High shutter speed accuracy?

[Ivan J. Eberle]

With the sound-card testers using the graphing in a sound editor like Audacity, it's rather simple then to average the mean opening and closing times, e.g. and measure only that range. Easiest way to do this I've found (assuming opening and closing propagation times are the same at either end) is to measure from the first instance of opening to the first instance of closing.

For critical use, it's probably best to reduce the aperture (or the distance of the light source) so that the signal from diode doesn't swamp the input ADC, and remains within range of the graph.

I bought an $8 phototransistor for my assembling own sound-card tester because it has a much faster on-off response than the cheapest ones. Don't have the spec handy at the moment but IIRC it's switching time is on the order of a few microseconds. (The resolution of the A/D recording sampling chosen in Audacity will be more of a limiting factor here.)

[Jorrit]

Struan, thanks for your pointers and insight.

[Ivan J. Eberle]

While I do like the ring of Struan's elegant answer much better than my own oversimplification, I don't yet know what the practical application of this is for quick and dirty shutter testing.

My limited understanding (admittedly imperfect) is that the taking aperture won't matter as much as it might have back in the 19th Century, because leaf shutters are now effectively placed at the nodal point of the lens-- with the light bending around the iris blades there to form a full and not truncated image, so that we're only changing intensity, not significantly truncating the image forming area described by the opening and closing of the shutter as might happen with a Packard type.

My own testing with the graphing seems to confirm this at different apertures, at least that the timing is unaffected and that the mean opening and closing times are unaffected.

But if I'm mistaken on this point I eagerly await someone to educate me further.

[Jorrit]

Ivan, I didn't see your post before I posted my reply. Sorry about that.

From a quick test just now, things seem to improve after desaturating or unswamping the signal. From my quick unscientific testing it seems that the afterspikes or shutter recoil as I so vividly imagined, seem to be an ill after effect of the swamping.
The after-spikes disappear with an unclipped signal.

I just realised that my test light will be perceiced by the phototransistor as a small point source, because I'm using a rather wide angled lens.
Perhaps we can get a better impression of the exposure when the entire field of view of the lens is covered with - say a softbox. Only then something can be said about it's influence on exposure. I'm not sure my modeling lights are strong enough though, but perhaps a quick sheet of paper will do just fine instead.

[bobwysiwyg]

I've been following this thread with interest. I use the photoelec and Audacity to check my shutters, including the focal plane shutters in my Nikons. It was interesting that of all my Nikons, the only one that exhibited the 'after spike' was my older Nikon F. I never could explain the cause to myself, yet it didn't seem to have any relationship to the true accuracy (or exhibited exposure) of the shutter.

[rdenney]

I just realised that my test light will be perceiced by the phototransistor as a small point source, because I'm using a rather wide angled lens.
Perhaps we can get a better impression of the exposure when the entire field of view of the lens is covered with - say a softbox. Only then something can be said about it's influence on exposure. I'm not sure my modeling lights are strong enough though, but perhaps a quick sheet of paper will do just fine instead.

The photo-transistor system isn't designed to measure intensity. It only measures switching, and any intensity effect there might be is filtered out by the DC filter on the audio input. That was Struan's main point. Thus, you can't measure light intensity.

To do a proper test of shutter efficiency, one needs to test the total light coming through the lens as the shutter opens and closes, integrating the total exposure over that time. There are, I expect, shop-grade shutter testers that do that (in addition to measuring opening and closing curtain speeds for focal-plane shutters). Given that few of us use the fastest speeds or the widest apertures, it's probably something for technicians to test and not required for those of us who just want to know if our shutters are way off.

I have tested a range of focal-plane shutters using a Calumet tester, and by positioning the sensor I was able to determine if the shutter open time varied across the frame. These were largish shutters and doing so was not that easy. It required using a DC light source that didn't pulse (as AC light sources do) and attenuating it sufficiently to just exceed the trigger threshold of the tester. But the Calumet tester uses the same technique as the transistor tester, with the output coupled to a micro-second timer that is started on the positive pulse and stopped on the negative pulse. Those testers cannot measure leaf shutter efficiency very easily, but they sell for a lot because they are useful nonetheless.

Rick "who rarely uses speeds faster 1/50 or 1/60 where it just doesn't matter" Denney

[Jack Dahlgren]

While I do like the ring of Struan's elegant answer much better than my own oversimplification, I don't yet know what the practical application of this is for quick and dirty shutter testing.

