DIY: Durst M601 LED head

[ucsugar]

This was done before, but for those trying to do something similar this might be some inspiration:

I built a LED head for a Durst M601 enlarger. It serves as a 'drop in' replacement for the original CLS66 color head. No modifications on the enlarger itself are required, and also no 3d-printing involved. I must admit, the project stopped in the 'first working prototype' phase, because the first prototype does the job surprisingly well.

The light source:
* 3x 3W LED green
* 3x 3W LED 'actinic blue'
* 2x 3W LED white for focus
* 1x 3W LED red
These are LEDs pre-fitted on a 'star-aluminium' board, I have mounted them on a 6x6cm aluminum board, around that a cube, consisting of 4 acrylic-glass mirrors and a plexiglass acting as a diffusor.

The original "Sivobox" (a cube with a reflecting inner side, I assume it acted for mixing/diffusing the light that came from the CLS66 head) is still in place, light exits the acrylic-mirror-cobe just above the Sivobox. I never did a scientific test, but there are no visible hot-spots.
The mirrors are just glued together (still in place after 18months), the board with the LEDs and the plexiglass diffusor screwed and a PVC-board - and that' it mechanically - it really comes handy that the M601 was designed with a quick-changeable light source in mind.

Electronics:

Meanwell LDD, LED drivers, PWM controlled by a ATTiny. Two channels, a relais switching between green/blue and white/red. A I2C connection (yes) to a control box, where a ATmega88 does the rest.... PCBs designed with Kicad. An other connection goes to a I2C controlled red-light (just a bunch of red LEDs)

It has two modes, one emulating different filters by mixing the green and blue LEDs and a "split-grade-mode" using the green and blue color independently.

What I like:

* saves a lot of space (transformer and timer)
* makes split-grade-printing much easier and less error-prone
* about 1 stop faster than the original CLS66 with 100W halogen bulb
* oh, and the controller can serve as a development-timer, saving even more space! (My dark'room' is a dark corner in my bedroom, space is at a premium)

Pitfalls:
* paper seems to be very sensitive to blue (actinic blue, 430nm). I "calibrated" the unit so, that equal time green/blue gives the same contrast like a filter 2: blue operates with ~25% of the power as green, so its' really dark for the human eye, making dodge/burn with blue a bit difficult.

 

[koraks]

Nicely done!

I must admit, the project stopped in the 'first working prototype' phase, because the first prototype does the job surprisingly well.

I think we have a similar philosophy to these things, hehe.

paper seems to be very sensitive to blue (actinic blue, 430nm). I "calibrated" the unit so, that equal time green/blue gives the same contrast like a filter 2: blue operates with ~25% of the power as green, so its' really dark for the human eye, making dodge/burn with blue a bit difficult.

Yes, I recognize this. There's two things you could do (if it bothers you enough, which I doubt):
1: Move to 450nm blue instead of 430nm. There's no real advantage to wavelengths shorter than 450nm, but things get progressively worse in terms of human visibility as you go down in wavelength.
2: Turn on the red LED along with the blue ones during exposure. B&W paper shouldn't care about this (test for fog and reduced contrast!), but it'll make a difference w.r.t. being to see what you're doing.
Btw, in all honesty, I'd probably just leave it as it is.

A I2C connection (yes) to a control box

Does that work well for you? I just re-did a project that involved a similar setup, but the 'long-distance' I2C was just too buggy to be comfortable. This was kind of a hostile environment and worst-case scenario, though. In a prototype setup it worked OK, but once noisy LED power supplies were around, things went down the drain pretty fast.

 

[ucsugar]

Nicely done!

Thanks!

I think we have a similar philosophy to these things, hehe.



Yes, I recognize this. There's two things you could do (if it bothers you enough, which I doubt):
1: Move to 450nm blue instead of 430nm. There's no real advantage to wavelengths shorter than 450nm, but things get progressively worse in terms of human visibility as you go down in wavelength.
2: Turn on the red LED along with the blue ones during exposure. B&W paper shouldn't care about this (test for fog and reduced contrast!), but it'll make a difference w.r.t. being to see what you're doing.
Btw, in all honesty, I'd probably just leave it as it is.

Thought about it already, but I made this mistake of only two channels and switching between them, so I can't add red while blue is on. I just have received my first aluminuim-PCBs for another project which was surprisingly easy; so I would build it with a lot of e.g. 0.5W LEDs instead a few 3W. But it works too good as is.

Does that work well for you? I just re-did a project that involved a similar setup, but the 'long-distance' I2C was just too buggy to be comfortable. This was kind of a hostile environment and worst-case scenario, though. In a prototype setup it worked OK, but once noisy LED power supplies were around, things went down the drain pretty fast.

To my surprise it does, this was also something intended for prototype, but good enough for use. I have only two short cables, <1m. The red-light once in a while loses a packet, the head itself never did. Runs with 5V on lowest speed, no serious currents involved in this tiny setup.