Sorry for my numerous attempts to create this thread but it looks like there is bug somewhere redirecting us to another group when we access to this thread.
This problem seems to not occur which short thread titles.
A new led light source for my Omega D5 condenser enlarger.
This is my second attempt with making a proper light source for my Omega D5 condenser enlarger based on color LEDs. My first attempt consisted in a large network of green and blue leds (diffusing mode) combined with two highly powerful 15W green and blue leds used as punctual source (placed in the middle of the board). Both green and blue channels were current controlled independently. The head could be used as a diffusing light source with a piece of diffusing plastic window instead of the 3x condenser lenses or as a condenser light source when the light source was replacing the standard light bulb.
Omega D5 new light source V1 by stormpetrel_geek_mode, on Flickr
This light source can be seen working on the video below. Sorry it is in french but the interesting bits are at 1m30 and at 2m35.
At this time, I was not enough aware with the principle of a condenser enlarger. There were no problem with the diffusing setup but at the end I realized I would prefer to use my enlarger mainly as a condenser enlarger. Unfortunately I did not take into account the formation of the image of the light sources on the film plan (slight offset between the green picture and the blue picture) , said differently the light source should be perfectly uniform which would require the use of a diffusing dome as the two main LEDs do not act as a perfect unique punctual source despite the short distance separating them). Unfortunately there was not an easy way to integrate such diffuser on my first prototype and also due to the asymmetrical design. The diameter of the diffusing dome is also critical in this application as the condenser creates a magnified image of the source which have to cover the full surface of the film. (small dome => bad film covering). This is not a surprise for the experienced "darkroomers". Ideally the size and the position of the diffusing dome should match the size and the position of the light bulb recommended by the enlarger manufacturer.
The second problem was the wavelength of the LEDs (navy blue and green). As we will see below they are not the most suitable wavelengths, so I have done a bit of research optimizing the new light source.
I would like to share those information with the community as it might be useful for your future own design.
* photographic paper sensitivity
I did an extensive review of the the light sensitivity of all the photographic papers available on the market nowadays. Unfortunately it was not possible to get the datasheets from all the manufacturers but still we have enough information to optimize the wavelength of our leds.
I have split the compilation into 3 groups: normal variable contrast papers, warmtone variable contrast papers and fixed grade paper.
The light sensitivity unit is an arbitrary unit. The interesting information here is at which wavelength each paper is the most sensitive and at which wavelength the control of the grade is done.
Normal variable contrast papers:
Variable contrast paper sensitivity by stormpetrel_geek_mode, on Flickr
As you can see on the graph, most of the papers are quite sensitive in the violet domain (400-425nm). Note the light sensitivity is a log scale and the sensitivity of many paper drops significantly in the blue. Regarding the control of the grade, this happens between 500nm and 575nm but mainly between 520-525nm (yellow green).
Warmtone VC papers:
Warmtone variable contrast papers sensitivity by stormpetrel_geek_mode, on Flickr
For the warmtone picture, the Ilford paper behaves like the normal VC papers however this is not the case with the other papers. Grade control is performed between 520-530nm for the Foma paper and at above 545 for the Forte paper. In the low side of the spectrum, we see that the Ilford paper is more sensitive in the violet area compared to the 2 others papers which are more blue sensitive (I wonder if there is a mistake in the Forte paper datasheet).
Fixed grade papers:
Fixed grade papers sensitivity by stormpetrel_geek_mode, on Flickr
Fixed grade papers are completely different beasts. There are very sensitive to the green light (475-500nm).
Based on all those information, I have selected the following wavelengths for the LED of my new light source: 420-425nm violet and 520-525nm (yellow green). Those are the wavelengths where I will have the best bangs for my bucks based on my use (variable constrast papers). The new light source consist in 12x 3W 520-525nm green LEDs and 9x 420-425nm violet 3W LEDs spread on a 6cm of diameter PCB. The second board with 4x far red 1W led is stacked onto the first board (the red LEDs are used as an inactinide light source). This boards is mounted on the first board.
Omega D5 enlarger head - leds V2light source by stormpetrel_geek_mode, on Flickr
Omega D5 enlarger head - leds light source V2 by stormpetrel_geek_mode, on Flickr
The diffusing dome comes from a Philips energy saving 18W bulb (OD=69mm)
philips-ambiance-soft-e27-18w-100w by stormpetrel_geek_mode, on Flickr
Next post: illumination modelling.
Ok this model is over killing but it was for me a way to check that the diffusing dome + LEDs light source act as a uniform source. This would have saved me $200 with my first prototype! It is also an exercise for a more ambitious project (a light 8x10 enlarger using a Fresnel lens from an overhead projector).
