LED Split Printing Enlarger Lamphouse

[Leukonarc]

Wow it looking good, it'll work like a charm.

 

[koraks]

No PWM inside the Buckblocks but it looks like they are fully capable of pulse or strobe.

I see. The implication is that there is a slight non-linearity as the current/brightness relationship of leds is not perfectly linear. But I don't think it's a problem in this application.

 

[albada]

It would be nice to have an LED controller with two knobs: brightness and contrast.
That would allow a worker to change one without affecting the other.
"It's perfect, but needs a little more contrast." No problem, just turn the contrast knob a little. Same thing when contrast is good but brightness needs to change a little.
Here's my hand-scribbled schematic of a prospective controller with those features:

There is a 3-position paddle-switch near the upper left having these positions:

Timer - LEDs are off (controlled by timer),
Focus - LEDs are on at 100% PWM, maximizing brightness (but preserving contrast, as if that matters when focusing),
Preview - LEDs are on with normal PWM.​

Rg and Rb control contrast. I'll start with two pots, and mark their scales with grades 00 to 5 in half-grade steps. If that works well, I'll measure resistances at those positions and buy 24 resistors and a 12-position rotary switch.
The 555 chip controls brightness via PWM, which is about 100 Hz. It's on a regulator because the 555 is rated to only 16 volts, well below the 24V needed for the LEDs.
Koraks suggested using NPN transistors to gate the Buckblocks. They're here. He also mentioned an LM350-T. I'm using an LM340-T, but the LM350-T would work equally well.
But my big question is: Did I make a mistake? I am a software engineer, not a EE, so I designed this using Ohm's Law and Internet searches, so I'm not confident in this design.
Mark Overton

 

[koraks]

In general terms it should work, in principle (I haven't verified the schematic in detail, only its general operation principle).
One question and one remark:
Question: What's the purpose of being able to control brightness? Usually you set exposure with the timer or, in an alternative system that's barely used by people these days, by varying the lens aperture. In practice I have never found a need to vary the overall brightness of an enlarger light source. How will it make your life easier? Not saying it's wrong; just curious.
Remark: keep in mind that the contrast-exposure relationship of papers is not linear. If, for example, you expose for just hitting dmax in the deepest shadows at varying grades, you'll find that there's a curvilinear relationship between the contrast grade and the required exposure. So the lowest and the highest grades require the most exposure, while the middle grades tend to require the lowest exposure. This does depend a bit (much) on the actual output of the LEDs, but overall, with today's blue and green LEDs, this curvilinear relationship presents itself. This may or may not pose a problem in practice for you; the implication is that whenever you change the contrast, you'll also have to change the exposure by varying the time, the aperture of the lens or of course the brightness setting on your circuit.

 

[radiant]

What's the purpose of being able to control brightness?

If you expose more faster material such as film you need dimming?

 

[albada]

koraks asks: "What's the purpose of being able to control brightness?"
I quickly discovered that today's papers are more sensitive than those of 40 years ago. For example, my 50 mm lens is sharpest between f/5.6 and f/8, and at f/8 with a grade-2 filter in the enlarger and a 100-watt (equivalent) LED-bulb in the condenser light-head, correct exposure is about 7 seconds. That's not enough to use dodging or burning tools, so exposure needs to be lengthened, and I don't want to stop down the lens past f/8 because that will reduce sharpness. So dimming the light is appealing.
Also, I plan to calibrate the contrast adjuster to yield the same overall density at all contrast settings, which I think should compensate for the curvilinear relationship you mentioned. But that relationship surprises me, as Ilford states that their MG filters require no change in exposure (though they must be doubled at grades 4 and up). Are you using Ilford's filters for selecting grades?

Mark Overton

 

[MattKing]

But that relationship surprises me, as Ilford states that their MG filters require no change in exposure (though they must be doubled at grades 4 and up).

That feature is based on speed matching of a particular high midtone/low highlight tone.
So if you aren't analyzing other details based on that target, it won't apply.
By the way, that need to double exposure at higher grades has been modified with the newest Multigrade RC papers.
I wonder whether a toggle switch - low, medium, high - might be a more useful/repeatable way of modifying light intensity.

