Is it possible to make a DIY LED splitgrade system?

Hello,

I had a chance to briefly see the Heiland Splitgrade system in action and I would really like it for some applications. But the price of the solution and the light source is out of my reach right now.
But I wonder how difficult it would be to put something like that together? I'm assuming it wouldn't be that fancy and comfortable.

It doesn't even have to be a fully automatic system. Let's say I have a basic electronic and programming skills and some of the following equipment:

Darkroom Automation Meter /RH Designs ZoneMaster/ another meter for measuring light levels.
Reflection densitometer for measuring density on paper.
Light source with suitable blue and green LEDs (maybe red for easel work) with PWM dimming.
Microcontroller (ESP32, Arduino MEGA) for light control and exposure management.

So i have the ability to measure/calibrate the paper and control exposure to blue and green light.

How to find out the parameters for splitgrade? Does anyone have an idea?
Thank you very much.

Although "Splitgrade" doesn't have a formal definition, the technique traditionally revolves around a system of making two test strips, one with all green light and one with all blue light, to determine exposure and paper grade/VC filtration. You then make the final print with two exposures to blue and green light.

Using blue & green LEDs with PWM intensity control isn't splitgrade - it is an LED variable contrast head. You would have a dial or Arduino setting to control the ratio of blue & green light intensities to give a reasonable range of paper grades. Of course nothing precludes using an LED head to do traditional splitgrade printing.

You might find the Darkroom Automation application note on metered splitgrade printing interesting http://www.darkroomautomation.com/support/appnotesgmeasured.pdf - along with an application note on how VC papers work http://www.darkroomautomation.com/support/appnotevcworkings.pdf

Forget about official "grades" as they are just fairly arbitrary ranges of ISO-R. To "calibrate" your grades, all you need is a Stouffer step tablet and some paper. You can define the grades as a combination of blue and green exposure within your specific implementation of a light source.

W.r.t. the light source, for b&w multigrade you can use whatever arrangement of blue and green LEDs you like. These can be strips, stars, individual (smd) LEDs on a PCB, COB etc. Whatever you choose, you'll have to make a light path that results in even exposure of the negative. Duffusion-based systems are easy to make; just experiment a bit with LED arrangement and e.g. milky plexiglass.

As to the controller, use any old Arduino etc of your choice with and LCD hooked up to it and buttons, rotary encoders and keypads as you see fit. Making a simple timer with two PWM outputs isn't very complicated.

This is a pretty straightforward project that many have undertaken; it's definitely feasible, just get cracking and have fun!

Nicholas Lindan said:

Although "Splitgrade" doesn't have a formal definition, the technique traditionally revolves around a system of making two test strips, one with all green light and one with all blue light, to determine exposure and paper grade/VC filtration. You then make the final print with two exposures to blue and green light.

Using blue & green LEDs with PWM intensity control isn't splitgrade - it is an LED variable contrast head. You would have a dial or Arduino setting to control the ratio of blue & green light intensities to give a reasonable range of paper grades. Of course nothing precludes using an LED head to do traditional splitgrade printing.

You might find the Darkroom Automation application note on metered splitgrade printing interesting http://www.darkroomautomation.com/support/appnotesgmeasured.pdf - along with an application note on how VC papers work http://www.darkroomautomation.com/support/appnotevcworkings.pdf

Click to expand...

But I really meant splitgrade. From the Heiland site: Exposure in 2 phases with soft and hard light allows dodging in either the hard or the soft gradation phase.

That's what I'd like to achieve. Measure the scene and get separate values for blue and green exposure that will lead me to a solid work print.

koraks said:

Forget about official "grades" as they are just fairly arbitrary ranges of ISO-R. To "calibrate" your grades, all you need is a Stouffer step tablet and some paper. You can define the grades as a combination of blue and green exposure within your specific implementation of a light source.

W.r.t. the light source, for b&w multigrade you can use whatever arrangement of blue and green LEDs you like. These can be strips, stars, individual (smd) LEDs on a PCB, COB etc. Whatever you choose, you'll have to make a light path that results in even exposure of the negative. Duffusion-based systems are easy to make; just experiment a bit with LED arrangement and e.g. milky plexiglass.

