Focal Plane Shutter Speed Tester Calibration

[ic-racer]

Without going into detail about how to make a shutter speed tester (there are many ways to make them) this thread is about calibrating or validating the values obtained when an audio analog-digital interface is used to record the shutter speed.

Audio analog-digital converters commonly sample at 44100/sec, which is the sample rate for audio compact disks.

In that case, 1/1000 of a second would be represented by 44 samples.

However, as anyone that has used this method would know, a correctly functioning shutter at 1/1000 produces a waveform with many more samples than 44.

So, there are a number of errors in the system for which need to be accounted.

 

[ic-racer]

There are numerous methods to validate the readings, I'll only describe a few methods here, and lastly present a simple way that I'll be using as my gold standard.

The first method I used was to calculate the error I would obtain based on the width of my light sensor.

The focal plane shutter is designed to expose each molecule for the designated time. Since no sensor is as small as the molecules in the silver halides, the sensor will give an error because it is not a point source.

It works out that the width of the sensor adds time to the result such that the focal plane shutter slit is effectively enlarged by the width of the sensor.

For example a 1mm sensor makes a typical 1/1000 second shutter slit (2.25mm) behave as if it is 3.25mm.

With a typical cloth focal plane shutter that travels across the film plane horizontally 36mm in 1/60 of a second, the extra 1mm of slit width works out to 0.4 milliseconds extra exposure read by the tester.

So, a 1/1000 exposure would be represented by 44 samples at 44100 samples per second plus the error of .4 milliseconds which is about 62 samples total.

This is pretty close, but still an estimate, because the actual sensor size is hard to measure in the TO-18 package, so 1mm is really a close approximation and the manufacturer does not give a dimension of the actual sensor.

BTW the sensor I'm using has a clear plain glass opening, without a convex lens and I'm using a laser beam, so the distance between the shutter and the sensor does not matter as the rays are not diverging from the light source.

 

[ic-racer]

What I describe next is ten cent device that can be used to simulate the progression of the focal plane shutter slit across the laser beam/sensor combination.

To create this device we need to know these items:

Film Gate Horizontal Distance
X-synch Speed
Slit Width at the tested shutter speed

In the case of the cameras I need to test, the Film Gate is measured to be 36mm.
The X-synch Speed is 1/60th. Therefore the shutter curtains need to cross the film gate at 36mm per 1/60th of a second or 2160mm/sec to achieve a correct exposure to the film.

Each speed after 1/60 sees the shutter curtains halving the film gate size. So 1/125 sec = slit width of 36mm/2 = 18mm, and so on so we find 1/1000 of a second would need a slit width of 2.25mm.

 

[ic-racer]

The device consists of a rotating disk with slits which mimic the slits produced by the two curtains of a focal plane shutter.
The disk needs to rotate at a correct RPM so the angular velocity of the slit is 2160mm/sec. This is not that difficult to calculate.

My motor runs at 0.1652 seconds per revolution. This is easily determined by spinning the motor with some sort of flag in between the shutter tester beam. The resulting waveforms total duration are measured and divided by the number of waveforms present in the measured group.

Knowing the motor takes 0.1656 seconds per revolution, we can calculate the distance our shutter needs to travel in 0.1656 seconds.

In one second the shutter travels 2160mm, so in 0.1656 seconds the shutter travels 356mm.

Knowing the relationship between a circles circumference and its diameter (2 * Pi * r) allow us to construct a circle of 356mm by drawing a radius of 56.65mm.

 

[ic-racer]

The circle is printed 100% and its dimensions are verified.
The paper circle was placed on mat board and a punch used to identify where openings should be placed in the mat board.
After cutting the appropriate over-sized openings in the mat board, an adhesive was sprayed on the two items to affix them together.

 

[ic-racer]

The mat board was cut in a circle and precision gaps in the disk were created with fine-edged slide masking tape.