 

[Mal Paso]

The multigrade filters incorporate neutral density to make the exposure come out right. Printing with blue and green is very different. You can see each part of the emulsion has a different speed and range.

I like to print 3-4 stops down with exposure between 10 and 20 seconds. I can extend the time if I need more. I like the continuously variable source and 10 turn pots give me repeatability. I'm going to see how this works before I add another layer of control. Even with an analyzer B&W is still about test prints.

 

[albada]

Mal, you are getting exposure-times of convenient lengths, so you don't need PWM.
A question: Have you considered using a logarithmic pot? Here's a suitable one: https://www.digikey.com/product-detail/en/tt-electronics-bi/P092N-QC15AR20K/987-1286-ND/2408863 .
This might expand the low-resistance (dim LED) end of the scale, uncrowding the numbers enough that multi-turn pots won't be needed to get sufficient repeatability.
I'd like to use two single-turn pots (G and B) with scales marked with contrast-numbers (00 to 5), and I'm hoping a log pot will space the numbers on the scale more uniformly.

Matt: Thanks for the info, especially that the doubling-exposure rule now works poorly. I haven't tried high contrast yet, so I didn't know.

Mark Overton

 

[koraks]

koraks asks: "What's the purpose of being able to control brightness?"
I quickly discovered that today's papers are more sensitive than those of 40 years ago. For example, my 50 mm lens is sharpest between f/5.6 and f/8, and at f/8 with a grade-2 filter in the enlarger and a 100-watt (equivalent) LED-bulb in the condenser light-head, correct exposure is about 7 seconds. That's not enough to use dodging or burning tools, so exposure needs to be lengthened, and I don't want to stop down the lens past f/8 because that will reduce sharpness. So dimming the light is appealing.

I see your point! Yes, it's certainly possible and not that complicated, so why not eh?

Also, I plan to calibrate the contrast adjuster to yield the same overall density at all contrast settings, which I think should compensate for the curvilinear relationship you mentioned. But that relationship surprises me, as Ilford states that their MG filters require no change in exposure (though they must be doubled at grades 4 and up). Are you using Ilford's filters for selecting grades?

Well, firstly, there's of course the somewhat difficult issue of 'same overall density' if you change the contrast. The same dmax for the clear areas on the negative, the same highlight density in a certain spot, or the same density somewhere in the middle of the curve? So it's important to choose what you'd like to keep constant. Personally, I decided to optimize for the black point, i.e. trying to maintain the same exposure across the various grades based on just hitting dmax on the paper. It's a personal choice as I find that 9 times out of 10, this is my starting point in determining exposure of a print (along with contrast, of course). If you find that you mostly optimize for the highlights, then your choice may be different from mine.
Secondly, the Ilford solution with their filters is apparently that they vary the density of the filters so that exposure remains constant at least across grades 1-4. So part of the 'linearization' is actually built into their filters. Whether they optimized for (keep constant) the shadow density, some midtone or highlights, I don't know. It's probably in the user manual of the filters.
Thirdly, papers are not created equally. They all follow roughly the same pattern - very sensitive to blue light, less so to green light, and a sort of sensitivity peak in the middle grades where blue and green are combined. But how big the differences are, depends a bit on the paper. So you may get it just right for one paper, but then you'll be off the mark for another paper. Turns out that especially if you use several papers, you'll still have to either do some kind of metrology or resort to tried and tested test strips (which is what I do) to really nail it.
Finally, as to the approach I use: in my case, it's LEDs. I built a RGB light source with a control unit for both color printing and B&W. As part of the B&W calibration I did some informal/quick tests on which I base the above observations. I can also tell you from first hand experience that the LED approach works fine for B&W (and arguably also for color, but it's more complex). But you already knew that!
In my post here you can read also a few comments on the B&W optimization I did, but this was of course within the 'limitation' that color was my primary purpose and B&W was sort of a secondary goal: https://www.photrio.com/forum/threads/craziness-using-leds-to-print-ra4-and-b-w.171911/

 

[Mal Paso]

Great discussion! E6 was my thing. I had a C41 tank line and access to a Kreonite dry to dry paper processor and like my current digital setup better. For me the LED light source was all about making a better black and white print.

