Universal Sensor Case and Other Modifications

[User-462]

As I worked on the sensor assembly, I realized that the sensor construction is fairly critical for good testing results. One way to ensure reliable construction is to reduce the amount of hand crafting. I recreated the sensor .STL files as Solidworks files, and then made several modifications to the designs. I also designed some new assembly tools to reduce variations in the final build. I hope others will find this useful.

FYI, @srosen already demonstrated the beginnings of some of these ideas in the original plans on Github.

Here are the highlights:
• The face plate and pcb form an integrated unit that is easy to inspect.
• New assembly tools position the diodes correctly for soldering.
• No need to file away part of each sensor diode.
• The sensor apertures are built-in as pilot holes that are precisely located.
• A universal sensor case and rear lid fits the new Sensor 1, 2, and 6 face plates.
• Tie-down posts prevent each sensor cable from twisting or pulling.

In addition, I found some 3D printer slicer settings and techniques to improve the accuracy and function of the printed models.

[User-462]

Sensor Cavities and pilot holes
Each sensor diode must align with its 0.8 mm view port hole and then remain in place. The original design called for removing the alignment tab from the front of the sensor to create a flat surface for mounting against the face plate, and only a 2 mm recess positions the diode lens. Filing away the tab creates a risk of scratching the sensor lens, especially given the tiny size of the sensor. It also takes additional time.

New Features:
• The new modification provides a cavity in the face plate that fits the sensor alignment tab.
• The cavity also contains a V-shaped countersink for the lens. The countersink makes it easy to use a 1.5 mm round bit to deepen and clean the recess for the lens (which is also 1.5 mm in diameter).
• A slightly undersized pilot hole continues through to the front of the face plate. The hole serves as a solid guide for drilling and cleaning the final .8 mm hole.
• The added benefit of the pilot hole (in addition to accurately spacing the sensor holes) is that a solid tube is formed around the pilot hole during printing. Without the pilot hole only a front and back surface prints, leaving an unpredictable honeycomb support in between that could interfere with drilling a clean hole.

[User-462]

Sensor Assembly Tools
These tools address two concerns about aligning the diode and the pcb during soldering along two dimensions:
• Lateral centering of the sensor over the guide hole in the pcb.
• Vertical distance of the sensor from the pcb

Both dimensions must be accurate to increase the ease and reliability of assembly.

[User-462]

Lead Bending Tool
The lead bending tool sets the bending point on the legs to establish the correct lateral distance from the soldering holes to the sensor lens. It includes a cavity for the rear of the sensor, and grooves for the three legs.
You place the sensor on the tool and then bend the legs into the grooves. I use a piece of plastic to form a tight curve on the leads as I bend them.

There are two variations of the bending tool based on differences between sensor pcb designs:
• One tool for Sensors 1 and 2, which positions the sensor lens 6 mm laterally from the soldering holes.
• Another for Sensor 6, which positions the sensor lens 7 mm laterally from the soldering holes.

[User-462]

Assembly Positioning Tool
The Positioning Tool holds the diode correctly over the board while you solder the diode into position.

The Positioning Tool includes three parts plus a clamp. The two plastic parts hold the front and back of the diode. A pin locates the two parts precisely over the alignment hole. A clamp holds everything in place while you solder.

Assembly order:
1. Place the pin through the alignment hole in the pcb and place the bottom half of the Alignment tool on the pin.

2. Place the diode with pre-bent leads through the solder holes and down into the cavity of the tool.

3. Place the top half of the Alignment Tool on the face of the diode and into the matching recess on the bottom part.

4. Clamp the sandwiched pcb, tools, and diode between a clamp.
5. Solder the diode to the pcb.

[User-462]

Note: The pin needs to be a 0.8mm wire or rod to fit through the pcb alignment hole, and the hole in the bottom half of the alignment tool. This rod can be purchased from Amazon. While I waited for it to arrive, I discovered that a small paper clip was just the right size. I pre-bent the wire in order to clamp it on the bottom.