As to the controller, use any old Arduino etc of your choice with and LCD hooked up to it and buttons, rotary encoders and keypads as you see fit. Making a simple timer with two PWM outputs isn't very complicated.

This is a pretty straightforward project that many have undertaken; it's definitely feasible, just get cracking and have fun!

Click to expand...

I've been browsing through Photrio and seen various tutorials (including yours), which are great.

But no one has addressed splitgrade, i.e. divided exposure and after calibration without the need to make test prints.

Silvergrain Classics have an article on the widget you are referencing. https://silvergrainclassics.com/en/2021/09/what-is-the-heiland-splitgrade-controller/

It feels a bit like it's on the wrong side of technological diminishing returns. If you can calibrate you print setup (so know your filter factors for your grades), you can easily determine a standard grade for a single grade print with a single test strip (DistPhoto on yourtube has some videos discussing this with is Perfect Test Strip printer, but his widget isn't needed).

For multigrade you don't have to stick with just 00 and 5, you cam mix whatever you need with dodging and burning. But if you need to split grade, you're already way beyond the point where you are going to get away with one test strip, regardless of how much wizadry is in the box.

It just doesn't feel like the computer is winning you much. It's also seems to suggest you can do away with testing because it has some paper and developer know how, but realistically you're have to test because there are just too many other variables (your water, your dev temperatures).

kuhyraco said:

But no one has addressed splitgrade, i.e. divided exposure and after calibration without the need to make test prints.

Click to expand...

The whole point of split grade is to be able to burn & dodge during either or both of the separate exposures. That's almost by definition an empirical, intuitive process based on trial & error. If you want to do measurements on the negative to determine the single ISO-R grade that will encompass these measurements within the tonal scale on a single print, then that's not necessarily a split grade endeavor; it's basically just a multigrade/variable contrast approach where you replace a series of test strips by a measurement combined with calibration. Whether you then make the final exposure as a single exposure with both blue & green at the same time, or consecutive ones, really doesn't matter from the viewpoint of the paper. It does matter of course if you're going to burn & dodge, but then the question is what the added value of the initial measurement and the whole calibration circus would have been.

tcolgate said:

It just doesn't feel like the computer is winning you much.

Click to expand...

Indeed, I agree.

I do not want to step into the discussion of pro/con split grade and how it works - others are more experienced in that.
The only reason that Mr. Heiland used the split grade system was technical. At the beginning of his system, no usable LEDs were available. For the traditional enlarger head modifications that he partially still provides, it was much more easier to implement 2 filters that switches than 14 for half grades. And also calibration is not as complicate.

I think you might be undeestimating the software involved in ÍSplit/Grade system but you shouldbe able to put a VC LED head together. The cost of the Heiland has its justification. It includes years of R&D.

kuhyraco said:

Microcontroller (ESP32, Arduino MEGA) for light control and exposure management.

Click to expand...

Having recently just implemented a darkroom timer on an atmega328p, given the choice, I'd opt for something slightly chunkier in terms of ram and flash support. Just about anything really bigger than an arduino really.
If I wanted to keep things tiny and cool I'd give the BBC microbit a go. I think it has everything I'd need and just a bit more than the atmega has to give headroom for more features without all the hassle of a Pi.
I think you can do a lot on an arduino if you want to stick with C/C++ (I'm spending my spare time writing C again though), but there are some things I'd like to do that start building up a bit. Like I'd quite like to be able to keep print profiles on an SSD that I could quickly switch in and out. Drivers and filesystem for that are going to eat into 32kb pretty quick. Honestly though, the idea of doing anything in C nowadays feels more than I can be bothered with for a hobby project nowadays.

tcolgate said:

Just about anything really bigger than an arduino really

Click to expand...

You're referring specifically to the 'old-school' Arduinos like the Nano, Uno (first version) and Pro Micro. Note that there are many Arduino boards that have far more powerful controllers with more flash and RAM. 'Arduino' is really just a brand name and the product range is very extensive.

tcolgate said:

the atmega

Click to expand...

