Initial Shutter Tests

I should mention that at this point I’ve done some work on the project and am slowly updating the blog, so if you’re working on your own project, don’t be discouraged by the appearance that I am working at a feverish pace.

The first thing I started to work on was the shutter. A few years before, I had bought an Ilex size 3 (an Ilex Oscillo-B no. 3) shutter on eBay and, to my luck, it was actually a shutter with no timing mechanism – simply an open/close lever.

The photo above shows the shutter intact. Below is an image of the shutter with the case off, showing how the blades are actuated via the lever’s turning of ring-shaped slotted cam which physically moves the blades themselves.

Although this shutter was not in a condition to be as frictionless as possible, I figured I could at least do some tests with it, actuating the lever with solenoids for both opening and closing the shutter, thus having digital timing on the mechanical shutter. The idea is to use small solenoids, and voltages as high as possible to maximize the amount of current going through the coils. It’s fine to exceed the ratings of the solenoids in this case because we are pulsing them. If we exceeded the ratings for too long, the wire insulation may start melting and if the coil makes contact with itself you start to lose turns and the solenoid becomes weaker.

The solenoids I chose I bought for $1 at the local surplus shop (which is now sadly going out of business). I think they were rated for 24 volts, but whatever the rating, they had 100 ohms of resistance. The biggest power supply I had at my disposal went up to 60 volts. There was another that went much higher, but it couldn’t source the current the solenoids would draw, so there’s no point. In fact getting large amounts of current like this is not trivial; in pulsed devices (such as camera flashes) it is achieved by charging a capacitor.

The solenoid current was switched by some power MOSFETs that were rated for 20 amps at 400 volts, or something similar. The very simple circuit would allow me to open and close the shutter with TTL signals, where the pulse widths were the “on” times of the solenoids, and of course the time between the leading edges was the delay between me opening and closing the shutter. The timing signals I got from a PCI-6602 card from National Instruments (mind you I was doing my PhD at Caltech at the time, which is where all this equipment was coming from) via a custom program that I wrote myself.

At first I tried to measure the shutter lag by shining a flashlight at the shutter and having a photo-diode on the other side, so I could look at the trace of the photo-diode voltage relative to the trigger pulse from the PCI-6602 on an oscilloscope to measure the shutter lag. I got some numbers, but this didn’t really tell me when the shutter was fully open, and since I knew someone in the group was borrowing some high speed digital cameras, I came in on the weekend and took some footage at 10,000 frames per second. The photo below shows the setup, with the shutter case closed (the videos were taken with the case open).

You’ll notice it’s quite rudimentary. I drilled a hole (two actually, at different distances from the pivot point) through the actuating lever, and put a #2 screw in there, and then made hooks out of paper clips to connect the lever to the solenoids. The solenoids were spaced so that the bodies were just beyond where the plungers would need to be for the shutter to be fully closed or fully open. I’ve uploaded one of the videos here (you may have to download it to your computer to play it properly and you may need the Indeo 5.x codec, available here); the rest are all similar so it’s irrelevant. They are not very informative either, but you’ll notice I reduced the play between the lever and the ring actuator by sticking a piece of round tubing in there. By counting the number of frames between first detecting any movement and the shutter being fully open, I could estimate the shutter lag (since I did not have a view of when the trigger pulse was actually sent, I could only assume that the solenoid started moving immediately). On a side note, I compressed the video with Virtual Dub after trying with both ImageMagick and ImageReady CS2 to create optimized animated gifs (since most of the frame doesn’t change at all) and Virtual Dub was the only one that could create a small file size without crashing.

According to the videos, the performance was not too enthusiastic. When the solenoids were attached to the hole farthest from the pivot of the actuating handle, the time from initial movement to shutter full open was 12.1 milliseconds. Thinking about it now, it actually doesn’t sound too bad. This is an old shutter; I didn’t really clean it or lubricate it, and the blades are pretty thick – in other words, maybe rebuilding it or making a new one from scratch, making everything is frail as possible would yield a reasonable delay. According to this discussion, the electromagnetic force on the plunger of the solenoid is proportional to the square of the current, so perhaps by running 120 volts through the solenoid (1.2 amps) the force would be four times as much, which, ignoring friction, would quadruple the acceleration. In general,

[tex]\frac{d^2x}{dt^2} = \frac{F}{m} \rightarrow x = \frac{F}{2m}t^2 \rightarrow t = \sqrt{\frac{2xm}{F}}[/tex],

so, interpreting m as the inertia, lag time will be inversely proportional to current in the solenoid: doubling the current will cut the lag time in half (remember, this is neglecting friction). But again, at the time I only had a 60V power supply, so I could not do tests at higher currents. I’m scared of trying to build high energy capacitor circuits, so that’s where I left it (though maybe writing this will inspire me to try it again).

At the time my goal was to build this camera much like Frans had done at fotoopa, that is, use a standad camera body and lens (in my case, a Hasselblad) and shutter it with this shutter. This size 3 shutter was going to be cutting it close in terms of clear aperture, and I figured, “All Hasselblad [V system] lenses have shutters in them, so I should try to use them.” After all, Frans was using a solenoid simply to fire a shutter with a built-in timing mechanism. The tests regarding the lens shutters were disappointing, as will be described in the next post.