[srozum]

Looks like you preparing for mass production :-)
I just used piece of cardboard

[User-462]

Keeping in mind "simple and repeatable!"
:-)

[User-462]

Actually, of the 10 or so diodes I've soldered in, none of them required adjustment when I fitted the pcb to the face plate. The diodes just slipped into the cavities on the faceplates - even sensor 1 with three diodes. The fit is good enough and the leads are stiff enough that I did not add foam supports after the first one, although the foam would be a good idea if you need to use the sensor as a hammer!

[User-462]

Unit Sensor Face Assembly

In the original design, the pcb is mounted in the sensor body, and then the face plate is aligned and glued to the body.
In this modified design, the pcb mounting surfaces are on the face plate. The gives us a couple of advantages.
1. You can clearly see that the diodes are properly seated against the face plate when you fasten the pcb and face plate together.
2. The face plate fastens to the body with screws for easy disassembly plus secure mounting.
3. An extra extrusion forms a light seal and adds rigidity.

[srozum]

Hey, where is mounting points for Medium Format adapters?

[User-462]

Still on the drawing board...

[srozum]

Good. If you going to repeat my clip-on design, then follow Sensor 6 mask design. It has notches oriented in opposite directions, so you clip on by pushing on one side and remove it by pushing the opposite side.

[User-462]

Sensor Cable Location and Tie-Down

The original design places the sensor cables in the center of one wall of the body and lid. This requires additional routing requirements for the cable wires because of the mounting bushing in the back lid on sensor one. Only one of the sensor bodies included a tie-down post that allows you to zip tie the cable as a strain relief and prevent stress on the wires soldered to the pcb.

The new design addresses securing the cable and eases the wire routing inside the case.
• The design shifts the cable hole slightly to ease the wire routing around the lid feature for the 1/4-20 mount in the back of the lid.
• A rounded fillet inside the lid on the mount eases the space for wiring as well.
• The back lid includes a modification that forms a half-round hole for extra grip on the cable.
• An updated post next to the cable hole allows you to zip-tie the cable.

The back lid retains the original snap-in feature for easy removal.

[User-462]

CAD Workflow
To modify the existing CAD designs I used the .STL files from Github as a reference, and rebuilt them from the ground up in Solidworks. With the everything in Solidworks it is fairly easy to make modifications and check the fit of all the parts in the full assembly.

The full sequence goes like this:
1. Create new files in Solidworks using the original .STL files as a reference.
2. Create or download all related parts like diodes, screws, pcbs, etc. to check the fit in the final assembly.
3. Save parts as .STL using maximum quality and face count settings.
4. Use Cura 5.5.0 or higher to slice the .STL files and save as .gcode files. Use the Super Quality profile, and modify certain settings to assure clean prints.
5. Use the .gcode files in the 3D printer.

[User-462]

3D Printer Slicer Settings

During print testing for the new CAD designs, I had difficulty getting clean prints. I use a Creality Ender 3 v2 with the stock nozzle, which is a fairly common printer. The PLA filament I use is HatchBox Matte Black. I typically warm up the printer for 20-30 minutes before starting a print, and then let the print cool on the bed to room temperature before removing it.

Leveling
I level the bed using the following method and the settings are fairly precise:
1. Autohome the printer.
2. Set the nozzle Z-height up by 0.1 mm after auto-homing.
3. Use a 0.1 mm feeler gauge to level over each bed adjustment screw as usual.
4. Autohome again

I found the method here: https://youtu.be/_Ic00W18_ck?si=5NNhl95AmBayRcxR

.STL files
When I save the Solidworks file as .STL, I set all the quality settings to the maximum for face count, etc. I figured that if we are printing and object with holes that are a fraction of a millimeter, and other cavities that must fit the diodes properly, the more faces in the file the better. Tiny round holes would be round and not some polygon that ruins the intent.

[User-462]

Ender printing bed
One happy discovery was I found that the glass bed forms a flat and slightly textured surface on the print that touches it. It appears that the molten PLA molds against the bed and leaves no trace of the extrusion to the touch. However, you can see the path of the extrusions in the plastic because of slight color or shading changes. Perhaps this is because the matte black PLA contains a dulling dulling additive that is responsible for the shading change. In any case the glass texture on the print looks quite nice – a bit like carbon fiber.

