Measuring focal plane shutter speeds with a turntable and distortion due to the rolling shutter effect

[Smokwawelski]

Hello everyone. I've been meaning to test my Nikon F's shutter speeds, but I've found that the shutter speed testers currently on the market are either too expensive or not accurate enough. So, I looked towards some more DIY solutions. An interesting way I've found involves photographing a record with a line drawn from its center to its edge spinning at a known rate on a turntable. The line creates a wedge shape due to motion blur from the record spinning, and by measuring the angle of the wedge and doing some simple math (eg. 7.8°: 78rpm = 1.3 rates per second = 468°, divide by the angle you found: 468°/7.8° = 60, or 1/60 of a second) you can find out how fast your shutter is. I've seen many discussions about this method here, but no mention of how it applies to focal plane shutters. Distortion caused by the rolling shutter effect is the primary concern. From what I understand, if the record is travelling with the shutter, the image will stretch, and if it is travelling against the shutter, it will compress. This will affect the angle of the wedge. Attached is two pages published in the September 1967 edition of Popular Mechanics detailing the method described above. To account for the rolling shutter, they suggest simply drawing lines from the edges of the wedge to the center of the circle and measuring the angle of the resulting wedge. Does this work? If not, is there any way to apply this method to focal plane shutters? Thanks.

 

[Dustin McAmera]

I can see two difficulties.
First is that many people, if they have a turntable at all, don't have one with 78 rpm. Lower rpm won't give a thick wedge for fast speeds. Even with 78 rpm, they're asking you to measure angles of less than a whole degree for your fast speeds.
Also, while the 'shutter closes' line is easy to draw, the 'shutter opens' line is a tangent to a curve, which has to be drawn at the end of that curve; not 'simple' at all.

 

[Smokwawelski]

I can see two difficulties.
First is that many people, if they have a turntable at all, don't have one with 78 rpm. Lower rpm won't give a thick wedge for fast speeds. Even with 78 rpm, they're asking you to measure angles of less than a whole degree for your fast speeds.
Also, while the 'shutter closes' line is easy to draw, the 'shutter opens' line is a tangent to a curve, which has to be drawn at the end of that curve; not 'simple' at all.

There is no reason why you need to use a turntable specifically. You could buy a small motor and attach a cardboard disk to it and achieve the same result. Of course the faster the motor the more accurate the measurements will be. Doing the calculations, a speed of 200 rpm seems like a good compromise for both higher and lower speeds. 1/1000 of a second would be 1.2° which, while small, is by no means impossible to measure accurately. For example, if you measure a 1.2° wedge as being 1°, you'll only get a shutter speed that's 20% faster, or 1/5th of a stop. It becomes even easier if you enlarge the image with an enlarger. Also, I dont think drawing that tangent line would be that hard. What I'm wondering is if drawing these lines and measuring the angle they create would correct for the distortion?

 

[Mark J]

If you're just going to buy a motor and attach a cardboard disk, then you somehow need to know ( accurately ) how fast that's spinning. With a record player, at least it's designed to be pretty accurate. How would you measure the rpm of your motor & disk ?

You got me thinking now, though how to use my digital camera and my turntable, set to 45rpm , to test a new LF lens shutter.

 

[Dustin McAmera]

My turntable has silver and black stripes around the edge, with a tiny lamp that flashes at mains frequency (I think), and the two form a strobe. When it's turning, you should see a static pattern of light and dark panels. If they creep forward, the turntable's too fast (and if they creep backward, too slow). So the turntable speed can be very accurate. For me though, that manual drawing of a tangent to a curve makes the whole scheme unreliable.

 

[Smokwawelski]

If you're just going to buy a motor and attach a cardboard disk, then you somehow need to know ( accurately ) how fast that's spinning. With a record player, at least it's designed to be pretty accurate. How would you measure the rpm of your motor & disk ?

You got me thinking now, though how to use my digital camera and my turntable, set to 45rpm , to test a new LF lens shutter.

That's a good point. I suppose I could record a couple videos of it spinning for, say, 15 seconds and then play each video back at 0.25x speed (so its slow enough to count) and count the number of revolutions. I would then figure out how many revolutions per minute each video was and average the numbers out. Its not an elegant solution by any means, but it works.

