Leaf Shutter High Speed

[Doremus Scudder]

Robert's comment about aperture size is quite relevant here. At a fast shutter speed where the opening/closing of the leaf shutter takes so much of the total "open time," the efficiency will be markedly different between a wide-open aperture (least efficient) and a very small aperture (most efficient). I would think it would be important to do your shutter testing at the aperture(s) you use most.

I know that I rarely, if ever, make a photograph at an aperture smaller than f/22. I'd say more than half of my exposures are made at an aperture somewhere around f/32 (give or take a third of a stop), so I'd likely test at f/32.

Having some way to measure the total volume of light instead of just the time between the first glimmer of light to make it through the opening shutter and the instant it goes dark would seem to be necessary to really get meaningful exposure data with a leaf shutter at high speeds. Maybe Dan's suggestion about asking your reversal/transparency film is going to end up being the most practical after all...

Best,

Doremus

[gnd2]

I find this an interesting curve. It is indeed more what you would expect. Could you please give the horizontal axis and what setting the shutter was set at?

Horizontal axis is sample number. Sample rate is about 9615.3846 samp/sec. ADC is in free-run mode, 16 MHz clock source / 128 prescaler / 13 clock cycles per conversion.

Lens was a Mamiya Sekor C 127mm, wide open at f/3.8 and 1/400 shutter speed (it's fastest setting).

[Havoc]

Horizontal axis is sample number. Sample rate is about 9615.3846 samp/sec. ADC is in free-run mode, 16 MHz clock source / 128 prescaler / 13 clock cycles per conversion.

Lens was a Mamiya Sekor C 127mm, wide open at f/3.8 and 1/400 shutter speed (it's fastest setting).

Ok, so about 0.0001s between samples and your shutter time should be 0.0025s or 25 samples. Now if you take that the slopes are about symmetrical around the 100 level (50% level) then for "total light" if you measure between the 50% points you get something like 38 samples or 1/263s. A bit faster than half speed. Very rough from the graph on the screen.

[gnd2]

Ok, so about 0.0001s between samples and your shutter time should be 0.0025s or 25 samples. Now if you take that the slopes are about symmetrical around the 100 level (50% level) then for "total light" if you measure between the 50% points you get something like 38 samples or 1/263s. A bit faster than half speed. Very rough from the graph on the screen.

Yes, but this is still an imperfect measurement. As the measurement improves, the slopes lean in towards each other making a tighter peak at the top. Assuming a perfect measurement would be very close to an ideal triangle, the effective shutter speed would be 1/2 the total time.

For the total time, extrapolate the linear approximation of the slopes down to 0 amplitude and determine the time between those points. It's tends to hover around 1/180.
As the slopes approach an ideal peak, the 50% level approaches 1/2 that, or 1/360. Close enough to 1/400 for rock and roll, eh?

As you stop down from wide open, the effective shutter speed would decrease from 1/360 to nearly 1/180 (the actual shape with the aperture closed down would start to look like the plots I have).

[Eric Woodbury]

The "half-max" levels are not well-known here as the photo sensor is probably acting as a switch and certainly not linear between OFF and ON. Perhaps the 'switch' is fast enough and this doesn't matter, or perhaps the rise time is dominated by the RC time constant of the circuit. A way to check is to measure the ON time of a strobe flash. Strobes have extremely fast risetimes.

If the photo sensor were a photodiode and the circuit a TIA, then the data would hold useable amplitude information.

[gnd2]

The "half-max" levels are not well-known here as the photo sensor is probably acting as a switch and certainly not linear between OFF and ON. Perhaps the 'switch' is fast enough and this doesn't matter, or perhaps the rise time is dominated by the RC time constant of the circuit. A way to check is to measure the ON time of a strobe flash. Strobes have extremely fast risetimes.

If the photo sensor were a photodiode and the circuit a TIA, then the data would hold useable amplitude information.

I took your previous post at face value because I didn't want to get into the device characteristics, but since you've mentioned it again...

No, it's not acting as a switch. It's wired in a common collector configuration with the emitter resistor value set to keep it in the active region up to the maximum light intensity received. So it is actually pretty linear, except obviously near dark, which you can see in the plots.

It should be obvious from the first plot that the response time is plenty fast enough for our purposes.

[mmerig]

Yes, but this is still an imperfect measurement. As the measurement improves, the slopes lean in towards each other making a tighter peak at the top. Assuming a perfect measurement would be very close to an ideal triangle, the effective shutter speed would be 1/2 the total time.

For the total time, extrapolate the linear approximation of the slopes down to 0 amplitude and determine the time between those points. It's tends to hover around 1/180.
As the slopes approach an ideal peak, the 50% level approaches 1/2 that, or 1/360. Close enough to 1/400 for rock and roll, eh?

As you stop down from wide open, the effective shutter speed would decrease from 1/360 to nearly 1/180 (the actual shape with the aperture closed down would start to look like the plots I have).

Why would a perfect measurement be necessarily triangular? This could be so for very fast shutter speeds, but not for slower ones (say 1/30 s or slower), unless the blades opening and closing time is proportional to the shutter speed, which would be a lousy shutter. Eventually, with a higher sampling rates, and good linearity,the shape of the curve would stabilize, and it could have a flat top.

The second graph you have may be pretty close to reality. The system you have seems well thought-out, and you have most of the bases covered.

[gnd2]

Why would a perfect measurement be necessarily triangular? This could be so for very fast shutter speeds, but not for slower ones (say 1/30 s or slower), unless the blades opening and closing time is proportional to the shutter speed, which would be a lousy shutter. Eventually, with a higher sampling rates, and good linearity,the shape of the curve would stabilize, and it could have a flat top.

I was referring specifically to the fastest shutter speed with the triangle comment. At the fastest speed, it starts closing again as soon as it reaches fully open, so a triangle. At slower speeds it lingers fully open for a period of time before closing and becomes trapezoidal. The slope remains the same, but as the shutter speed decreases, the time between open and close gets longer and the slope becomes less significant.

[mmerig]

Thanks, I did not realize you were referring to the fastest shutter speeds only.

Maybe I am missing something, but I wonder why manufacturers don't publish shutter efficiencies or the graphs themselves for their shutters. There is plenty on lens performance. Even Ansel Adam's "The Camera" book has only a theoretical graph demonstrating shutter efficiency (although he does report a range of efficiencies). Digital storage oscilloscopes have been around since the 1980's, and something like this could have been used to do the job.

[Andy Eads]

The efficiency of the shutter plays into all this. The 1960 edition of the Focal Encyclopedia of Photography defines it as the ratio of the light (the shutter) transmits during the exposure to the amount that it would transmit if it were fully open for the whole duration of the exposure. So a shutter might be open for 1/500 sec. but only transmit the amount of light expected at 1/1000 sec. That shutter would be 50% efficient. The rate of speed the blades or curtains can achieve is the limiting factor. Lower mass blades and higher force improve efficiency. The PQS shutter Bob Salomon cited is a fine example of a high efficiency shutter. In a 50% efficient shutter, lots of activity can be going on that would register as exposure, negating the usefulness of the 1/1000 sec. shutter speed for many applications. My memory is foggy here but Zeiss made a very efficient leaf shutter for its aerial survey cameras by having multiple rotating blades that did not "uncap" till they were up to speed. I remember the surprise I had as a 14 year old seeing the stopping power of a short-duration electronic flash compared to what was possible with ordinary shutters.