[srozum]

I see you still had to drill holes to clean them. Then why bother?

[User-462]

Printing small pilot holes plus a countersink has some advantages:

1. The locations are precise and centered on the faceplate - no template required
2. Printing a hole forms a reinforced tube around the hole that makes a solid drilling surface with no hidden surprises because of internal support structures. (assuming an FDM printer)
3. The pilot hole keeps the fine drill going straight, so the exit hole is lined up with the entrance hole.
4. The printed countersink keeps the larger drill (1.5mm to match the diode lens) well centered.

[srozum]

Printing a hole forms a reinforced tube around the hole that makes a solid drilling surface with no hidden surprises because of internal support structures. (assuming an FDM printer)

I was building a tool which is meant to be used for years. So I printed all sensor parts 100% solid. You can keep your holes as you like them.

[User-462]

@srozum Did you see any warping with the solid print? It's not that thick, but I was using the default slicing with internal fill pattern to avoid a solid print.
In any case, having a pilot hole really does ensure success by keeping the hole perpendicular to the front. I hand drilled these holes easily. and could feel the pilot hole guiding the drill bit.

[srozum]

Did you see any warping with the solid print?

It all depends on filament and printing temperature. With sensor parts I had no warping issues. Front panels for the controls were PIA to print.

[User-462]

Cura Slicer Settings
Even with the glass bed there were issues about the print lifting at the corners and occasionally failing to stick to the bed. I solved this by setting the bed temperature to 60 C, and the nozzle temperature to 210 C.
The next issue was the surfaces on the bottom of parts that were above the bed and therefore held up by printed support structures. Those surfaces looked rough with loose extrusions and rounded corners that should be square. The solution for this was to change the slicer settings (Cura 5.5.0) governing the spaces between bottom support and the part to 0.1 mm.

Here are the Cura 5.5.0 custom settings added to the Super Quality profile (as they appear in an export of the custom settings). Your printer and slicer may require different changes, but this may inform your troubleshooting:

adhesion_type = brim
material_bed_temperature = 60
support_enable = True
support_structure = normal
support_type = everywhere
ironing_enabled = True
material_print_temperature = 210
skin_monotonic = True
small_skin_on_surface = True
support_offset = 1.2
support_z_distance = 0.1

[srozum]

Hi @User-462

Nice work, looks good.

But you missed key principle of a design: it should be simple and repeatable, thus parts can be printed using basic FDM printer with standard 0.4mm nozzle and no to a minimum supports.

Here are some responses to you highlights:

No need to file away part of each sensor diode.

To accommodate for diode key your printer should be finely tuned. Thus it's simpler to file the key and print flat front plate

New assembly tools position the diodes correctly for soldering.

PCB has drilled holes in it which you can use as a template. Press Diode lens against PCB and bend legs upwards where they align with soldering holes

The sensor apertures are built-in as pilot holes that are precisely located.

Again, requires precise printer and some slicing experience. Printing blank front plate and then drilling opening, using PCB as a stencil, produces cleaner results. Also, not all sensors have same opening diameter, so better to print same front plate and use different drill.

A universal sensor case and rear lid fits the new Sensor 1, 2, and 6 face plates.

Sensor 2 has smaller face plate for a reason. Not all cameras allow you to fit big sensor. Old cameras have sprockets or other pins close to film rails.

Tie-down posts prevent each sensor cable from twisting or pulling.

Tie-down "eye" placed on a lid because it doesn't require supports. Its 4mm gap allows for small tie and can be printed without support.

The Positioning Tool holds the diode correctly over the board while you solder the diode into position.

Yeah, that could be useful. In my design Diode's soldering holes placed a bit far away to allow for longer leads and some bending, so soldering height is not critical.

FYI. Sensor 6 design was finalized when Sensor 1 was already on sale. Otherwise I would have change Sensor 1 PCB to fit into same enclosure as Sensor 6.