Likewise, there are ATMega controllers with more flash and RAM than the '328P/PB.

tcolgate said:

the idea of doing anything in C nowadays feels more than I can be bothered with for a hobby project nowadays.

Click to expand...

AI is getting pretty good at writing embedded C++.
If you want higher-level language support, it's indeed a good idea to scale up the hardware.

I've run a color enlarger controller based on an Arduino Nano for a few years; it ran 3 color channels, timer programs for b&w and color mode and read out several encoders. This ran easily within the memory constraints of a '328P. However, given the negligible cost of far more powerful hardware (e.g. ESP32 platforms), I agree it's unnecessarily limiting to stick with the very old 328P/PB platform.

koraks said:

[adrduino etc]

Click to expand...

absolutely true, I've not looked at the alternative boards. I saw Arduino recently put out a non-avr board too with a higher spec. Most of my experience (and gear) in this area is 15 years old.

koraks said:

AI is getting pretty good at writing embedded C++.
If you want higher-level language support, it's indeed a good idea to scale up the hardware.

Click to expand...

reading regurgitated stolen C++ isn't on my TODO list of an evening either. I've been writing Go for so long now everything else feels like hard work.

koraks said:

I've run a color enlarger controller based on an Arduino Nano for a few years; it ran 3 color channels, timer programs for b&w and color mode and read out several encoders. This ran easily within the memory constraints of a '328P. However, given the negligible cost of far more powerful hardware (e.g. ESP32 platforms), I agree it's unnecessarily limiting to stick with the very old 328P/PB platform.

Click to expand...

I managed to get multi-exposure and f-stop support (with test strips), into about 24kb of flash, but got dangerously close to the 2kb RAM (as in about 2032 bytes). Improving the UX and adding a G+B graded exposure option wouldn't need much more, but enough that I need to start bigger refactors. I don't think I'd get an sd/mmc driver in there either. I /think/ I could do it on a microbit, and it has enough GPIOs, but I've not had a try (need to get on and actually do some printing). The microbit actually has bluetooth support too, giving hte option of moving print profiles on with a phone app... but mobile development is definitely not on my "doing that for a hobby" list either.
I do think a Pi is probably overkill, and then you have to faff about maintaining the OS, but for less eperienced people maybe python etc are an easy route in.

tcolgate said:

regurgitated stolen C++

Click to expand...

It's really not that, but YMMV and to each their own. I've written a decent amount of C++ over the past few years and have recently begun incorporating AI in these projects. Most software engineers today have embraced AI in some shape or form because it just speeds things up massively. The quality of the code may not be excellent right off the bat, but if you know what you're doing, you can accelerate things dramatically using AI. Previously I would spend hours writing a simple ISR using the reference manual for some chip; now I just ask AI to come up with a skeleton demonstrator for the platform involved (while feeding it the reference manual or parts thereof) and then also use it for troubleshooting. The net result is that I can get stuff done within an hour that might have taken me several days to figure out before.

tcolgate said:

Go

Click to expand...

Yeah, that's pretty high-level. May work OK for this sort of thing, but for many things that are closer to the hardware layer, the net benefit is marginal. Again, to each their own and there are plenty of ways to skin that cat.

tcolgate said:

I do think a Pi is probably overkill

Click to expand...

I think given the cost of hardware these days I wouldn't consider it overkill necessarily. Sure, the same thing can be crammed into tiny little STM8F003 or $0.25 CH32v003, but if you look at the bigger picture, you need to factor in the time spent on making stuff work. If you can get there with two days' work with a $75 Pi while it would take a week of cursing on a bare metal approach, it's clear that the Pi wins. Assuming it has no soul, it won't mind being bored 99.995% of the time, hah!