The Multigrade Filters are an interface. They constrain paper to 12 grades to put a handle on things. The problem with interfaces is they get in the way of what is possible. The next step for me is to plot the power levels and see what they do.

I wish I had my old color analyzer, the probe was fairly full spectrum. I'd like to quantify the light hitting the easel. I have a Phillips 2000 with probe but don't know the color response. I may play with that today. It's not smoky in the darkroom.

 

[koraks]

I played around a bit with a Lici Colorstar and my led light source, but no luck. I think the pwm output of the leds doesn't play nice with the sampling frequency of the analyzer. Highly variable and inconsistent readings.

 

[Blythenti]

Looking good, it'll work like a charm.

 

[MattKing]

Whether they optimized for (keep constant) the shadow density, some midtone or highlights, I don't know. It's probably in the user manual of the filters.

They are speed matched on a lower highlight tone. This works well for matching Caucasian skin tones.

 

[koraks]

Thanks; yes, that makes good sense.

 

[albada]

Matt's post made me look again at Ilford's RC curves:

Close examination of these curves reveals the following paradox. When comparing grade 5 to grade 00:

1. To match Caucasian skin-tones, Blue must be increased to around 2X.
2. To reach D-Max, Green must be increased to about 6X. (Wow!)

You must boost blue in one case and green in the other. So karaks was right: When "matching" density, it makes a big difference which density you want to match. People are prominent in many of my photos, so should match densities Ilford's way.

I'm leaning toward *not* using PWM on the first go-around, as I realized that I can change density without changing contrast by marking the scales of the G-B pots with stops as well as grades. Thus, my circuit will be almost identical to Mal Paso's. Also, I think I'll buy both linear and log pots, and see which works out best.

Mark Overton

 

[MattKing]

It is important to understand that the way that multigrade papers work is that they use two or three different colour sensitive emulsion components, with each component having essentially the same shape of characteristic curve, but differing as to sensitivity (speed). The way that we adjust contrast is by varying how the densities add up, with the blue sensitive (high contrast) emulsion component being the highest speed, the green sensitive (low contrast) component being the lowest speed, and the cyan sensitive component (where one is used) having a speed somewhere between the two.
If you want to do an interesting experiment, make a good contrast, full range print of a normal full range subject using split grade printing techniques - one exposure using a 00 filter for the low contrast exposure and one exposure using a 5 filter for the high contrast exposure.
If the result is something like 12 seconds with the 00 filter and 4 seconds with the 5 filter - a not unusual result - next make two prints from the same negative without changing any setting, with one print being just the 12 seconds with the 00 filter (no 5 filter exposure), and the other print being just 4 seconds with the 5 filter (no 00 filter exposure).
You will most likely observe that the print using the 00 filter only is a bit flat and grey, but otherwise relatively close to the initial good full range print, while the print using the 5 filter only offers a relatively small amount of relatively faint detail.
Yet when the two exposures are combined, the result is a full range print with good contrast.

 

[albada]

The deed is done! I copied Mal Paso's design, and made my own LED-head like his.

I'm using five Cree LEDs of each color (Red, Green, Royal blue), connected in series, arranged around the perimeter of the circular barrel for a condenser head (bought from http://www.glennview.com/index.htm) that I'm adapting to use LEDs. Each LED is attached with two screws.

Everything is mounted to an L-shaped piece of sheet aluminum that is 1.5 mm thick.

Don't laugh at the picture below. I attached a pasteboard to the front to serve as a temporary control panel until I decide how to make an attractive one. I plan to put the control circuitry in the top instead of having a separate controller.