I appreciate your work, and hope someone will use it.

[User-462]

Thanks, @srozum , I was also cognizant of the need for a standard printer and setup. The printer I used is an Ender 3 v2 with the standard 0.4mm nozzle. I understand that printer is a derivative of the Prusa i3 printers. The settings I mentioned are fairly minor tweaks. So, it does take some care in printing, but is probably attainable by most.

By the way - should l simply attach the .STL files here? I'm new to Github and its conventions.

[srozum]

You have few options:

1. attach them here
2. fork repository, commit STLs, link here or create PR.

Since you are new to Git, then option 1 is OK

[Surreal62]

Very useful additions and I am interested in giving them a try, so I will have some SLA printed at JLCPCB early in the new year. Really high quality and very competitive pricing.

[User-462]

It will be interesting to see how an SLA print compares to an FDM printer. Please show us the results and describe the drilling process and results for the printed holes and the fit of the diode cavities, etc. The STL files (still to upload) are tweaked for a common FDM printer (Prusa family - Ender 3 v2). The tendency on an FDM printer is for the plastic to spread a fraction of a millimeter, So I accounted for that. An SLA print may need a slightly different .STL file. Otherwise the pilot holes may be larger.

…

On Sat, Dec 30, 2023 at 5:13 PM Surreal62 ***@***.***> wrote: Very useful additions and I am interested in giving them a try, so I will have some SLA printed at JLCPCB early in the new year. Really high quality and very competitive pricing. — Reply to this email directly, view it on GitHub <#47 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BCJMKEWMXMNV2JCYNRO52N3YMCNYFAVCNFSM6AAAAABBHYTEECVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TSOBQHEZDE> . You are receiving this because you were mentioned.Message ID: ***@***.*** com>

[Surreal62]

I will have five sets of assemblies printed as I have 5 sets of PCB's with me already so can experiment if need be. 👍

[Surreal62]

Just a note to let you know that the first delivery of your designs printed in black resin.
Dimensionally absolutely superb.
The photosensors fit exactly with no discernable free-play.
The holes in the faceplates printed absolutely clean and are as close to 0.8mm as I can tell.
A 0.9mm drill bit will not enter the holes and a 0.8mm bit is a quite smooth sliding fit.

So to sum up, there are no issues with the parts being printed by SLA process.

Super parts that will lead to greater accuracy.

[User-462]

That is fantastic to hear. Sounds a lot less problematic than FDM printing. Interesting about the holes, too. I believe the CAD ended up at 1.0 mm for the holes because of the spread I saw on FDM printing.
Did you also print the bending and assembly tools?
Please keep us posted on the assembly and testing.

[Surreal62]

Assembly of the printed parts was a breeze - the fit is perfect. The electronic mponents themselves were the issue.
I bought in a batch of 10 photocells and only after soldering in the three for sensor 1, discovered they are mixed types! Three out of the ten are measuring dark time, and one went into my assembly causing headaches as of course, the firmware didn't know what to do! 😂😂😖
I should have 'scoped them all first - and suggest anyone using these things does so before use if they can.

[User-462]

STL files here. I am including a screenshot from Cura to show which side goes up when printing. Please review the slicer setting suggestions in this discussion for a starting point.

[User-462]

Sensor 6 face universal

snsr 6 Face-universal.zip

[User-462]

Sensors 1 and 2 bending arbor

Sensor 1 and 2 bending arbor.zip

[User-462]

Sensor Alignment tool


alignment tool.zip

[User-462]

Sensor 6 bending arbor

Sensor 6 bending arbor.zip

[User-462]

Sensor 2 face universal

snsr 2 Face-universal.zip

[1N4003]

Looks awesome. I’ll give the stls a try 👍

[ic-racer]

My sensor, assembled with much care, still gives slightly different readings depending on the orientation. Otherwise very good, but maybe could be better if I were using the sensor housing presented here.

I'd be having JLCPCB print it in black resin.