Another option would be waiting for the Dektronics Printalyzer from Derek Konigsberg that will have a DMX512-compatible expansion port. Should be possible to connect it to a custom LED head. If I remember right from his last video, he also played with an Intrepid Enlarger Kit interface hack. There is no pricing available yet, but should be less expensive then a Heiland LED cold light head + Splitgrade Controller + LED controller unit. And you safe the effort for an Arduino or Raspi or whatever and the programming...
I am thinking about selling all my Heiland and Analyser zoo if Derek's solution keeps his promises.

Pegajoy said:

Another option would be waiting for the Dektronics Printalyzer

Click to expand...

The Enlarging Timer, specifically.
And yes, assuming that the device will be capable of calibrating itself to render different paper grades using the probe and data from the paper manufacturer(s). AFAIK that's not really clear as of yet although undoubtedly @dkonigs has been working on this. Perhaps he can comment on the status and intended direction in this regard.

Thanks for the interesting debate.

Maybe I should have written at the beginning why I would like something like this.
More and more often I get my hands on other people's negatives, which I print in larger quantities. Different negatives, different sized prints. Which do not need to be top quality, as if I were doing something of my own and dedicated to it. So I need to get to a decent work print as quickly as possible and with the least effort, and for better negatives straight to the final print.

I had the feeling that with the Heiland solution, you can measure the negative and without a single test print you will get exactly what I wrote above. With other workflows, you have to think about the negative, where you want to have what tone, etc.

Nicholas Lindan said:

You might find the Darkroom Automation application note on metered splitgrade printing interesting http://www.darkroomautomation.com/support/appnotesgmeasured.pdf - along with an application note on how VC papers work http://www.darkroomautomation.com/support/appnotevcworkings.pdf

Click to expand...

Thank you for the links full of knowledge!

Gotcha; in that case forget about 'split grade' as such; what you're doing is basically a single multigrade exposure using blue & green (which can be a concurrent or a staged exposure; it doesn't really matter), based on a measurement of negative contrast. Yes, this is feasible; the most tricky bit is to make or interface with a measurement system.

The most low-tech way of doing this is to take an existing baseboard meter such as the one you already have and translate the readings to grades using the ISO-R definitions, then dial in that grade with regular Multigrade filters or a dichroic head. You don't have to build any additional hardware for this. I've done this for a time; the approach as such works and is quite fast, and will give you decent prints especially if you standardize on one or two papers.

This is exactly why I worked on my f-stop test strip setup. There's a rule of thumb that says that 1/3rd stop expisure difference between correct highlight and shadow is about a 1/2 grade on the Ilford filter.
I use a test strip printer to expose an area of the neg that includes some highlight and shadow I want retained. The strip uses the same area but at 1/3 stop intervals. If I get good highlights on one patch, and good shadow detail on another, I can use the number of patches between as a guide to adjust the grade, adjusting the exposure to compensate for the filter.
I've used this approach a few times now, I'm no expert, but it gets me close fast.

I guess its time that I joined the thread

For the timer itself, I've been keeping this thread around:
(Its been going on-and-off for a long time now.)

As far as status, my current goal is to get it over the hump of safety and EMC testing. Right now I'm delayed on one final hardware update outside my control (enclosure vendor taking a while on some final changes), but once that's done I'll immediately move forward.

Once I'm past those hurdles I'll build a small number of pre-production units to send out. Initially I'd like to work with a few technical test users to get all of these specific use cases ironed out, then over time I'll expand that to a few more typical users. Hopefully I'll get to actual production sometime later next year.

Pegajoy said:

Another option would be waiting for the Dektronics Printalyzer from Derek Konigsberg that will have a DMX512-compatible expansion port. Should be possible to connect it to a custom LED head. If I remember right from his last video, he also played with an Intrepid Enlarger Kit interface hack. There is no pricing available yet, but should be less expensive then a Heiland LED cold light head + Splitgrade Controller + LED controller unit. And you safe the effort for an Arduino or Raspi or whatever and the programming...
I am thinking about selling all my Heiland and Analyser zoo if Derek's solution keeps his promises.

Click to expand...