 

[kl122002]

I have seen this article before and did tried with my pals. It works! Unlike today, turntable was one of most common hifi gear back in 1970s and so most people could do this by themselves. In fact today I believe one can use other methods to do the same test.

The limit here is about the speed accuracy of the TT, since wheel, belt or direct drive TT's speed accuracy are different. The best is probably using quartz controlled direct drive TT, if that is still working today.

But still, as our pals agreed in the past, we develop & make our prints (B&W) so most of us not actually cares about the accuracy of the shutter too much.

 

[Chan Tran]

I have a small stepper motor which I can make it run at quite precise speed but the problem with this approach is that you have to shoot on film and process before you can tell the shutter is accurate.

 

[Jim Jones]

This is the digital age. An accurate audio oscillator flashing an LED lamp provides a target that contains precise information for determining shutter speed. Either the camera or the LED is moved to record a series of dots that can be counted, and from that count the shutter speed can be calculated. It wouldn't be too difficult to have a second LED flash at 1/10 the rate of the first LED, and perhaps a third LED flashing at 1/100 that rate to make counting easier. Movement of the dots across the film or sensor should be at a right angle to the movement of a focal plane shutter. This system should be accurate for focal plane shutters, but less so for between-the-lens shutters at highest speeds. For these, plotting the image brightness against time with an oscilloscope may be better. Better yet, but more difficult, is to measure the total light accumulated while the shutter is open, just as film does. Charging a capacitor proportionally to the amount of light passed by the shutter should achieve this.

 

[reddesert]

You can use the inexpensive "Photoplug" tester together with a smart phone and their app to test a focal plane shutter. The persnickety will point out that a focal plane shutter should be tested in two places to measure curtain speed-up/slow-down, which is in principle true, but the turntable doesn't test that either. Besides, it's a Nikon F, it's probably fine.

The app for the photoplug called "Shutter-Speed" is free and you can even use it with sound without the plug, but the sound method works better for leaf shutters and speeds slower than 1/60 sec.

I know I'm not answering the turntable question, but in terms of accuracy, the turntable method requires adding significant complexity to achieve precise measurements at high speeds.

 

[mshchem]

What's the point? It's going to be obvious by looking and listening if things are too goofy. A phone app and this "photoplug" gadget is probably adequate.

 

[Bill Burk]

Hello everyone. I've been meaning to test my Nikon F's shutter speeds, but I've found that the shutter speed testers currently on the market are either too expensive or not accurate enough. So, I looked towards some more DIY solutions. An interesting way I've found involves photographing a record with a line drawn from its center to its edge spinning at a known rate on a turntable. The line creates a wedge shape due to motion blur from the record spinning, and by measuring the angle of the wedge and doing some simple math (eg. 7.8°: 78rpm = 1.3 rates per second = 468°, divide by the angle you found: 468°/7.8° = 60, or 1/60 of a second) you can find out how fast your shutter is. I've seen many discussions about this method here, but no mention of how it applies to focal plane shutters. Distortion caused by the rolling shutter effect is the primary concern. From what I understand, if the record is travelling with the shutter, the image will stretch, and if it is travelling against the shutter, it will compress. This will affect the angle of the wedge. Attached is two pages published in the September 1967 edition of Popular Mechanics detailing the method described above. To account for the rolling shutter, they suggest simply drawing lines from the edges of the wedge to the center of the circle and measuring the angle of the resulting wedge. Does this work? If not, is there any way to apply this method to focal plane shutters? Thanks.

I’ve done that. For a Retina I. It’s only useful for a small range of times. You can scan and use Photoshop “measure” tool to get the angle.

I’d keep looking for a better tester.