My only potential stumbling block is not knowing the logic of the new sensors I'll need to buy. Currently all my sensors run on Firmware 2.9.0.

Where are the files located? Ok, I see, they the .zip files attached to the posts.

I'll probably countersink the hole, as I did in my other sensor, as I don't have a drill press, and the holes might not go straight.

[User-462]

@Surreal62

Having a deep countersink would be a step towards inaccuracy.
The 0.8mm hole should end up as close to the surface of the photodiode as possible.
The countersink is best done with the bullnose bit as you say and to the depth only of the photodiode lens.
Keeping the hole, countersink and photodiode concentric is the key.

What would it take to test this? Watch the diode response on an oscilloscope? - then look for a not-so-square wave?
(I've been wondering how to build a sensor tester, especially sensor 1 - Perhaps use a summing op amp and observe the staircase as the diodes respond.)

About the distance from the 0.8mm hole to the sensor - it is a sensor lens we are talking about, and it is a hemisphere that gathers any light that hits is and focuses it down to a small point where the microchip lives. Wouldn't it be an advantage for the entire lens to capture more light? Is it a concern about the time it takes for a shutter curtain to traverse the 0.8mm aperture? Trying to understand why it would be less accurate and not more.

[srozum]

Trying to understand why it would be less accurate and not more.

I see you weren't good at trigonometry class.

[User-462]

@ic-racer

My only potential stumbling block is not knowing the logic of the new sensors I'll need to buy. Currently all my sensors run on Firmware 2.9.0.

I had the same question, so I tested the laser sensors first and then selected the software version that supports it. To test the diodes, I put them on a breadboard, carefully observing the pins and polarity of the 5v connections and reading the output pin with a voltmeter. It would swing between 0 and 5v when I waved my hand in front of it.
It's a good idea to test them first anyway to screen out bad diodes.

[User-462]

@srozum I phrased the question to clarify the cause of the inaccuracy mentioned by @Surreal62 - Is it a concern about the time it takes for a shutter curtain to traverse the 0.8mm aperture? A simple yes or no would be efficient.

[User-462]

@srozum

Sensor 6 is quite opposite. It's opening should be 1.5-1.7mm straight hole.

I did not see this in the wiki assembly instructions. The instructions for sensor 6 only say it is similar in construction to sensor 2, so it is logical to assume a 0.8mm hole. Double the diameter should make quite a difference in calibration and sensitivity.

[User-462]

@Surreal62 @srozum Since the type of printer (SLA vs FDM) and other variables could drive changes in the design dimensions, namely the small pilot holes, I'm considering posting the Solidworks files. It could be a narrow audience, but this would let the community push any innovations.

Solidworks is normally expensive, but they now offer a low-cost hobbyist license. So, it is a learning opportunity for those who want to develop a new skill. I've used both Fusion 360 and Solidworks and I find Solidworks a much more mature, consistent, and intuitive product.

@srozum - Would it be more appropriate to go with the fork or PR method, or do you think the file attachment method is fine? I do think it is a good idea to maintain the open source licensing on all the material. If the file attachments in the discussion thread still fall under that umbrella, then that is the path of least resistance for me.

[Surreal62]

@User-462 the files would be greatly appreciated. I had started inroads to make alternative sensor housings to facilitate components and sensors being fitted with plugs/sockets so everything is 'swappable' for replacements/upgrades. More of a modular approach.
I have an order in place at JLCPCB with your STL's and will see if there are issues.
(Incidentally, some components are unable to be FDM printed by them, as some small dimensions are beyond their capability)
Pricing with them is such that if they prove unsuitable dimensionally, which I seriously doubt as the holes etc. can be filled and remade, it is no great loss anyway.

[srozum]

Since the type of printer (SLA vs FDM) and other variables could drive changes in the design dimensions

My STLs were designed for FDM printers and have 0.1mm tolerances added where needed. People who printed using SLA or JLCPCB reported good fit, so It seems that FDM designs work well with other types of prints.

Would it be more appropriate to go with the fork or PR method, or do you think the file attachment method is fine?

