Project 10, an unorthodox method to set up a focal plane shutter

Back

Project 10, setting focal plane shutter speeds.

When a camera left the factory, one of the final processes was to set up and adjust the shutter speeds. Everything was new, clean and each technician prior to the test knew exactly their job in the production chain. Test equipment was available, calibrated and optimized for each particular task.

Wind on many decades and the camera has seen years of use, has uneven wear, dirt, possibly tampered with and has been fitted with non original parts. No test equipment is available, much less calibrated.

So how do I get a passable result from all these old relics? It's unorthodox, and I tend to keep it quiet, as all those serious camera technicians out there will be appalled... but it's time to confess.

This method is used for two blind focal plane shutters that aren't tied together. Such as those employed in the Agiflex at left, the reflex Korelle, Leicas, Feds, Zorkis, Praktisix and many others. Is it accurate? Nope, not entirely, but it should give you consistent curtain speeds - providing there is nothing seriously mechanically wrong with either curtain.

Firstly, let's look at how these shutters work in broad principle.

They comprise two curtains, they are not connected and travel independently. The velocity of each curtain is governed by its own spring, coiled up inside one roller to which it is attached. The shutter relies on the springs behaving in a consistent and near identical manner to each other. This is why dirt, lack of use and stiffening curtains cause issues - this consistency collapses. The shutter curtains always accelerate at the same rate, regardless of the shutter speed used. The gap between the first and second curtains is what determines the length of time the film is exposed to light. The narrower the gap, the shorter the exposure. For the Agiflex, above, the curtain exceeds 1.4m/sec in less than a blink of an eye.

After the shutter is set and ready to fire, this is what happens.

The shutter trigger is pressed and the first curtain is released, accelerating from a standstill it opens a gap and travels across the focal plane, exposing the film in the process. Initially, the second curtain is held back by the shutter speed selection mechanism but in turn is released, allowing it to accelerate with identical characteristics to the first curtain, trailing behind it, the two curtains expose the film though the gap between them. The first curtain is brought to a stop by the mechanism and the second curtain catches up finishing the exposure and closing the gap completely in a condition known as "capped". Given all the mechanics, rollers, cloth and tapes, there's a fair bit of mass to accelerate and consequently a fair bit of energy to dissipate as the shutter stops. This is why focal plane shutters tend to be rather noisy, always regarded as one of their disadvantages. But this is what I exploit to adjust these shutters.

Each of the events described above has a particular noise signature, can be recorded and analysed.

Some means to record the camera and output an MP3 file will be needed. This could be a downloaded application for a mobile phone or a dedicated recording device. I use a cheap little Ruizu X20 MP3 player bought on ebay, but there are dozens of options available.

Next up, you will need some sound software to open the MP3 file and view it graphically. I use an elderly Apple MacIntosh running SoundBooth, but I am sure there are many applications that do much the same for PCs.

Cock the shutter on the subject camera and set the shutter speed to B, no need to record this.

Choose a quiet location. No ticking clocks, no refrigerators humming, no TV in the background, no computer fans whirring away - as silent as you can get it. Set the recorder going and release the shutter, let the first curtain go, wait a second, then let the second curtain go. Stop the recording. Now download and examine the result.

Here is a recording of a Reflex Korelle being fired as above. This Korelle has had new curtains fitted and is running really quietly and smoothly, but all the events described can be seen clearly, it takes a bit of practice though. Here you can clearly see the loud report as each curtain is brought to a rapid stop, producing the two spikes. Let's zoom in and look at the first curtain more closely.

The first thing to do is to get the two curtains travelling with near identical characteristics. Looking at the sound recording. let's break it down into it's elements. Zoomed in on the first curtain. Loudness is on the Y axis (vertical), and time on the X axis (horizontal).

