Build a shutter tester for Focal Plane shutters - Cheap, Easy & it Works

[OAPOli]

Thanks. I changed the SPI speed to 80MHz and I was able to flash the code.

I set my tester for 6x6 SLRs. I spaced the sensors at 51mm vertical and horizontal (I'm guessing the standard is frame size minus 2mm on each side?). You have to put the lasers closer than that because the lens flange opening is smaller than the frame. I still have to make the sensor aperture plate.

This Kiev 88 has a slower first curtain at the high speed settings. ~1/2 stop difference, although this is asymptomatic on the negatives.

Is there a way to test leaf shutters on this setup? Maybe have a mode where only the central sensor is used?

 

[Niglyn]

Hi Oapoli,

Glad you got the flashing sorted.

I like your Lego towers )

Having the sensors in a diagonal pattern is a pain to align. If you are only doing horizontal shutters, then having all sensors horizontal makes it far easier.

Yes, the tester does leaf shutters. Just ensure sensor M and 2 are covered. Then the tester will use only senor 1. It gives a raw value for when the sensor sees and does not see the laser.

Now a slight problem, the tester is designed for 35mm cameras only. Service manuals specify a standard sensor spacing of 32mm for horizontal or 20 mm for vertical shutters.

In the setup screen, one selects either horizontal or vertical shutters. My clever code then uses an algorithm based on this setting to compute the actual shutter speed, removing the error that so many other testers suffer from, where the tester measures both slit width and sensor width. If the sensors are not set at 32 or 20mm, the speeds shown will be incorrect.

I have not looked at a medium format service manual, to see if it gives sensor width spacing. If it does, it will be a small matter to add it to the code.

 

[OAPOli]

Thanks @Niglyn , glad to hear one can use a leaf shutter on the tester.

I haven't been able to find specific on medium format shutter testers. There are vertical travel 645 cameras, horizontal and vertical 6x6s, and one horizontal 6x7. Here is what I found:

Mamiya M645	Mamiya 645 Pro	Pentax 645	Bronica S2 (6x6)	Hasselblad 202FA/201F	Hasselblad 203-205	Pentax 6x7
1/500s or 1/1000s, vertical	1/1000s, vertical	1/1000s, vertical	1/1000s, vertical	1/1000s, horizontal	1/2000s, horizontal	1/1000s, horizontal
12-12.5 ms	13.3 +/- 0.2 ms	1st: 11.95+/0.1 2nd: 12.05+/-0.1 ms	15mm/54ms (??)	9.6-9.9 ms	9.5-9.7 ms	22.9 +/- 0.3 ms

I think there is a typo in the Bronica service manual, should be 15ms over 54mm. That's the only clue I found regarding distance.

 

[ic-racer]

Nice job on the medium format FP tester!!

Every medium format camera I own has a leaf shutter so I have not had the need for one of those.

Speaking of leaf shutters, since it was mentioned above. Not sure if you [all those following this thread] have figured it out, but closing any lens down two stops makes the aperture 1/2 as big as wide open.

So...with the Nigyln tester you can measure the speed at the edge of the diaphragm stopped down by two to get a very close approximation of Te (effective shutter speed) with a single measurement. (see diagram below)

The "Film Camera Tester" makes two measurements (one at the center and one at the wide-open edge) and does the calculation, but making a single measurement, stopped down by two, gives essentially the same value.

 

[OAPOli]

@Niglyn I did some digging in all the 135 format service manuals I have. Very few specify the sensor spacing. All I found was 1) some Canon models at ~11ms/34mm; 2) some Minolta models at ~6ms/21mm; and 3) some Pentax models at ~6ms/20mm.

So I suppose there is no standard and each manufacturer used a specific tester.

Can I suggest to make additional versions of the code for 120 format? As far as I know all the 645 cameras have vertical shutters, the 6x6 have both and the 6x7 is horizontal. There is a focal plane 6x9 but it's very rare. Maybe stick to the logic of sensor spacing = frame size minus 4mm (2mm on each side). So 38x51mm for 645, 51x51mm for 6x6 and 51x65mm for 6x7.

BTW, how sensitive are your timing corrections with respect to the sensor spacing? In other words, if the actual spacing is off by 1-2mm from expected, how different would the timings be?

