High shutter speed accuracy?

[Jack Dahlgren]

On my first 90mm f/5.6 in Copal 0, I've found the 1/500 to be around 1/270.
It doesn't bother me (yet), and I did find some interesting shutter "recoil" at high speeds. I'm wondering if that could be used to some benefit. In practice it would likely be like a weaker second exposure, with a delay.

I also found the working aperture to be of no influence on the speed. 1/270 stays 1/270. I've tested from wide open to closed. The pic above is a cropped screen capture from Audacity using a photo transistor. So from the graph it looks like the aperture only has influence on the width of the first pulse.

I'm having trouble understanding what those pulses (especially the ones to the right) mean. What is it you are measuring and how? Voltage level through a photo-transistor? What is your setup? I've been thinking of building my own shutter tester and am curious about what you have put together.

[Jorrit]

I'm having trouble understanding what those pulses (especially the ones to the right) mean. What is it you are measuring and how? Voltage level through a photo-transistor? What is your setup? I've been thinking of building my own shutter tester and am curious about what you have put together.

Ah sorry, yes the y-axis is voltage (in unknown scale), and x-axis is time.
It's this (PDF) circuit belonging to this thread. You basically hook the transistor to the mic input on your soundcard, cock shutter, shine a bright light on the lens, release, and use a sound capture prog to do the measuring.
To be honest, I didn't build it myself, I was lazy and ordered one on Ebay after searching for "shutter speed tester"

The phototransistor seems only to register changes in input signal. So a continuous light will yield a flat line, as will continious absence of light.
First spike is shutter open, second negative spike is shutter close. The third and fourth spikes can only be another open-close sequence. I assume it's a sort of recoil from the "main" release. It only happens at high shutter speeds. When checking a slow 1 second speed, there are no after-spikes, so to speak.

[venchka]

The secodary spikes vary with aperture. That's curious.

[Jorrit]

The secodary spikes vary with aperture. That's curious.

Above is a new capture: tried 3 times @1/500, with the same aperture @wideopen. So by keeping aperture the same, the result is also variable. In both width and amplitude.
So far for a reliable hidden Copal multi exposure feature.

[GPS]

...
First spike is shutter open, second negative spike is shutter close. The third and fourth spikes can only be another open-close sequence. I assume it's a sort of recoil from the "main" release. It only happens at high shutter speeds. When checking a slow 1 second speed, there are no after-spikes, so to speak.

Or rather the other spikes have their source somewhere in the electronic circuit itself...

[Jorrit]

You're right to succesfully attack my rather too strong proposition that the other spikes "can only be" such-and-so. I got carried away by the possibility of a secret undocumented 1/500 = multi-exposure feature

[GPS]

Didn't want to attack, just knowing that mechanically it would be next to impossible to have such a recoil... :-)

[BetterSense]

I also found the working aperture to be of no influence on the speed. 1/270 stays 1/270.

I think you are not thinking about this fully. Your photodiode probably turns on as soon as the shutter begins to open. The fact is that it takes a finite amount of time for the aperture blades to FULLY open. During the time in which they are in the process of opening, the effective aperture of the system is not the full marked aperture, because the shutter blades are blocking the aperture partially. At small apertures, this effect will be negligible, but at full apertures the shutter blades can block the aperture a significant amount of the time during their opening-and-closing. Thus even though the "shutter speed" may not change with aperture, the amount of exposure the film gets WILL change with different apertures, and thus sometimes high shutter speeds are deliberately slow, in order to give the expected exposure at all apertures, with the penalty of slight overexposure at smaller ones.

[Jorrit]

Your photodiode probably turns on as soon as the shutter begins to open. The fact is that it takes a finite amount of time for the aperture blades to FULLY open.
[...] thus sometimes high shutter speeds are deliberately slow, in order to give the expected exposure at all apertures, with the penalty of slight overexposure at smaller ones.

I agree with you that the shutter needs some time to fully open.
Let's look closer at the following graph. Copal 0 shutter set at 1/500, aperture at f/5.6.



My proposition is that the first pulse (measuring some 1/3000s) is the traveltime of the shutter going from closed to open. The phototransistor seems to only register CHANGES in light. So while the shutter is traveling, the phototransistor registers changes in light, and at the end of the 1/3000s doesn't so much anymore. The line is more or less flat. Meaning no changes in light = open shutter.
Now the duration of this more-or-less flat line is some 1/290s.
After that, the transistor starts to register changes again butin the oposite direction: the shutter is closing.

So even if we almost entirely exclude the 2 shutter travel times, the shutter stays open a lot longer than the advertised 1/500, leading to over-exposure at any aperture value.

The weakness I see in my own proposition is the lack of information about the y-axis. Are the values on a lineair scale? A log scale? Or perhaps an exp scale? That matters, because I've placed the white line at the end of the first pulse at an arbitrary position.
So we don't know for sure whether the end of 1/3000s is really a more-or-less open shutter. I asume it to be around f/5.6. But i can't tell. It may very well be still f/16 or something.

The second set of open-close is another story that intrigues me.

[Struan Gray]

Jorrit, I don't know where your second pulse is coming from, but like GPS I suspect it is electronic in origin.

There are two things you need to realise to understand your curves. First, that your sound card is not DC-coupled. In fact, it includes a circuit designed precisely to reject DC. This is because DC is bad - sometimes very bad - for audio amplifiers and loudspeakers/headphones. This is why you only see changes: your detection setup can only measure changes.

Second, the response time of the phototransistor is limited, as is the input circuitry of your sound card. In an ideal world you would be able to take your traces and integrate them to find the instantaneous light level on the detector, but the limited bandwidth of the detection system makes that impossible (even if you added some ND to avoid the saturation you now have).

As others have said, a leaf shutter takes a finite time to open and close. A plot of the light intensity versus time will not be a perfect square pulse, but will have sloped lines at the beginning and end. At full aperture, the leaves only just reach the edge of the aperture stop in time to turn around and start closing again: the pulse shape becomes, roughly, a triangle. The total exposure is the integrated light intensity over the whole pulse. Since the area of a triangle is 0.5*base*height, it follows that the start and end times for a triangular pulse need to be twice as far apart as they would be for the same exposure from a square pulse from some ideal shutter that could open and close instantaneously. It therefore makes perfect sense that your measurement - which only accurately measures the start and stop point of the shutter sequence - gives a shutter speed that appears to be a factor of two out.

At small stops the shutter blades don't move in any different way, but they clear the aperture stop much faster. The same slope is there on the ends of the pulse, but now the blades spend most of their time moving outside the clear hole of the aperture, so the slopes are just a small correction to the total exposure. If you had a true plot of light intensity vs time, a small stop just cuts most of the top off the triangle seen at wide apertures to form a flat-topped pulse with sloped ends. There is very little difference between that and an ideal pulse, and the exposure is longer than it 'should' be - again, by a factor of two or so.

The marked shutter speed can only be correct for one stop. Convention seems to be to make the marked speed that which gives the same exposure at full aperture as a perfect shutter that can open and close instantaneously, even if that speed does not correspond to any actual timing within the shutter. How this convention was arrived at I don't know, but you can find discussions of the effect going back to the C19th.

In your original traces, all the information about these effects is hidden in the flat-topped saturation of your measurement system. If you add some ND, or turn down the brightness of your test light, you should see changes in the shape of the curves as you change the aperture. Remember though that the shape may be largely determined by the bandwidth of your phototransistor/sound card combination, so the changes may still be subtle. The same phototransistor and a real oscilloscope should show the effect clearly though.

If you want to google, these effects come under discussions of shutter "efficiency".