My limited understanding (admittedly imperfect) is that the taking aperture won't matter as much as it might have back in the 19th Century, because leaf shutters are now effectively placed at the nodal point of the lens-- with the light bending around the iris blades there to form a full and not truncated image, so that we're only changing intensity...

That is correct. With a wide aperture you are changing intensity over the duration of the shutter opening. This directly determines how much light is transmitted while the shutter is open and thus the effective exposure of the film.

This is why Struan mentioned integrating your measurements - in essence summing the light intensity over the time that the shutter is open.

Imagining the shutter as an iris might make it easier. As it opens over a period of time, the intensity goes from 0 to the maximum. Then depending on the shutter speed it lingers open for a time (longer for long shutter speeds, maybe almost nothing at short shutter speeds). If you plot the intensity over time it will be a mound of some kind with a pointy top for the shortest shutter speed and a flatter one for long shutter speeds. In the short shutter speed case, the total amount of light passing through the lens is less than it would be if the sides of the mound were not sloped. Unfortunately, a shutter is mechanical so it takes a material amount of time to travel the distance.

Now one way to shorten the time it takes to go from fully closed to fully open is to shorten the distance that the shutter has to travel. One way of doing this is using a smaller aperture. As soon as the shutter blades are clear of the aperture, the shutter is effectively as open as it is going to get, even if it has not physically reached its limit. The same goes for closing.

From looking at your charts though, I'm not really certain it is a big problem. The amount of time that the intensity is varying (your first pulse) appears to be pretty short in comparison to the duration of the exposure. I don't think you can trust it at all for measuring that opening time, but it seems like it is close enough to rule out that as being something you have to compensate for in your exposures. It is easy enough to be a half or a whole stop off in an exposure due to changing light, mis-measurement, reciprocity or other factors. I think it is interesting but not really significant for normal use.

Since it is a slope, you could pretty simply average it out by measuring from the peak of the pulse to the peak of the other pulse instead of from the start of the first to the finish of the last. But as pointed out, there are other unknown things going on in the devices you are using to measure it, so it may still have some error.

[GPS]

...

The amount of time that the intensity is varying (your first pulse) appears to be pretty short in comparison to the duration of the exposure. I don't think you can trust it at all for measuring that opening time, but it seems like it is close enough to rule out that as being something you have to compensate for in your exposures.

...

I think it's correct to say so.
One way to see it photographically (with no electronics involved) is to expose a piece of film put at much a shorter distance behind the lens than is its focal length (or put the film just a few mm of the lens exit pupil). In this way the film will get exposed just on a illumination circle not too wider than the lens rear diameter. Make a short (the shortest) exposure with the widest aperture and develop the film. You can see then if the center of the illuminated circle has much a different exposure than the edge of it. The difference would be given by the time the shutter blades were taken to open completely and to close completely. All the exposure in between those 2 actions would have no influence on the illumination difference between the edge and the center of the film (provided you didn't over saturated the film with the light).
I think the difference must be negligible.

[Struan Gray]

The amount of time that the intensity is varying (your first pulse) appears to be pretty short in comparison to the duration of the exposure.

The variation time is almost certainly set by the DC-rejection of the sound card. I doubt it will have much to do with the shutter blade travel time. Traditionally, audio circuits are handwavingly specced at 15 Hz - 20 kHz, but with an unspecified computer sound card the position and sharpness of the lower frequency cutoff is an open question.

I have tested some shutters with a lab-grade photoreceiver (from New Focus, if anyone cares). These are designed to give an analogue signal proportional to the average intensity falling on them, and have a known, and more than adequate, measurement bandwidth. I didn't have a storage scope with a data link, so I can't show you any traces here, but they were pretty much as expected.

The measurement is largely insensitive to exactly what you have as a source object, or the focus distance, but convention says that apertures are defined for on-axis light from infinity. You do need an extended source if you're trying to do radiometry, but for measuring the light intensity as the shutter opens and closes even a point source will do. Similarly, the focal length is not a big issue.

The only exceptions are if the light is bright enough to locally saturate the detector, or if wild bokeh varies the shape of the light spot so that parts of it miss a small area detector. The former requires a non eye-safe laser, which I hope nobody here is daft enough to try and use for this application; the latter is I suppose possible, but anyone experimenting with such lenses is likely to welcome any further contributions to out-of-focus-wierdness from the shutter inefficiency.