Building A Professional Grade Shutter Tester

[ic-racer]

How are you measuring the output? The LEDs in the left of the picture look like the ones I wound up using.

 

[billywei]

How are you measuring the output? The LEDs in the left of the picture look like the ones I wound up using.

I used a Minolta F spotmeter shooting directly agaist the center of the diffuser panel. It reads in 0.1EV step. I also had a closeup len on my spotmeter and it didn't make any different w/wo this close-up len. I'm thinking about borrowing a Luminance meter from a friend to measure it with higher accuracy.

I also noticed that the spotmeter reading was the same with my Nikon DSLR. However, the readings of all my manual focus body (FM2 F3 F5, etc) were one stop lower than the spotmeter and DSLR. In the light source section on wiki, it says "Because of the Color Correction Factor of LED, old light meters like selenium meters, photoresistors or LDR, and some photodiodes read exactly 1 stop lower. More modern light meters, including digital cameras, read LED lights correctly." So I surpose that is expected, right?

 

[FM2]

Here comes the photo of both LEDs I get, the new arrival on the left and previous one on the right.

In my eyes the new LEDs still look like the wrong ones. The darker gap on the side of the LED lens looks untypical. I’d try to ask one of the AliExpress sellers if they have a taobao store and go with the ones they sell.

 

[ic-racer]

I used a Minolta F spotmeter shooting directly agaist the center of the diffuser panel. It reads in 0.1EV step. I also had a closeup len on my spotmeter and it didn't make any different w/wo this close-up len. I'm thinking about borrowing a Luminance meter from a friend to measure it with higher accuracy.

I also noticed that the spotmeter reading was the same with my Nikon DSLR. However, the readings of all my manual focus body (FM2 F3 F5, etc) were one stop lower than the spotmeter and DSLR. In the light source section on wiki, it says "Because of the Color Correction Factor of LED, old light meters like selenium meters, photoresistors or LDR, and some photodiodes read exactly 1 stop lower. More modern light meters, including digital cameras, read LED lights correctly." So I surpose that is expected, right?

Make sure the K is set correctly. For Nikon it is K= 14.03. Some meters (like Sekonic) are K = 12.5
Also, when testing Nikon, I use center weigh mode rather than Matrix. A complete illuminated field might confuse the matrix meter.

 

[billywei]

Make sure the K is set correctly. For Nikon it is K= 14.03. Some meters (like Sekonic) are K = 12.5
Also, when testing Nikon, I use center weigh mode rather than Matrix. A complete illuminated field might confuse the matrix meter.

But wikipedia says both Nikon and Sekonic are using 12.5. I'm wondering which one is the correct one.

Exposure value - Wikipedia

en.wikipedia.org

 

[ic-racer]

That might be for older classic Nikon. I'll try to find the reference.

 

[billywei]

That might be for older classic Nikon. I'll try to find the reference.

Since the light unit is using PWM pulse to drive the LED, I think it is possible to insert another Arduino Nano to measure the PWM signal output from the shield board and then output a corrected PWM signal corresponding to my light unit. The code should be fairly simple and I would like to give it a try.

 

[ic-racer]

Since the light unit is using PWM pulse to drive the LED, I think it is possible to insert another Arduino Nano to measure the PWM signal output from the shield board and then output a corrected PWM signal corresponding to my light unit. The code should be fairly simple and I would like to give it a try.

Exactly, I was going to suggest the same thing but wasn't sure people would understand it. You should try it!

My interest in doing that would be so I could use my REAL CREE lamp with closer to 4700k output. The intensity was ok in the middle range but the linearity was off.

 

[ic-racer]

ISO 2720 on exposure meter calibration specifies 4700k, but also says you can use a filter to check to spectral sensitivity. There is no standard, just the difference between 2855.6K and 4700K responses are reported as a ratio. The smaller the better.

 

[ic-racer]

That might be for older classic Nikon. I'll try to find the reference.

The reference to modern Nikon and K14 was in a personal e-mail with the creator of the project:

ic-racer, In your test table I see mix of old and new cameras, which were designed to meter differently. For example, Nikon F100 calibrated to K14.03, and it uses modern photodiodes, so Light Source should be set to K14 and 0EV shift in settings. Also, pay attention that F100 has different metering modes, you might need to switch it into average mode, not centered. And when calibrating, you should use 50mm lens on your Nikon f100, and place it directly against the screen at the center. --Serhiy Rozum

 

[billywei]

Exactly, I was going to suggest the same thing but wasn't sure people would understand it. You should try it!

My interest in doing that would be so I could use my REAL CREE lamp with closer to 4700k output. The intensity was ok in the middle range but the linearity was off.

OK. I went one step further by using a little scope to measure the PWM output from the shield board. The result is quite interesting. The project creator is not using a pure PWM, instead, it is a mixture of PFM and PWM. Here comes the result I get:

This make it a little bit complex for the Arduino program since now it has to measure both the frequency and pulse width of the incoming control signal.

I'm looking forward to a national holiday vacation with my kids. So I will start working on the Arduino program after the trip.

 

[ic-racer]

Fantastic reverse engineering!