I understand that it's a bit of learning, but forking (making a copy of this repository into your account) and committing new files there is the best approach. Thus you will have the ability to keep different versions/iterations of your files organized. For example, this fork with schematics.

https://github.com/aconbere/film_camera_tester/tree/master/schematics

[User-462]

@srozum Thanks for the link - that is a good illustration. I agree about the fork - it's more organized that way. I'll read up first and then pursue it.

[Surreal62]

srozum said: "My STLs were designed for FDM printers and have 0.1mm tolerances added where needed. People who printed using SLA or JLCPCB reported good fit, so It seems that FDM designs work well with other types of prints."

Indeed the SLA printed parts I have from JLCPCB are a very good fit indeed. I only had to make very minor adjustments of the frosted screen surround to fit to the case I bought, which was to be expected as I couldn't find the identical case used in the original.

[User-462]

Update on forking the repository - I've been wading through the instructions, which keep calling for additional utilities to complete it.
Life intervenes as well, so it will be completed when I can, and then will post the Solidworks files to allow tweaking.
Happy New Year!

[ic-racer]

Maybe to increase accuracy the photosensor could be closer to the front opening. The issue with a 'straight' hole to the photosensor is that only the center sensor receives light straight ahead. The other sensors need to accept light at an angle. There are not enough photons from the light source face straight ahead of the two outer sensors; they need to accept photons from the entire face of the light source, and most of these photons will come in at an angle. This is all theoretical. Some testing could tell for sure one way or the other if countersinking the holes is a good idea or not.

[srozum]

Dear @ic-racer, I don't know who you are or what is your expertise, but you keep pushing some ISO quotes without seeing the obvious. This Tester uses diffused light and measures WITHOUT the lens. It's a different system and works differently. Please, stop confusing people.

[Surreal62]

@ic-racer
srozum beat me to it.
Only a single-point light source would have the issue you speak of.
Diffuse light is even (or should be if correctly assembled) across the whole frame with this tester, as it is with the likes of Kyoritsu testers

[ic-racer]

Sorry, not to confuse anyone. Actually what I was referring to was that if I have my camera too far from the light source, there is not enough light to trigger. Why would that be the case if the photons went straight from the light source to the hole in the face of the sensor. (of course with a laser, the distance does not matter). So, as I mentioned theoretically, the hole in the sensors receive light from the entire screen, otherwise there is not enough light to trigger.

This is a picture of the sweet spot where I get the most reliable readings and readings that match known calibrated shutters and I presumed you knew the camera cannot be placed too far away as the instructions read:

"...place camera in front of [light source] approximately 5-10cm away"

[srozum]

So you saying that your sensor (which you drilled all the way in) works better at the distance of 15cm instead of 5-10? Have you tried to assemble one according to the instructions? Anyways, if you know your tool and it works, then what's the question?

of course with a laser, the distance does not matter

I am sorry to disappoint you, but inverse square law applies to lasers just like to any source of light, even the stars. That's why we have seasons like winter and summer. If you don't see the difference using "laser" LED, you just not far enough.

[ic-racer]

if you know your tool and it works, then what's the question?

The closer my sensor and shutter are to the light source, the longer the recorded exposure times. My best correlation to what I believe is a correct shutter is at the distance in the picture. It behaves as if the readings reach a asymptote as the distance increases.

I do get a small discrepancy based on the orientation of the sensor. I believe that is from the sensor/hole alignment. I had been ready to make another when I saw this thread giving me the possibility of making one different from the first.

I did make one where I glued the sensors in place, then ran wires through a hole in the board, but the glue spread into the hole so I abandoned that technique, though was considering trying it again.

Have you tried to assemble one according to the instructions?

Yes, the issue I had was the hole in the PC board was not centered on the lens of the sensor.

[ic-racer]

Agree the Kyoritsu is good to model. I don't know much about the design of the sensor though I have tried to find out as much as I could without actually owning a Kyoritsu. I have found some good pictures of the Rollei tester's sensor:

[User-462]

Are the vertical slits the apertures for the sensors?