 

[ic-racer]

The device was then attached to the precision drive motor (Makita 6095D) and spun in between the laser and sensor of the shutter speed tester and the results recorded and analyzed with free Audacity software.
The waveforms were recorded and saved to be used to compare those produced by the camera shutters undergoing speed testing.

 

[ic-racer]

Now that I have calibrated my shutter speed tester, I can make error tables for my shutters like this one.

 

[ic-racer]

This is a camera that needs to be cleaned.

 

[gone]

I used to use Audacity in combination with electronic shutter speed testers I bought on fleabay and elsewhere. But my negs were seldom ideally exposed no matter what camera I tested or which light meter I used. Finally just went to Nikon film cameras w/ electronic shutters and good meters, including spot. Then the entire roll was usually exposed on the money. Not sure what any of this means, all I know is it worked.

 

[monopix]

Not sure it makes any difference as I got a bit lost in your calculations but you seem to be making assumptions about the curtain speed, and hence the slit width, which may not be accurate. Using the flash sync speed and the distance of travel to make an estimate of the curtain speed should get you close but it probably won't be accurate. The shutter will be set to ensure the full width of the frame is uncovered at the sync speed but there will be some tolerance on this and that means the curtain speed might be slower than you think.

Something I did with a similar setup was to cover the sensor with foil with a pinhole in it. This reduces the sensor width considerably and reduces the errors. Using a laser still provides enough light through the pinhole for the sensor to read.

 

[Nodda Duma]

Good posts and fairly spot-on.

At the faster shutter speeds, there are two sources of errors that will or may begin to contribute significantly.

You are already aware of the detector size being an unknown. PIN diodes come in all sizes, unfortunately, so the only way you could know is if the TO can has a part # or by asking the manufacturer. In this type of application, 200 um square diodes are commonly used, but that’s no better of a guess than 1mm. A very small error is that the sensitivity falls off near the edge of the material, but that’s not an issue here.

The geometry of the source vs. shutter will be the other source of error, as the light incident on a film grain from any point in object space is coming in distributed across a cone of light. The incident angle is of course related to the f/#, and the different geometry of a wide open lens relative to your test setup will contribute an error. This is separate from vignetting. Easiest solution to eliminate this potential error is of course to test through a lens. Theoretically, this error will be larger if your shutter curtain speeds vary through their range of travel.

Good stuff!

-Jason

 

[ic-racer]

Not sure it makes any difference as I got a bit lost in your calculations but you seem to be making assumptions about the curtain speed, and hence the slit width, which may not be accurate.

The slit width and curtain transit time values are specified in the service literature for the cameras. However, most cameras (in which the speed and slit width are not specified) will be similar due to similar construction.

The information on focal plane shutter adjustment and testing is kind of scattered all over the "Learn Camera Repair" site. For example a lot of good information is hidden in the Pentax repair manuals. I'll try to make a list of the good resources from that site.

The Pentax service manuals are especially useful for the Rollei SL35cameras (rather than the well documented Leica and early Nikon shutters) because the Rollei SL35 is somewhat a copy of the Pentax; inside and out.

 

[monopix]

The slit width and curtain transit time values are specified in the service literature for the cameras. However, most cameras (in which the speed and slit width are not specified) will be similar due to similar construction.

True. But you said...

In the case of the cameras I need to test, the Film Gate is measured to be 36mm.
The X-synch Speed is 1/60th. Therefore the shutter curtains need to cross the film gate at 36mm per 1/60th of a second or 2160mm/sec to achieve a correct exposure to the film.

Which is what I was referring to.

Also, just because a curtain speed is quoted in the manual, you shouldn't assume that's what the curtain speed actually is. On old cameras it's unlikely to be correct.

 

[ic-racer]

The shutter transport speed is set to that value with the opening shutter spring.

I have a thread on building a shutter tester, this thread on testing the shutter tester, and repairing Rollei SL35, but no thread dedicated just to FP shutter adjustments.