PS. I'm unable to test leaf shutters. When I block sensors M and 2 and test with sensor 1 I get error messages ('flicker or SSmico1')

 

[ic-racer]

Link to shutter travel times with associated measurement distances for many 35mm cameras and some medium format cameras:

Curtains Travel Times

DIY Film Camera Tester. Contribute to srozum/film_camera_tester development by creating an account on GitHub.

github.com

 

[Niglyn]

IC, thanks for the link

OAPoli, the tester was designed to measure focal-plane shutters. Additionally, it has a simple one senor reading. I use it for a Box Brownie.

Not sure why you are getting the<100 microseconds warning, This indicates that the shutter speed was recorded as less than 100 microseconds. This is most often caused by LED lighting which is interfering with the reading. I get this warning if my tester is too near the computer screen.

I will check my tester & see if it works ok, using one sensor and post back.

It is very easy to add additional sensor spacing. I will probably do it by adding a rotary encoder, to save having to bush a button multiple times.
Also easy to add better leaf shutter testing, as per IC's diagram using two sensors centre and edge of the lens.

It is very difficult to keep support for the LCD, due to the increasing complexity of user input & displayed output. I have started to replace it with a simplified TFT display, showing just the basic results but in a much larger font. The user can toggle between the original & simplified screen, so I probably will release a V4 version with added rotary encoder, depreciated LCD, variable sensor width and two-sensor leaf shutter testing.


I'm quite busy with paid work at the moment, so may be a few weeks until I can get round to it.

Thanks for the constructive feedback, it is always most welcome & helps me make changes and improve the tester.

 

[OAPOli]

Thank you very much @Niglyn, I'm very grateful. I'll get a TFT screen eventually for sure.

Awesome reference @ic-racer, thanks.

PS. I was wrong: the later Mamiya 645 models have a horizontal shutter.

 

[Niglyn]

Oops, Sorry Oapoli, you are right, the single senor mode did not work. A slight bug crept in.

Fixed now & new firmware on github.

Single senor gives raw data, not calibrated for sensor width.

Adding the leaf shutter calculation is quite straight forward, however it is a bit messy to select it, so will probably leave it until version 4, which will allow any sensor width to be selected.

Version 4 will also include a rotary encoder for ease of selection, as I am not a fan of having to push multiple button combinations to select things.

 

[OAPOli]

That's beautiful @Niglyn. Just downloaded the latest version and it's great.

I have a TFT display and rotary encoder on the way. Looking forward to the next iteration. Thanks again!

 

[Niglyn]

Version 4 Firmware - Beta version Now Available.

Hi Guys,

Version 4 of the ESP32 Shutter Tester firmware, is ready to be posted on Github. Keep an eye there.

Leaf shutter testing has been added, using the calculation as detailed by Icracer. Not having a leaf shutter, I have only been able to test it with a homemade leaf shutter. It seems to function correctly.

A rotary encoder is now required. this is the only hardware change. So to move from V3 to V4, just add the encoder.

I'm now trying out JST HX connectors, which come in different pin lengths, 5 or 8 pin for example, unlike Dupont, which are single. It makes the wiring neater, but the pins in the head-shell do need to be moved around, which is a bit fiddly. Builders can choose which to use, but the schematic wiring colours will change to the JST colours.

Full details below.


Version 4 Beta Firmware

Hardware Changes:-
Rotary Encoder added (mandatory)
LCD - Depreciated.

Wiring colour changes to match JT HX pin-headers & specified multicore cable. Makes wiring neater.
Dupont wires can also be used.

Note to convert version 3 hardware to version 4, just add the rotary encoder )

Firmware Changes:-

Baud rate changed to default value of 115200.
Leaf shutter testing added.
Sensor spacing is now user-set.
Sensor Reading validity checking updated.
Test result analysing updated (bounce, slow curtain etc).

Setup screen replaced with all new screen.
LCD updated to show & allow change of options.
Press encoder to select setting & rotate to change.

Horizontal/vertical shutter selection removed.
Replaced by user set sensor spacing.

Auto single-sensor sensing removed.
Replaced by user selecting Single Sensor, Focal-Plane or Leaf shutter.

On first use, encoder testing screen is displayed, to ensure correct functionality.
User needs to press & rotate encoder to match screen prompt.