So yes, there's a DMX512-compatible expansion port. I expect this to be an area of on-going development, as I work through all the cases of contrast control configuration. But it already allows some pretty nifty implementations with the right hardware connected to it. Not only can it be used to control the enlarger, it can also be used to control the safelighting. (in addition to the traditional switched outlets, of course)

One thing I absolutely do want to provide is a "reference implementation" for as many existing products as I can find. Right now that's at least the Intrepid LED head and the Heiland LED head. (And it just occurred to me that the Ilford 500 MG Head might also be worth exploring. No idea how many are out there, but they are easy to find on eBay.)

The demo I cobbled together in that video was a little bit of a hack, but only on the software side, and then mostly due to Arduino library limitations. The issue was really juggling timing between the DMX receiver code and the code that controls the Intrepid head (which is basically a big Neopixel array). Basically they kinda stomp on each other. That is fixable, with some work on the libraries involved.

The Heiland is a bit more "old school" in how its controlled, as you'd need to send PWM signals to it (albeit at 12V, and over a really funky connector). I haven't tackled this yet, but I will. The real issue there is that I really want 16-bit PWM (better for logarithmic control) and I'd need a newer Arduino (e.g. R4) to get that. But the DMX libraries are all really old and unmaintained and don't run on the newer boards. So it'll just be a bit more work if I want to stick with an Arduino-based "reference adapter".

For more of your DIY LED heads, this problem actually gets a lot simpler. You can simply buy off-the-shelf LED drivers that speak DMX and support 16-bit control. One I've done some preliminary testing with is the LT-820-5A, as an example.

Just as a side-note, I don't personally use Arduino for anything. (The timer itself uses an STM32F446RE coded directly in C /w FreeRTOS.) Its just that it is the most convenient choice when you need to use "someone else's hardware" and expect other people to be able to pick it up and tinker.

koraks said:

The Enlarging Timer, specifically.
And yes, assuming that the device will be capable of calibrating itself to render different paper grades using the probe and data from the paper manufacturer(s). AFAIK that's not really clear as of yet although undoubtedly @dkonigs has been working on this. Perhaps he can comment on the status and intended direction in this regard.

Click to expand...

The current approach for calibration involves using a meter probe to measure enlarger light, a calibrated step wedge to make an exposure onto a piece of paper, and then a reflection densitometer to measure that paper once processed. (And yes, I plan to include such a device, though any should work fine.)

From this it should be possible to collect all the necessary data to determine paper Dmax, Base exposure (Dmin+0.04), Speed (Dmin+0.6), and Contrast range (ISO-R). In other words, everything to produce the characteristic curve in ISO 6846. And then this curve is used as the basis for everything else.

Unfortunately this does need to be done for each contrast grade, but there's really no obvious way around that.

Probably enough rambling here for my first reply.

dkonigs said:

)
One thing I absolutely do want to provide is a "reference implementation" for as many existing products as I can find. Right now that's at least the Intrepid LED head and the Heiland LED head.

Click to expand...

This all sounds brilluant! One thing to check with the Intrepid is that they changed to physical connector from their timer to their light some time last year. My old timer used a round plug, the modern one is sorted of triangular. They ship an adapter cable with the timeless 4x5 kit (and kindly sent me one when I asked).

kuhyraco said:

So I need to get to a decent work print as quickly as possible and with the least effort, and for better negatives straight to the final print.

I had the feeling that with the Heiland solution, you can measure the negative and without a single test print you will get exactly what I wrote above. With other workflows, you have to think about the negative, where you want to have what tone, etc.

Click to expand...

As has been discussed there are several alternatives but another solution is to get a Heiland splitgrade-controller and use it without their specialized lightsource or automated filter-module. Heiland calls this Manual Splitgrade and it's much more affordable.
The advantage compared to other analyzers is that data for available papers is already integrated and Heiland provides updates if necessary, and still support even their first device from 1998. Several friends of mine swear by the system, manual as well as full featured. For your application, providing a decent work-print as fast as possible, it may be the right solution.