 

[ic-racer]

monopix, getting back to your questions. Maybe it has to do with Nominal vs Actual shutter speed. As you know the progression of the numbers on the speed dial are not mathematically 'correct.' These differences are less than the 20% tolerances so I'm ignoring them.

Nominal 1/60 = Actual 1/60 (this is set per the instructions from the camera service guide)
Nominal 1/125 = Actual 1/120
Nominal 1/250 = Actual 1/240
Nominal 1/500 = Actual 1/480
Nominal 1/1000 = Actual 1/960 (this one is adjustable but changes all the others up to 1/60 based on the shape of the spinning cam).

 

[RalphLambrecht]

I'm not very talented with soldering, but interested to buy one of these made up can they be bought?

 

[ic-racer]

I'm not very talented with soldering, but interested to buy one of these made up can they be bought?

This post shows the schematic and how I wired it. In fact there are three different schematics that all do the same thing. I assembled mine with parts I had laying around. I used a buss bar. Currently mine is disassembled. Next time I build it I might make it more permanent.

Shutter Speed Testing Focal Plane Shutters

I just finished evaluating some focal plane shutters and wanted to share the information as I found very little on this topic from the internet. Before building the tester, one needs to know which information is needed. For a focal plane shutter, the fast speeds need special consideration...

www.photrio.com

 

[Andreas Thaler]

Excellent topic!

 

[ic-racer]

This post shows the schematic and how I wired it. In fact there are three different schematics that all do the same thing. I assembled mine with parts I had laying around. I used a buss bar. Currently mine is disassembled. Next time I build it I might make it more permanent.

Shutter Speed Testing Focal Plane Shutters

I just finished evaluating some focal plane shutters and wanted to share the information as I found very little on this topic from the internet. Before building the tester, one needs to know which information is needed. For a focal plane shutter, the fast speeds need special consideration...

www.photrio.com

Buss bar, I mean terminal block...

 

[Niglyn]

Hi,
I was directed here by ic-racer, who has spent much time & thought on this subject & posted some most useful information and experiment results. Thanks! )

It seems to me (thinking aloud)

There there will always be an error with any shutter tester, due to the width of the sensor used.

The error is large enough to be a problem when measuring exposure at shorter times.

As the curtain slit width gets smaller (exposure decreasing) the error between measured and actual exposure time will increase due to the slot width & sensor size ratio reducing

The error can be calculated with a known slit width (as demonstrated by ic-racer), however, slit width at any exposure will vary depending on the camera's sync speed (1/30s sync would be 36mm, 1/125s sync would also be 36mm)

It would be easy to add a sync speed selector to a modern microprocessor based shutter tester, which would add an offset for a known slit width.

Setting up a camera, one has no idea what the initial slot width will be as it would depend on spring tension and how gummed up the works are.

Using one sensor, to measure 'seen' and 'not seen' causes the problem we face as the sensor has width, thus the two measured values have a slight difference between them.

Using two sensors in a shutter tester, one measuring 'seen' and the other measuring 'not seen' would remove this error, however, precise manufacturing would be required, or fine mechanical calibration adjustment, outside the scope for most of us.

Using two sensors, speed of a passing object can be measured without sensor width error, as each sensors 'broken' time would be used.

As a shutter tester uses two sensors (or three) the travel time, thus average speed, of each passing shutter curtain can be measured without sensor width error.

A simple calculation can be used to calculate the curtain transit time across the sensor, based on curtain travel speed.

The time of sensor transit being known, this can be used as a calibration value to correct the error caused by sensor width.

An assumption has to be made that the shutter curtains move freely and do not hang or lag. However any such fault will show as an imbalance of exposure on the tester. Similarly an assumption has to be made as to the width of the sensor and how much of this width has to be exposed or covered to trigger the sensor.
------------------------
Following my thoughts & musings above, I have added some test code to my version of a shutter tester and it seems to be correcting the measured exposure value, with the corrected value getting progressively smaller as the exposure time decreases.
More tinkering is required, but my hopes are high for a more accurate shutter speed tester.