On first use (after encoder testing) a one-time lifetime passkey is required.
Passkey is unique to each tester and will be issued free of charge upon request.
It is valid for the life of the tester.

Future updates.
Below are things that may be added.
Add simplified but larger font tft display option - to replace LCD layout.
Add PWM controlled LED light source.
Improve on current sensor reading validity and testing analysis.

 

[LIK]

Version 4 Firmware - Beta version Now Available.

Hi Guys,

Version 4 of the ESP32 Shutter Tester firmware, is ready to be posted on Github. Keep an eye there.

Leaf shutter testing has been added, using the calculation as detailed by Icracer. Not having a leaf shutter, I have only been able to test it with a homemade leaf shutter. It seems to function correctly.

A rotary encoder is now required. this is the only hardware change. So to move from V3 to V4, just add the encoder.

I'm now trying out JST HX connectors, which come in different pin lengths, 5 or 8 pin for example, unlike Dupont, which are single. It makes the wiring neater, but the pins in the head-shell do need to be moved around, which is a bit fiddly. Builders can choose which to use, but the schematic wiring colours will change to the JST colours.

Full details below.


Version 4 Beta Firmware

Hardware Changes:-
Rotary Encoder added (mandatory)
LCD - Depreciated.

Wiring colour changes to match JT HX pin-headers & specified multicore cable. Makes wiring neater.
Dupont wires can also be used.

Note to convert version 3 hardware to version 4, just add the rotary encoder )

Firmware Changes:-

Baud rate changed to default value of 115200.
Leaf shutter testing added.
Sensor spacing is now user-set.
Sensor Reading validity checking updated.
Test result analysing updated (bounce, slow curtain etc).

Setup screen replaced with all new screen.
LCD updated to show & allow change of options.
Press encoder to select setting & rotate to change.

Horizontal/vertical shutter selection removed.
Replaced by user set sensor spacing.

Auto single-sensor sensing removed.
Replaced by user selecting Single Sensor, Focal-Plane or Leaf shutter.

On first use, encoder testing screen is displayed, to ensure correct functionality.
User needs to press & rotate encoder to match screen prompt.

On first use (after encoder testing) a one-time lifetime passkey is required.
Passkey is unique to each tester and will be issued free of charge upon request.
It is valid for the life of the tester.

Future updates.
Below are things that may be added.
Add simplified but larger font tft display option - to replace LCD layout.
Add PWM controlled LED light source.
Improve on current sensor reading validity and testing analysis.

Hi, Niglyn. I made my prototype based on your device today. I only greatly reduced it in size for portability. I wanted to tweak your code a little for my tasks, but I couldn't find the code itself without compiling it on GutHub. Is he even there? If not, can you provide it to me in the 3rd and 4th versions?

 

[Niglyn]

Hi, Have not quite got round to posting V4 yet. Should be this weekend.

I'm planning to also update the build documents, showing the JST connectors, sockets for sensor heads, USB etc.
These will follow along later.

Firmware is available only as compiled binary.

If you have any ideas for updates, improvements or tweaks, let me know & I can add them.

 

[OAPOli]

Thanks @Niglyn, I've installed the TFT screen and rotary encoder. Just wondering how to install the beta v4? I tried flashing the new partition.bin but I don't think that works.

Cheers.

 

[Niglyn]

Hi,
Sorry that file should not be there. V4 has not been put onto github yet. Will be there soon.....

 

[OAPOli]

There is a discussion on Reddit (via FB) about shutter testers. Discusses a bit about using a collimated light source like this tester vs. a diffused light source like the Reveni and its predecessors. Due to the small height difference between the curtains and from their offset from the film plane, there will be a difference in effective exposure between opposite sides of the frame. The SPT journal (Sept/Oct 1977) mentions a 5% difference for the Copal square shutter at 1/1000 (slower at the first edge), which is not much. But it would be larger for bigger shutter with fatter curtains.

I wonder if this would affect the curtain speed estimate as well.

PS. I think an analogous phenomenon occurs on digital cameras when electronic first curtain and fast shutter speeds are combined.

 

[Niglyn]

Hi Oapoli,

There is a lot of guff and discussion about shutter testers, going back to the dawn of time.