[Surreal62]

That isn't a sensor for a Kyoritsu.
I recall seeing that used on a 1960's EKZM tester or suchlike - three analogue meters on the front of it.

[srozum]

Here, google search works for me

[ic-racer]

Yes, EKZM is pictured, I don't have any info on the Kyoritsu's sensor. I have downloaded and gone through the Kyoritsu EF-8000 user's manual to find out as much as I could.

[ic-racer]

Are the vertical slits the apertures for the sensors?

Yes, in the EKZM, which is a device from the 1960s, I presume the sensors are larger and less sensitive than the sensors in this FILM CAMERA TESTER project, so a slit would allow more light to pass than a hole. I believe the EKZM uses a capacitive charging circuit to tell the time, but I have not seen the schematic.

[ic-racer]

Just re-testing my existing sensor and in addition to confirming the distance/speed findings in the graph below, I also note that the orientation did not affect the readings. Distance beyond 250mm did not activate the sensor

So, as was pointed out:

if you know your tool and it works, then what's the question?

I'll still build this new sensor as a spare and also to compare to the existing one. These sensor modifications in this thread do look like they will simplify construction. I will see.

[Surreal62]

One question. How were the original 'reference' shutter speeds on the test camera obtained?

[srozum]

You just demonstrated how inverse square law of light works.

In this case, the further you holding the camera from a light source the less light hits focal plane. Thus sensor (keep in mind that it has fixed amplifier and trigger) reaches trigger level at different time. Your graph demonstrates that it causes variation ±25µs (1/2000 is 500µs and 1/1800 is 550µs, so ∆55µs).

Let's use your data for range 6-18cm, then ∆36µs, or ±18µs. Now, if you hold camera within recommended 5-10cm range, then error would be within 10µs. Just like described in technical specs of my tester. Of course your data is not complete, but I hope you got the idea.

Kyoritsu and ZTS sensors use exactly same schematics: amplifier plus trigger. If you think that Kyoritsu doesn't have such effect, read its operation manual carefully. It describes proper testing conditions for every type of test and/or camera. And its calibration procedure consists of setting trigger voltage.

I don't have access to Kyoritsu tester, but I am more than confident that you will observe same behavior. Plus, because it tests WITH lens, the lens itself (aperture and focal length) will add to the error.

[Surreal62]

Kyoritsu/Arrowin/ZTS testers do behave in exactly the same way as Srozum's tester does. Physics dictates it to be so. As a matter of interest, from memory, none of those machines camera attachment stages will allow distances of >100mm from their light source to be used with them.
There is another similar tester available from a supplier in Canada - https://www.reveni-labs.com/ that will also behave in a very similar way.
It is good practice to have the camera as close to whatever calibrated light source is used so as to exclude extraneous light - (body only 5cm+/- and body with lens touching the diffuser screen) -otherwise there is little point in having a calibrated light source, as ambient light can and will affect accuracy.

[srozum]

One question. How were the original 'reference' shutter speeds on the test camera obtained?

He uses Nikon F6 with electronically controlled vertical shutter. Which also means that speeds are power of 2, and on his graph expected correct speed is 1/2048 (or 488µs). Even given its age, I personally, would trust that camera too. Although aggregation across various speeds would produce better data, because even on F6 speeds above 1/1000 are not guaranteed.

[Surreal62]

I have pretty much just repeated this experiment with distance from light source to camera and I do not see any appreciable changes in recorded shutter speeds as distance varies. Anything from touching the light source to 200mm+ away I get the same speed as set, and accurate to a known shutter. Moving away eventually caused no capture, but that is to be expected as the light level would be too low to trigger the sensor.
Ambient light level was very low during the test.

[Surreal62]

A variation towards a modular sensor setup.

With thanks to the work of User-462, I used them to build 'rebuildable' sensors.

[User-462]

@Surreal62 That's a cool idea. Thanks for the pics. The SLA prints look nice.
In the last image - are those connectors on the diodes?