There are three knobs and three switches, one to control each color. I marked each knob with the number of stops of light-reduction from full power, which is why the scales are backward (0 on the right is full power).
Calibrating the stop-marks on dimming dials was tricky. I tried to use my DSLR as a meter, but its numbers were so jumpy as to be useless because the PWM dimming confused its meter. I finally used two old Selenium meters in tandem to check each other for agreement.

Red is using the dimming feature in the BuckBlock LED-driver. But green and blue use a PWM circuit I made based on the 555 timer chip. Here it is:

Dimming blue using the BuckBlock has two problems: (1) It dims to about 3.5 stops, and that's not enough to reach grade 0, and (2) the dimming scale is affected by the temperature of the LEDs. That second problem is what made me switch to PWM. When the blue LEDs get hot, their voltage-drop changes, and that changes how much dimming you get for a given resistance across the two DIM wires, making it difficult to repeat results when printing. I verified that PWM does not have this problem.
The large flat white object in the picture above is a solderless breadboard: You just push parts and wires into it, so you don't need to solder them. Such a board is intended for prototyping, and I'll probably build the permanent circuits on a perforated board. When I build the final perf board, I'll add a PWM circuit for red as well. Note that the regulator has a heat-sink on it. Also, you only need one regulator to power all three PWM circuits. The regulator is needed because the power supply provides 24 volts needed by the LEDs, but the 555 is only rated for 16 volts max.

The circuit schematic I used has a couple of changes made to accommodate parts I had on hand. The 20K potentiometer was supposed to be 100K, but I had 20K pots on hand, so that's what I used. The .3 capacitor was supposed to be .1, but I tripled it (using three .1s in parallel) in order to reduce PWM frequency, because the 20K pots multiplied the frequency by about 5x. An oscilloscope shows that I'm at around 300 Hz, which is working well with the BuckBlocks (their data-sheet recommends 200 Hz). The 53.6K resistor should have been about 24K, but I only had 53.6Ks lying around, so that's what I used, and the transistor is switching fine anyway.

I suggest using potentiometers with a logarithmic taper instead of the usual linear taper. Linear taper would have crowded most of the stop-marks near the left end of the scale; the logarithmic taper spreads them around more uniformly (look three photos above), allowing you to dial your dim-levels more accurately. And accuracy is needed for the low (soft) grades because they need little blue light.

My plan is to run test-strips of Stouffer wedges to find green-blue levels that match the Ilford VC filters. That info, using the stop-marks on the dials, will give me the ability to adjust contrast and brightness independently of each other.

Mark Overton

 

[radiant]

I've been thinking of led head too. I bought two 1 meter strips of WS2812 (144 leds / meter). I was thinking of wrapping it around in my enlarger "light chamber" (Fujimoto G7). One meter should draw 43W of power so I think one strip could be enough?

WS2812 has the beauty that you don't need any external power limiting/adjusting, all can be done with microcontroller.

 

[albada]

I've been thinking of led head too. I bought two 1 meter strips of WS2812 (144 leds / meter). I was thinking of wrapping it around in my enlarger "light chamber" (Fujimoto G7). One meter should draw 43W of power so I think one strip could be enough?
WS2812 has the beauty that you don't need any external power limiting/adjusting, all can be done with microcontroller.

The spec's for this LED-strip look good. Its RGB wavelengths are 650-520-460 nm, versus 625-530-450 for my Cree LEDs. When all my Cree LEDs are on full power, they use 41 watts, which is close to your 43 watts. But I think this is too bright, and I'll probably run mine mostly at half-power so exposure-times won't be under 10 seconds.

You will need to choose how to enter times or PWM-values into the Arduino. If you use knobs and potentiometers, you will need to calibrate them. A decimal keypad allows you to enter numbers directly, but it's harder than knobs to change brightness-levels. An app on your computer or phone would work, but they emit light, so you would need to cover them or turn them off when exposing paper. The brightness of LEDs in the WS2812 can be set to 256 levels, and you will need to run test-strips (or use a light meter) to see what brightness-levels you actually get. For accurate work, I suggest buying a Stouffer strip.