Mechanical cameras will never be 100% accurate. Just a change in temperature will affect the viscosity of lubricants, fit of metal on metal parts etc. & will change the 'speed'. Trying to strive for a camera giving 100% accurate results is futile.

Next we have film, which has (other than colour transparency) a very wide latitude for over/under exposure, then add to this compensating developers and the printing process. Then the subject itself, unless under laboratory conditions, one will not get the exact 8 or 9 stops of light the film can resolve.
So trying to get a 100% accurate shutter 'speed' is pretty pointless.

The main problem I have found, as discussed & highlighted by icracer, is that shutter testers measure not only the slit width, but also the width of the sensor and combine them in the reading.

A sensor width of 1mm measuring a slit width of 2.5mm will give a reading of 3.5mm. Which is why so many testers get increasingly inaccurate at higher shutter 'speeds'. (This is without adding diffraction & diffused light into the equation and assumes full saturation of the sensor).

Old professional shutter testers used incandescent light sources because that was all that was available at the time & many think unless a tester looks like an old tester, it will not work properly. Which is far from the truth. I'm sure if a tester was made by a professional company today, it would probably use lasers & a microcontroller and look nothing like the old testers, full of 1970s CMOS logic chips.

The guy who makes the Revenue tester is a nice chap & we have exchanged thoughts. He, like myself is an electronics & firmware guy and has made a number of products. How he deals with the issue of diffraction and measuring both sensor & shutter slit width, I do not know. He has made a very nice ready-build product to take to market.

My tester is certainly as accurate as a professional tester & the Revenue. It is also far more accurate than any of the other diy ideas, using Audacity, or single sensors or indeed ones using multiple sensors, as they suffer from the inherent problem detailed above. The cost of putting an idea into production is very costly (hence the Revenue price), & I can't be faffed, so just publish my projects for free.

I had a previous product that I did sell, ripped off by Chinese manufacturing. Was funny when I got complaints from people buying the Chinese version ) as it did not work properly, but was 'cheaper than buying mine' Stupid thing is, I would have been happy to give all the details to the Chinese company for free, so the they could make them properly. Similarly if a Chinese company wants to make my Shutter Tester, making a nice case & pcb, they are most welcome to do so, just send me some )

Lasers or a diffused light source? Well there is only one answer. Lasers. All sensors will have an 'acceptance angle' of light it can see. Using a wide diffused light source will allow the full acceptance angle of the sensor to be used, thus reducing accuracy. A Laser gives a single collimated light source. Each sensor will only see it's associated Laser.

Many people want a 'film-back' as they see that as 'professional'. The problem with a film back, is that a wide diffused light source is required. It is almost impossible to line up a film-back with Lasers.

My tester gives two readings. The raw shutter 'speed' as seen by the sensors. So people can experiment by using tubes or pin-hole masks to limit the sensor width & also my patent pending computational algorithm (sounds posh, but one line of code ) that computes the actual shutter 'speed' based on the readings taken. This does of course assume that the shutter curtains and mechanical movement are in good condition.

I'm sure if you put a 'professional' tester, Revenue and mine, side-by-side, each would give a slightly different result as they will all use different hardware with different response curves and different firmware interpreting the sensors slightly differently. However the debate as to which one is 'accurate' will continue on reddit till the end of time )

 

[OAPOli]

@Niglyn I think the laser setup is accurate.

But if I understand correctly, at high shutter speeds (small slits) and with a large aperture lens, the film will be exposed to more light on one side of the frame if the slit is kept constant. This is due to the differences in the shadows of the curtains, which is caused by a small difference in their heights over the film plane. This is exacerbated by thicker curtains and larger frame sizes. So if one measures a small gradient in exposure and curtain speeds it might not be a sign of maladjustment.

This is likely mostly academic, since chasing an error under 10% is futile as you say. I just thought it was interesting; stuff gets messy at high speeds. I do wish there was more literature about this, none of those debates are new I suppose.

 

[Niglyn]

Firmware 4.0 beta is live on github.

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

 

[OAPOli]

Thanks @Niglyn! The v4 is great so far.

I have some questions regarding the screen output.