Optical modelling is quite useful to check a prototype however keep in mind that this model mights differ significantly from the reality.
As I could not find any information regarding the focal length of the lenses used in the condenser of my Omega D5, the focal lengths used in the model are based on the geometrical dimension of the real lenses.
Omega D5 enlarger - optical model by stormpetrel_geek_mode, on Flickr
The light source is this one described previously with 12x yellow green LEDs and 9x violet LEDs. The diffusing dome is based on the Philips energy saving bulb (OD=69mm).
The enlarger lens is made of single a converging lens (not necessary, just for fun).
The illumination modelling is performed with a ray tracing software.
Omega D5 enlarger - Ray tracing by stormpetrel_geek_mode, on Flickr
Here are the results:
* No plano-convex lens in the 135mm tray (150mm lens configuration).
Green light distribution on a 4x5 cut sheet film:
Illumination of a 4x5 cut sheet film with the top plano-convex lens placed in the 150mm tray (green leds only) by stormpetrel_geek_mode, on Flickr
Violet light distribution on a 4x5 cut sheet film:
Illumination of a 4x5 cut sheet film with the top plano-convex lens placed in the 150mm tray (violet led only) by stormpetrel_geek_mode, on Flickr
Both violet and green light distribution on a 4x5 cut sheet film:
Illumination of a 4x5 cut sheet film with the top plano-convex lens placed in the 150mm tray (violet+green leds) by stormpetrel_geek_mode, on Flickr
Those results look promising. The illumination is even on all the surface of the film and we can not see any offset between the green and the violet channels (at least at this resolution).
* Plano-convex lens set in the 135mm tray (135mm lens configuration).
Illumination of a 4x5 cut sheet film with the top plano-convex lens placed in the 135mm tray (violet+green leds) by stormpetrel_geek_mode, on Flickr
We can see the light coverage is not as good as previously however the illumination is stronger at the center which makes this setup a better option for 6x9 film enlarging.
Note the blueish color is different from the previous case due to a different mixing ratio of the green sources with the violets sources.
I have also simulated the illumination of the print just for fun. The focus is not properly done and the single lens suffers from all the optical aberration existing on earth but still, we can have a rough idea about the illumination of the print (here 40cm x 32cm):
print 40cm x 32cm - 150mm tray (violet+green leds) by stormpetrel_geek_mode, on Flickr
It is time to switch on the solder iron and check that this new light source works as expected!
The LED driver.
The easiest way to light a LED is to connect the LED to a voltage power supply with or without a limiting current resistor (depending on the voltage applied). PWM or pump charge circuit might also be used. Unfortunately those are not good solutions to have a constant light emission.
When a fixed voltage is applied to a LED, the current going through the LED drifts due to the temperature increase in the junction. As the number of photons emitted by the LED is proportional to the current crossing the junction, the light emission changes till the thermal equilibrium is not reached. The best way to stabilize the light emission is to drive the LED with a stable current source. This is how laboratory lasers are controlled (a semiconductor laser is very close to a LED)
I propose here a programmable current source using the OPA548T which can drive LEDs up to 3A at 20VDC
LED driver by stormpetrel_geek_mode, on Flickr
This designs works perfectly well as you can see on the video posted at the beginning of this thread.
Ideally with U/I power amps, resistors should be matched (making a perfect U/I converter) but in our case (unipolar design + OPA548 technology) it creates an unstable situation when the driver is switched on. This problem is solved by increasing slightly some resistors value (1.1k instead of 1k and 49.2 instead of 47k) introducing a shift in the function transfer (see simulation below).
LED driver transfer function simulation by stormpetrel_geek_mode, on Flickr
This simulation also shows the effect of the current limiting resistor programmed at 1.02A (the limiting resistor here is 56k instead of 39k as used in the schema above).
It is important to program this resistor value accordingly to the specs of the LED to avoid blowing up your LED!
The driver is fully operational and here is an example of a measured function transfer (the current limit is programmed here at a higher value)
LED driver transfer function measurement by stormpetrel_geek_mode, on Flickr
The driver can be used in different ways. An easy way would be to control the input voltage with a pot (manual control) but in my case I prefer to use a multi-channel 16bits Digital to Analog converter allowing the control of 5 LED drivers simultaneously.
This DAC is controlled by a home brewed PC software/µc board. Thanks to this solution, it was possible to go further in the control of the light emission by doing a linearization of the transfer function curve above. It is then possible to set directly the wanted intensity of light instead of adjusting the tension....