Uniformity is another problem. It is difficult to illuminate uniformly, and that's what makes Mal Paso's discovery valuable. You might need to try different arrangements of the strips to achieve uniform illumination.

Good luck,

Mark Overton

 

[radiant]

The spec's for this LED-strip look good. Its RGB wavelengths are 650-520-460 nm, versus 625-530-450 for my Cree LEDs. When all my Cree LEDs are on full power, they use 41 watts, which is close to your 43 watts. But I think this is too bright, and I'll probably run mine mostly at half-power so exposure-times won't be under 10 seconds.

Thanks for checking the wavelengths! And good to know I'm probably in the range of good power. I don't know if it is the condensing type of enlargers or my enlargers feature what but I have not ever experienced that my times would be too short (well if I'm not blasting from really close or/and using full aperture). So I'm still pretty skeptic if that is enough

ou will need to choose how to enter times or PWM-values into the Arduino. If you use knobs and potentiometers, you will need to calibrate them. A decimal keypad allows you to enter numbers directly, but it's harder than knobs to change brightness-levels. An app on your computer or phone would work, but they emit light, so you would need to cover them or turn them off when exposing paper. The brightness of LEDs in the WS2812 can be set to 256 levels, and you will need to run test-strips (or use a light meter) to see what brightness-levels you actually get. For accurate work, I suggest buying a Stouffer strip.

I have built myself a quite featureful enlarger timer with all kind of features such as automatic test strip exposing and stuff for multigrade/splitgrade printing. So it is pretty easy to implement the light control to that. And yes I'm actually using a full size keyboard to control my timer Well it is not a timer, it's a exposure computer?

I'm using 00 and 5 filters only currently so just plain green & blue light is enough for me to continue my workflow. I'm also looking for the red light to implement the pre-exposure red doge/burn-tool adjustment time-window etc. But I think I finally get stouffer wedge anyways, would be interesting to make comparisons at least!

Thanks Mark for the support!

 

[radiant]

.. and about illuminating uniformly.

My enlarger (Fujimoto G70's) system of projecting light is quite weird. The lamp is blasting light from a really tiny (1" x "1" or smaller) hole to this "condensing chamber". The light just goes straight to the wall of chamber and then bounces around until finally goes through the condensing lens. I don't understand why it is built like that, feels pretty stupid. See illustration below.

My plan is to circulate the chambers walls with the LED strip so no LED is directed to the lens at all. So it should maybe be similar system that the current light path is?

If you have any knowledge on how condeser systems or even how G70 works (and why ..) please let me know I know condenser makes even illumination for sure.

I was also thinking of LED "plate" on top of the chamber but I guess then I could run into non-even illumination problems more easily.

 

[David Brown]

.. and about illuminating uniformly.

My enlarger (Fujimoto G70's) system of projecting light is quite weird. The lamp is blasting light from a really tiny (1" x "1" or smaller) hole to this "condensing chamber". The light just goes straight to the wall of chamber and then bounces around until finally goes through the condensing lens. I don't understand why it is built like that, feels pretty stupid. ...

I know condenser makes even illumination for sure.

Even illumination comes easier from diffusion. Condensers focus light; they are lenses, after all. What you have in your Fuji is a very common way that color heads achieved diffusion. Many color heads used such an arrangement. Phillips enlargers had this 90 degree arrangement of bulb to "diffusion chamber". (Not condensing chamber) Many Durst are like this. Common color heads for both Beseler and Omega 4x5 enlargers have the light enter a diffusion chamber from the side. It's quite normal, even for some black and white heads.

 

[MattKing]

Current LPL enlargers use the same system.

 

[gijsbert]

I think the Durst M805 (with a similar condenser head) manual mentions that this 90 degree mirror light construction keeps heat from the lamp away from the negative stage. It also makes the head less high so one needs less overhead space.

If I remember correctly the G70 has a condenser on one side and diffuser on the other, you can pull it out and reverse it to switch between condenser and diffuse enlarger. I had a G70 but I couldn't find any of the negative carriers so I switched to a M805 when one popped up in my area.