1. first line has "EEPRON READ ERROR", not sure what that means;

2. the two DEV lines are decimal EV and fractional EV errors ( i.e. log2(val/ref) )? Some values don't seem to match;

3. the curtain travel labels R-M and M-L may be ambiguous, I would suggest replacing with S1-M and M-S2 or something analogous

It seems my Kiev 88 has uneven curtain movement, but on film this seems asymptomatic. I have to investigate further.

Also, does the tester expects a sensor mask with 1mm holes?

PS. it appears that the leaf shutter mode is not functional

 

[LIK]

Thanks @Niglyn! The v4 is great so far.

I have some questions regarding the screen output.

View attachment 381370

1. first line has "EEPRON READ ERROR", not sure what that means;

2. the two DEV lines are decimal EV and fractional EV errors ( i.e. log2(val/ref) )? Some values don't seem to match;

3. the curtain travel labels R-M and M-L may be ambiguous, I would suggest replacing with S1-M and M-S2 or something analogous

It seems my Kiev 88 has uneven curtain movement, but on film this seems asymptomatic. I have to investigate further.

Also, does the tester expects a sensor mask with 1mm holes?

Hi. I have the same problem. This is a test from a serviced Nikon FE. Therefore, there can be no such uneven movement. I also checked it with film, everything is fine. I haven't tried the V4 firmware yet, so far I've uploaded it to the V3 prototype. The sensors stand diagonally

 

[LIK]

I've done some tests. From 1 to 5, I released the shutter by 1/1000, and then turned the sensors 180 degrees and took a few more measurements. As you can see, this did not affect the results in any way. Right now I'm at a dead end and I can't figure out if the problem is with the shutter or the shutter tester

 

[LIK]

Firmware 4.0 beta is live on github.

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

Maybe you know what's the matter? Do you think this is a problem in the technical part, for example, in the location of the sensors. Or is the problem in the code?

 

[Niglyn]

Thanks @Niglyn! The v4 is great so far.

I have some questions regarding the screen output.

View attachment 381370

1. first line has "EEPRON READ ERROR", not sure what that means;

2. the two DEV lines are decimal EV and fractional EV errors ( i.e. log2(val/ref) )? Some values don't seem to match;

3. the curtain travel labels R-M and M-L may be ambiguous, I would suggest replacing with S1-M and M-S2 or something analogous

It seems my Kiev 88 has uneven curtain movement, but on film this seems asymptomatic. I have to investigate further.

Also, does the tester expects a sensor mask with 1mm holes?

PS. it appears that the leaf shutter mode is not functional

Hi,
Thanks, you have found the first bug. The top of the screen shows horizontal or vertical shutter in V3. This is gone in V4, replaced with user set sensor spacing. As a valid sensor spacing could not be read, my error checking assumes an EEPROM fault.
I have changed it to show sensor space.

Dev is deviation from centre. in mS and deviation as a vulgar fraction to the nearest 1/3 stop.
New V4 user guide details the readings. Have been taking the photos for the user guide today. Will be on github when finished.

Not sure which is better, right-middle-left or sensor1-middle-sensor2
Maybe R<M : M<L or S1<M : M<S2

You will never get mechanical focal-plane shutters to give 100% accurate results. We are measuring old mechanical movement with modern electronic testers with nS timing.

Further up this thread is the iso tolerance figures. Work on 20% up to 1/125s and 30% 1/250s and above.

Your screenshot shows nearest shutter speed from all three sensors is 1/256s which is good.
Deviation show right side at 1/3 stop more exposure.
Both curtains continue to accelerate across the frame, so will give a shorter exposure at the final end, with curtain 2 travelling faster.

I would say that the shutter is working fine and within tolerance.

 

[Niglyn]

I've done some tests. From 1 to 5, I released the shutter by 1/1000, and then turned the sensors 180 degrees and took a few more measurements. As you can see, this did not affect the results in any way. Right now I'm at a dead end and I can't figure out if the problem is with the shutter or the shutter tester

Hi,
not sure what you mean by 'I released the shutter by 1/1000', do you mean you set the camera shutter to 1/1000s ?
Your readings look fine. Well within iso tolerance.

The right side will often be a slightly higher exposure, as the curtains have to accelerate from zero, thus travelling slower on the right, speeding up as they travel across the film-gate.

Putting the tester in single-sensor mode, by covering sensor 2 and M, you can take single readings at different parts of the film-gate.