[Surreal62]

Yes - I desoldered the push connectors from the laser receiver boards, slightly adjusted the leads to fit the module PCB.
Careful forming of the photosensor leads and fitting into the connectors gave a really good fit into the faceplates so everything can be disassembled.
To fit the assembly to the faceplate only required the use of two of the plastic washers used on the light source unit and everything fits as you see.
The results I'm getting with the tester are very good indeed - the sensor module is a very good fit into the SLR's and rangefinders I have tried so far by the way.
I am starting building sensor 6 in the same manner, but am awaiting delivery of the correct photodiode.
I'll keep updates coming as and when. 👍

[User-462]

With a right angle connector in the right place, you could take advantage of the zip-tie to stabilize the cable. That would be a redesign.

One risk of adding connectors, especially the multi-pin connectors is that they add potential failure points with all the pins and crimping. Especially with all the handling these sensors will receive. On mine, I stabilized the soldered wires with clear hot glue, and they feel rugged. I also added staggered sizes of heat-shrink to the cable where it enters the case. It stiffens the cable from flexing at the entrance hole.
The diode connectors do have a nice tight fit and fewer parts to fail on the other hand. It would be helpful to have swappable diodes for initial testing and setup. Plus, soldering the diodes with higher heat for lead-free solder adds the risk of damaging them.

[srozum]

yeah, I am not getting the need for connectors.

[Surreal62]

@User-462 this is a work in progress. Cable strain relief is not in place yet. Those will be made to measure and IPC 620 compliant. When I build the 'working' cable sets I'll be using silicone insulated wire to give supreme flexibility, but these will do for testing
. It is handy having experience in defence electronics manufacturing and maintenance with test rigs running at up to 200G and temperature ranges of -80 to + 85°C for accelerated aging of 'things'. There are some very advanced cabling solutions in those rigs, that's for sure, although this setup won't have the hard use those will. Supplies of bits and pieces are easily available to build cables that could last a lifetime! 😂
As for connector wear, that won't be a real issue as the connectors are rated for 1000 insertions. I doubt there will be that many changes. If it were, I'd have opted for pogo-pins.
There is no lead-free solder used in my builds - and I have plentiful supply of DLMP22 solder that will take more years than I have left to use! 😉

[Surreal62]

[srozum]

Are you sure that you understand difference between radius and diameter? :-)
Those are definitely not a ø0.8mm.

[Surreal62]

That drill bit shaft is a quite firm sliding fit into those holes. It does not wobble about nor falls through
The next size down at 0.7mm easily falls through
The next size up bit at 0.9mm will NOT fit into the holes.

Edit: I wasn't going to type this originally, but I DO UNDERSTAND the difference between radius and diameter. I am no fool.

[ic-racer]

Nice pictures above of the Kyoritsu sensor!

FYI if anyone is interested, there is a Kyoritsu listed in Seattle Craigs list. I'm heading to Seattle next month, but I'm telling my self "NO" as I'm very pleased with the tester I made. I have tested about 30 shutters and adjusted about ten electronic and mechanical focal plane shutters so far with the film_camera_tester.

[Surreal62]

Stick to your 'NO' and save yourself the $2500 ono.
The Srozum tester does the same job very well indeed as the Kyoritsu to all intents and purposes - and it can be repaired should any component fail. 👍

[ic-racer]

One question. How were the original 'reference' shutter speeds on the test camera obtained?

Nikon F6, purchased new 2020 from B&H, maybe the last one they had. SN: 35,xxx. Also that camera was confirmed with an oscilloscope tester which was itself confirmed with a rotating focal plane shutter of known angular velocity and slit width.

[ic-racer]

You just demonstrated how inverse square law of light works.

Yes! Because the sensors receive light from the whole light source surface. If they were little telescopes, each only viewing 0.8mm of the light source, there would be no falloff within distances that could be measured indoors.

Again, the reason I brought it up was to clarify why I undercut the holes. In keeping with 'simple construction' and eliminating the need for a drill press.

One thing I did during construction was to check the module in both orientations to make sure one of the holes were not mis-aligned (giving a slower speed than the other).