I still prefer to work with test-strips and work my way to the final print with my own senses but on some days I suck at this and wish I had an analyzer again to make things easier.

skahde said:

As has been discussed there are several alternatives but another solution is to get a Heiland splitgrade-controller and use it without their specialized lightsource or automated filter-module. Heiland calls this Manual Splitgrade and it's much more affordable.
The advantage compared to other analyzers is that data for available papers is already integrated and Heiland provides updates if necessary, and still support even their first device from 1998. Several friends of mine swear by the system, manual as well as full featured. For your application, providing a decent work-print as fast as possible, it may be the right solution.

I still prefer to work with test-strips and work my way to the final print with my own senses but on some days I suck at this and wish I had an analyzer again to make things easier.

Click to expand...

You are totally right. I also use the Heiland controller with manual mode for one of my enlargers. Calibrating for any light source is quite easy, you need just 2 test strips for high and low contrast (and not 14 for all half steps like for the Analyzer).
If you prefer going the test strips way instead measuring, I would opt for the RH Designs Stopclock that have flexible test strip modes with dry compensation as well as 2 channels for splitgrade printing with automated time adjustment if changing the contrast grade.

tcolgate said:

This all sounds brilluant! One thing to check with the Intrepid is that they changed to physical connector from their timer to their light some time last year. My old timer used a round plug, the modern one is sorted of triangular. They ship an adapter cable with the timeless 4x5 kit (and kindly sent me one when I asked).

Click to expand...

Thanks for the heads-up on this.
The round cable they used to use is something sold as an "E-Bike Signal Connector" (or a Chogori Tech Micro-A 2-6 Pin), which is a little bit of an odd choice and took some effort to track down.
I'd definitely appreciate a lot more information on what connector they currently use. (even if you send it privately)

Coming up with the best way to attach the Intrepid connector to my adapter is something I've been thinking about. With the old cable, I was considering just providing screw terminals (or solder holes) where one could buy the cable off Amazon, slice it in half, and hook up the wires to the board. (similar to how its attached inside their timer). With the new cable, the same approach could be taken. Unless, of course, its something I can mount a real connector for (and then have adapters).

Regardless, I think my big questions are:

1. Did they actually change the design of their head, or just the cable?
2. What connector are they using now?
3. What is its pin-out?
4. Is it the same as the old one electrically? (I'm guessing it is, if they'll send you a simple adapter.)
5. How is the length of the cable split between their timer and their head, versus the old one?

skahde said:

I still prefer to work with test-strips and work my way to the final print with my own senses but on some days I suck at this and wish I had an analyzer again to make things easier.

Click to expand...

I often still work with test strips as well, though having an analyzer to get your staring exposure is a nice thing.

My main gripe with the old (80's/90's) style devices like RH, is that its very easy to lose track of what's going on in their user interface. So even if you are doing test strips, a more modern display makes it easier to see what is going on. (e.g. graphical image of a test strip, highlighting the current patch, and a more obvious way to change test strip parameters)

dkonigs said:

Regardless, I think my big questions are:

1. Did they actually change the design of their head, or just the cable?

Click to expand...

The head is, as far as I know, unchanged. I've run my firmware on my original Compact, and on the newer 4x5 head, and it works on both. It's the same LED string, and same number of LEDs in the 4x5 and compact.

dkonigs said:

1. What connector are they using now?

Click to expand...

looks like it might be one of these https://www.mutuallc.com/MINI-J-3-pin-signal-connector-pd45173085.html

dkonigs said:

1. What is its pin-out?

Click to expand...

I'm not sure. It's only a two wire setup, I'd need to open up my spare timer and poke around (I'm not going to have time for that for a few weeks).

dkonigs said:

1. Is it the same as the old one electrically? (I'm guessing it is, if they'll send you a simple adapter.)

Click to expand...

Yeah, the cable they send is just a straight conversion.

dkonigs said:

1. How is the length of the cable split between their timer and their head, versus the old one?

Click to expand...

About the same. I can't be 100% sure because I got my first timer with an old Compact, and the second with a new 4x5, so they could have shifted stuff around, but the two timers are essentially the same barring this connector. I assume they had issues with someone forcing it in the wrong way (hard to do, but as they say, "there's a story behind every warning sign").