 

[Bill Burk]

This is the digital age. An accurate audio oscillator flashing an LED lamp provides a target that contains precise information for determining shutter speed. Either the camera or the LED is moved to record a series of dots that can be counted, and from that count the shutter speed can be calculated. It wouldn't be too difficult to have a second LED flash at 1/10 the rate of the first LED, and perhaps a third LED flashing at 1/100 that rate to make counting easier. Movement of the dots across the film or sensor should be at a right angle to the movement of a focal plane shutter. This system should be accurate for focal plane shutters, but less so for between-the-lens shutters at highest speeds. For these, plotting the image brightness against time with an oscilloscope may be better. Better yet, but more difficult, is to measure the total light accumulated while the shutter is open, just as film does. Charging a capacitor proportionally to the amount of light passed by the shutter should achieve this.

We should be able to analyze a shutter with a phone video… opening curtain speed, closing curtain speed, integrate the time across the frame and indicate anomalies and bounce. Maybe even recommend maintenance based on the model.

 

[Smokwawelski]

I’ve done that. For a Retina I. It’s only useful for a small range of times. You can scan and use Photoshop “measure” tool to get the angle.

I’d keep looking for a better tester.

If you use a motor with a higher rpm and a cardboard disk instead of the record player you can capture higher speeds as noted above. Photoshop is a good idea too. I'm not familiar with the shutter of the Retina I, is it focal plane? I like the "mechanical" nature of this method, it's something that other methods lack.

 

[choiliefan]

Here's a printable strobe tachometer for higher speeds:

And source material:

Shop made strobe tachometer

After switching my lathe to a variable speed DC motor I had no way of knowing it’s range so rather than buying a digital tachometer for a on...

thewoodknack.blogspot.com

 

[Bill Burk]

If you use a motor with a higher rpm and a cardboard disk instead of the record player you can capture higher speeds as noted above. Photoshop is a good idea too. I'm not familiar with the shutter of the Retina I, is it focal plane? I like the "mechanical" nature of this method, it's something that other methods lack.

The old Kodak Retina is a leaf shutter.
It is satisfying. But as I say it was only good for a few speeds. I think 1/4 to about 1/50 sec the 78 RPM gave a measurable arc, but those speeds are well covered by the sound app. I had an electronic timer that was also good in that range. It’s the higher speeds that are hardest to measure.

I used three methods to check the speeds; electronic which was great for the longest times but dubious as shutter approaches 1/100 second, the record player “because I had to do it, and a sound app on the phone.

The fourth method is to use sensitometry and measure the density of test shots compared to the expected density

 

[Chan Tran]

Out of curiousity I put a CD on a stepper motor and I run the motor at 1800RPM. I shot this image at 1/1000 sec. If the shutter speed is correct the angle of the blurr should be about 10.5 degree. I can't really measure it precisely though.

 

[RalphLambrecht]

There is no reason why you need to use a turntable specifically. You could buy a small motor and attach a cardboard disk to it and achieve the same result. Of course the faster the motor the more accurate the measurements will be. Doing the calculations, a speed of 200 rpm seems like a good compromise for both higher and lower speeds. 1/1000 of a second would be 1.2° which, while small, is by no means impossible to measure accurately. For example, if you measure a 1.2° wedge as being 1°, you'll only get a shutter speed that's 20% faster, or 1/5th of a stop. It becomes even easier if you enlarge the image with an enlarger. Also, I dont think drawing that tangent line would be that hard. What I'm wondering is if drawing these lines and measuring the angle they create would correct for the distortion?

You could use a fan but how do you find the rpms?

 

[Dustin McAmera]

I found this, in the British Journal Almanac and Photographer's Daily Companion, 1889 (here:

https://archive.org/details/britishjournalph1889unse/page/498/mode/2up

you can read it on screen or download it, but it's a big download, and I have to say generally it's mostly nothing like as interesting as some later BJ Almanacs.
Anyhow, it turns out what you need as a visible pointer on something spinning quite fast is a mercury thermometer:

TIMING AN INSTANTANEOUS SHUTTER.

By D. E. Benson.

Some eight or nine years ago I purchased an instantaneous shutter, and, of course, the first thing to do was to time it.

A clockwork arrangement, with a finger turning round once in a second, not being obtainable, the best available means had to be adopted.

My bicycle, therefore, was taken out into the sun and turned over, so that the handle bar was on the ground and the large wheel free to spin.