I don’t know enough about this to comment as I thought Arduino only output 500 Hz

 

[billywei]

All right, I managed to finish my very first version of the "Tuner" before my vacation. And here comes the solution.

The hardware modification is simple, just grab another Nano (maybe the one you used as programmer), connect Vin and GND of the added Nano to the one on the shield board. And then cut the Light PWM wire connecting shield board and the light unit board in the middle. Reconnect the wire on the shield board side to D2 on the added Nano and the light unit side to D12 as shown bellow:

Now your are all setup for the hardware mod. Just download the attached ZIP file, unzip it and open the .ino file with you Arduino IDE. Look for the following section:

These are the parameters corresponding to each EV level you'd like to control.

Column 1 controls the clock divider. The larger the number the slower the control signal frequency goes. Just leave it untouched.

Column 2 controls the output frequency of the control signal. It's the comparator number in a timer. Once the number is reached, the counter will trigger a pulse output. So with the same pulse width, smaller number in column 2 means faster output pulse train and higher output power(brighter light).

Column 3 and 4 controls the pulse width. A larger number provides a longer pulse which leads to brighter light.

Try to play with column 2,3 and 4. Tune them and upload the code to your added Nano with Arduino IDE. Check the result, change the parameters, upload and check again. Repeat the process until you're satisfied with the output accuracy and linearity. You will have to tune the whole table for each EV but it shouldn't be a a big deal.

Last but not least, play at your own risk! I will absolutely take NO responsibility to any potential lost done by my code.

 

[billywei]

I forgot to mention that with this modification I managed to tune all EV output to exactly the same reading on my spot meter. However, it also turned out that my LED still required 13V minimum to output 16EV. I will just live with it.

 

[ic-racer]

Wow!

 

[FM2]

All right, I managed to finish my very first version of the "Tuner" before my vacation.

Looks great, congrats! I guess this could be the end of most diffusor problems, fake Crees and LED restrictions in general – at least as long as the Kelvin value fits and there is enough light output. Maybe you can share it on the GitHub discussion section.

 

[ic-racer]

Before the discussion board was temporarily closed I was going to suggest to Serhiy that the LED pwm parameters be placed in the Settings File so the end user had the option of modifying them if desired.

 

[FM2]

I guess the dev has already accomplished what he needed for his own use. I doubt he will advace the project any further.

 

[billywei]

I guess the dev has already accomplished what he needed for his own use. I doubt he will advace the project any further.

I think he has done a great job to squeeze the last bit of potential from a Nano. There is almost nothing left for him to develop more functions. I figured out why he had to combine PFM/PWM together to control the light unit only until I started to calculate the actual requirement which required quite a high resolution for the dynamic range.

 

[FM2]

I remember now that he mentioned the limited memory space on the nano:

"The thing is, the firmware takes so much space that it doesn't fit into regular Nano board. For example, v2.5 uses 31380 bytes of flash memory."

I'm not even sure if any more code additions would be possible.

(source: https://github.com/srozum/film_camera_tester/wiki/Firmware)

 

[vandergus]

I remember now that he mentioned the limited memory space on the nano:

"The thing is, the firmware takes so much space that it doesn't fit into regular Nano board. For example, v2.5 uses 31380 bytes of flash memory."

I'm not even sure if any more code additions would be possible.

(source: https://github.com/srozum/film_camera_tester/wiki/Firmware)

There are Nano boards with more flash. The Nano Every is cheaper and has an upgraded processor with 48K flash. There are other Nano boards with even better processors and up to 1MB flash. Some are built around 3.3V I/O (I think the camera tester is built around 5V) but there are a few that would be drop in replacements. Pin-to-pin and software compatible.

What I'm getting at is that there are solutions to the memory problem. And it would be nice if Srozum would publish the source code so the community could continue to improve and develop the design. Especially since he is no longer building and selling units.

 

[ic-racer]

The eeprom, where the settings file is loaded, is 512 and the 'settngs' don't take much room. That is why I thought that would be a good place to store the table of PFM/PWM values.

 

[ic-racer]

. I figured out why he had to combine PFM/PWM together to control the light unit only until I started to calculate the actual requirement which required quite a high resolution for the dynamic range.

Amazing detective work!

 

[FM2]

Finally found time to work on another sensor.

One thing I noticed, in the assembly instructions for „Sensor 6“ it says to connect „ID“ to „A“ and vice versa. I guess this is a mistake, and I went with the markings on the board.

However, I can’t get sensor 6 to work. I bought the Sensors from one of the AliExpress links here in the thread. On the back they have a „237LF“ mark. So I guess they are the right ones.

Without the photodiode soldered in the sensor is recognized by the tester and I can access the „exposure error“ mode. With the photodiode soldered onto the sensor-pcb the tester either won’t boot or crash on startup. Does someone have an idea what’s going on? Did I buy the wrong photodiode?

Assembly Instructions

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

github.com

 

[ic-racer]

"Sensor 6"

I noticed that it needs to be in it's enclosure to calibrate. I tried to test it (before fitting it in the enclosure) but just a little stray light messes up the calibration making it appear dead or non-functional.
After fitting it in the enclosure, it worked perfectly.