A. Finger pressing down on trigger, mirror moving upwards.

B. Mirror seating in light dams, mechanics grating slightly.

C. Click as first curtain escapes, followed by that sound decaying.

D. First curtain accelerates away, the increasing sound tells you its accelerating, otherwise the sound would reach a constant. The dotted drawn ramp shows you the trend. Note the curtain continues to accelerate for its entire travel. It's hard to spot, but the variations in sound during its travel have three identifiable peaks, equating to the three rotations the shutter roller makes as it unwinds. The noise is a combination of the spring roller pulling the curtain, a narrow guide roller spinning furiously, the curtain supply spool spinning and the plain cut gear teeth meshing.

E. The curtain's stop pin hits the stop, it's fair to assume the event has occurred as the sound registers. It's a fairly violent affair, and hence the largest sound spike.

F. The sound resonates and decays through the shell of the body producing a slight ringing, but this is of academic interest as the curtain finished its business at E.

The part that interests us principally is the duration of D.

Most cameras have a shutter speed that involves the first curtain arriving at the stop just as the second curtain is released. This will normally be the slowest shutter speed not employing a clockwork timed delay, if fitted. This will also normally be the flash synch speed (if fitted), as for a brief moment, the entire focal plane will be open, allowing the flash to expose the full image. This becomes our target speed. After some experimentation it seems that the duration of D needs to be around the same as the slowest shutter speed.

For this camera, that speed is 1/25th second. Or 0.04sec. My MP3/Soundbooth combination goes to a resolution of 1/1000th sec, so really only suitable for these earlier generation shutters.

Your sound software will come with a timestamp readout. The readout at E minus the readout at C will give you the duration. If its quicker, say 0.03, back half a turn of the first curtain spring or vice versa if too long.

The benefit with this approach is that it allows you to actually see what is going on. For this particular camera I thought the second curtain was a bit sluggish, but it turned out the second was okay, it was the first running too fast. It was running at 0.03.

The same applies for the second curtain, although it's sound signature will be different.

As the shutter was recorded on B setting, the first sound picked up is the finger relaxing on the trigger shown up in A2. The click as the curtain escapes is less pronounced, but you can see the sound increasing as the second curtain accelerates continuously, I have drawn on a dotted trend line, again this equates to the three rotations the roller makes as the curtain rolls off it. The duration on this curtain turned out to be 0.042sec, and there isn't sufficient finesse of adjustment to get it any closer.

Any adjustment follows the same theory, too fast slacken the spring half a turn or vice versa.

Then re-record the shutter on B setting and repeat until you have both curtains with duration as close to D=0.04 as possible.

In the Korelle example there are four rotating assemblies as the first curtain fires and five as the second curtain fires, all these have their own sound signatures superimposed on one another. No two models are the same, and no two cameras of the same model have the same overall signature, so don't worry if your's looks different. However, this method will betray a curtain that doesn't continuously accelerate or one with a tight spot. If you can't get the curtains to match reasonably closely, you have a mechanical issue to resolve.

Once you have the first and second curtains as close as you can you can check the result by recording the shutter at 1/25th sec. This recording is of the Korelle at 1/25th.

The accelerating second curtain is obscured by the collapsing sound of the first curtain stopping, but we are only interested in the time SS, or shutter speed. The time between the first curtain stopping and the second curtain stopping. In this case and agreeable 0.041sec. Since we have checked and adjusted both curtains independently, we know they are closely matched, and the duration at 1/25 is pretty close. If it's running a tad slow, you may be able to put half a turn on both curtains, but you will need to recheck them individually as each curtain has different static friction, drag and running characteristics. Particularly as much of the camera may be 70+ years old.

At this point you may be tempted to record a faster shutter speed, here lies the road to premature madness. You may just about get 1/50th, but after that things can get a bit confusing. Pop a film in it, pick a nice sunny day and use the good old sunny sixteen rule whilst photographing something suitably bland and mid grey.

I made this confession in the hope it may help others who wanted to find a way to get their old focal plane shutters working reliably and there's not much on the internet on the subject. It's hardly a professional approach, but it has a number of advantages over the spinning discs, as it allows you to examine each curtain in forensic detail. Also spinning discs need to be calibrated... a problem in its own right.

I hope this proves helpful to someone.

Next restoration Project page.