At the rim end of one of the spokes I attached a mercurial thermometer, removed from its wooden support. The bulb being globular reflected the sun, as a fine point of light, through the lens of the camera in whatever position the thermometer was, when revolving, with the wheel.

The camera was then placed directly opposite the centre of the large wheel, on the sun side of it, and focussed, the plate inserted, and the shutter adjusted ready for exposure.

The bicycle wheel with the thermometer attached was next spun round on its axis until it had a fair velocity, From the commencement of a minute the revolutions of the wheel were counted, at the half minute the shutter was let go, and the revolutions continued to be counted till the end of the minute.

The plate was then developed, and showed a fine black line extending part way round the circle made by the rim of the wheel. This was the reflected light of the sun, and showed the arc through which the wheel had turned while the shutter was open.

Now for the time value of that arc.

Suppose that during the minute the wheel had revolved sixty times, then at the half minute, when the shutter was let off, the wheel would be going at the rate of sixty revolutions a minute, or one revolution per second, and if the arc subtended by the fine black line was 36° (the tenth of a circle), the time of exposure would be one-tenth of a second.

Suppose the wheel had performed seventy revolutions in the minute, it would at the half minute be going at the rate of seventy revolutions a minute, or one in six-sevenths of a second. The arc being the same as before, the time of exposure would then be one-tenth of six-sevenths of a second, or three thirty-fifths of a second. Arcs greater or less than the above give times of exposure proportionately greater or less.

This method gives the actual time that the light is effective on the plate, and, as almost everyone either has a bicycle or tricycle nowadays, or knows a friend who would lend one, it is easily within the reach of all.

 

[Chan Tran]

You could use a fan but how do you find the rpms?

I can find the speed of a fan easily. People would use a non contact tachometer but I don't have one. I would use some type of sensors (I have choices among several) and a frequency counter.

 

[ic-racer]

An electronic shutter tester is probably easier to build than an electronic tachometer.

 

[Bill Burk]

Out of curiousity I put a CD on a stepper motor and I run the motor at 1800RPM. I shot this image at 1/1000 sec. If the shutter speed is correct the angle of the blurr should be about 10.5 degree. I can't really measure it precisely though.

That’s where you can take the image into Photoshop and use the measure tool to find the angle

 

[Chan Tran]

That’s where you can take the image into Photoshop and use the measure tool to find the angle

Thanks I didn't know that tool in Photoshop. I used Autocad but with either tool it's kind of difficult to measure accurately. I do think it's close enough and the error is 10% or less which I think is good for the 1/1000 shutter speed. I can try to spin the disk faster and see what happens.

 

[Niglyn]

Just build this.
Cheap as chips buying the parts from Aliexpress.
Omit the tft and LCD for an even cheaper build, probably around £10.

I cannot think how anybody could make a more accurate shutter 'speed' tester at any price.
(Shutter speed never changes it is the time of release of the second curtain that governs exposure)

There is also an active builders thread here on Photrio

GitHub - billbill100/Camera-Shutter-tester-Cheap-Easy-it-Works: Arduino / ESP32 based shutter tester using three lasers. This allows for correct focal plane curtain speed adjustment & even exposure across the frame.

Arduino / ESP32 based shutter tester using three lasers. This allows for correct focal plane curtain speed adjustment & even exposure across the frame. - billbill100/Camera-Shutter-tester-Cheap...

github.com

 

[Chan Tran]

Just build this.
Cheap as chips buying the parts from Aliexpress.
Omit the tft and LCD for an even cheaper build, probably around £10.

I cannot think how anybody could make a more accurate shutter 'speed' tester at any price.
(Shutter speed never changes it is the time of release of the second curtain that governs exposure)

There is also an active builders thread here on Photrio

GitHub - billbill100/Camera-Shutter-tester-Cheap-Easy-it-Works: Arduino / ESP32 based shutter tester using three lasers. This allows for correct focal plane curtain speed adjustment & even exposure across the frame.

Arduino / ESP32 based shutter tester using three lasers. This allows for correct focal plane curtain speed adjustment & even exposure across the frame. - billbill100/Camera-Shutter-tester-Cheap...

github.com

How do you test the shutter speed of a digital camera?