives you the exposure time. The drawback is you
need a good stroboscope.
3b. Pick one of the slit images somewhere in the middle of the frame
(keeping magnification into account if measuring an enlarged version
of the cameras' frame size divided into the corresponding value of
the enlarged version) and determine its size in comparison to the
width or height (depending on which direction the slit moves) and
that ratio multiplied by the X sync speed of the camera will give
you an approximate exposure time. This can be refined (as I do in my
class) to get more accurate measurements.
Obviosuly you always need a "standard of measurement" that you don't
question. Turntable rotation rate, LED flashing rate, stroboscope
flashing rate, etc.
Yet another way is to photographn linear motion (as I do in my high
speed class) such as a car traveling at a known rate (you must
assume the speedometer is accurate!) and then, placing a small light
(such as a bare bulb car taillight) on the side of the car,
photograph the moving vehicle. The lamp's filament will leave a blur
on the record. Divide the length of this blur (again relying on
knowing the magnification of the print you are using) by the rate of
motion of the car.
4. Finally there is yet another method that depends on the use of an
oscilloscope but not looking through the camera but, instaed,
monitoring the X sync connection timewise. In the X sync mode the
sync contact closes immediately upon the exposure starting and the
scope will detect this. The X sync contact opens again once the
second curtain ends the exposure. The time base of the scope is set
so as to produce a square wave "image" of the time elapsed between
the closing and opening of the X sync contacts.
I am sure there are several additional methods! BTW, these
approaches can be used backwards also by assuming one knows the
exposure time (I usually just assume the shutter has been first
calibrated) I mostly just assume this since students tend not to
bring the same camera to school, etc. ... and precision in my high
speed class is less important than the related thinking process).
I hope this is of some use ... if you discover another approach or
if I made a mistake in my suggestions please let me know. BTW,
electronically controlled shutters tend to be much more accurate
than the older all mechanical ones. I would trust their timing and,
in fact, would use them as precision timepieces for certain applications.
cheerio,
andy
James Schenken
