ILFORD Multigrade and E27 Off-The-Shelf (OTS) LED Bulbs

[Alexo]

I've started to experiment replacing the PH211 Opal with OTS LED E27's, and I will post results on this same thread including actual pictures developed with each one and other details.

To start I grabbed some commercial Off The Shelf bulbs from Carrefour and fetched their technical specs from the ENERG Website. I will later grab other E27 one like 270 degree one with plastic bases from cheap dollar-store-style brands and compare results.

Light Sensitivity of ILFORD MG

This is the baseline sensitivity curve for ILFORD MG Papers:

Source: https://www.ilfordphoto.com/wp/wp-content/uploads/2021/01/MULTIGRADE-IV-RC-Papers-060619.pdf


Light Curves of Typical OTS E27 Light Bulbs

Here are the baseline Carrefour DY96xxx series bulbs which are full glass 360 deg E27

Carrefour 6500K
Carrefour DY96190 75W, 6500K, 1055 lm, 7.8 W, 80 CRI, Flicker 1, Strobe 0.4, EI=D
Carrefour DY96202 60W, 6500K, 806 lm, 5.9 W, 80 CRI, Flicker 1, Strobe 0.4, EI=D

Carrefour 4000K
Carrefour DY96192 75W, 4000K, 1055 lm, 7.8 W, 80 CRI, Flicker 1, Strobe 0.4, EI=D
Carrefour DY96204 60W, 4000K, 806 lm, 5.9 W, 80 CRI, Flicker 1, Strobe 0.4, EI=D

Carrefour 2700K
Carrefour DY96194 75W, 2700K, 1055 lm, 7.8W W, 80 CRI, Flicker 1, Strobe 0.4, EI=D
Carrefour DY96206 60W, 2700K, 806 lm, 5.9 W, 80 CRI, Flicker 1, Strobe 0.4, EI=D

Source: https://eprel.ec.europa.eu/screen/product/lightsources
Terms: lm = lumens, W = Watts, CRI = R9 Color Rendering Index, EI = ENERG Rating

Curve Overlaps

Here I tried to overlap the curves on somewhat the same scale to get a rough idea of the best matching bulbs.
Keep in mind that these curves may not be on the same Y scale but still gives a visual representation of the match.

Overlap with 6500K

Overlap with 4000K

Overlap with 2700K

Conclusions and Next Steps
It seems obvious that 4000K matches almost exactly with the sensitivity curves of ILFORD MG Papers so this will be the baseline to compare to. I will also include some experiments to compare with an incandescent opal bulb as well, all in different contrast ranges.

In the next few days/weeks I will do some actual development and document development times f stops, nd and contrast filters, to get a good idea of how well these cheap E27 LED bulbs compare to incandescent opals. I will be using a Durst M300.

 

[Alexo]

Base Image from Incandescent 60W Opal Bulb

The image below is the baseline image, using a commercial off the shelf incandescent opal 60W light bulb. I think most opal bulbs are rated similar to the PH211 which is around 3000K. This picture was burned with the 2.5 MG filter at 12 seconds plus an ND4 filter at f8. This one is framed at 10x15cm standard ILFORD MGRC Deluxe Pearl. The experiment pictures below are framed at half sheets (10x7x5cm) to save paper.

Be sure to expand the images and examine closely !!

Results from the first batch

For these tests I started with the 75W equivalent Carrefour LED Bulbs specified above. Considering the wavelength overlaps, I decided to start with 4000K as a baseline and filters 0, 2.5 and 5 (doubling the time for 5). For all the pictures below I used the same 4 seconds time burn at f8 with an ND4 filter in the drawer as well (both the ILFORD MG filter and the ND filter are before the lens, inside the Durst M300 filter compartment).

4000K Result

The 4000K LED seemed to work pretty well. Maybe I should have dialled back the burn to 3 seconds to get it closer to the baseline image, but the results, overall seem tu suggest that the contrast levels work as expected. I might do another saline test with the incandescent bulb but the burn times are much slower and I did not want to change any variables. In other words for the rest of the experiments I maintained the same burn time, f stop and filter, except for the last experiment at 2700K where I had to halve the ND to get a decent image.

6000K Result

The 6000K is very interesting. Definitively higher contrast at #5 but a lighter image at #2.5. I will research the filter curves of the ILFORD MG and publish them here to understand what happened here. But overall, seems 6000K works just as well as 4000K but with higher contrast in the higher filters and what seems to be either less contrast (or brightness, not sure) in the lower MG filters.

2700K Result #1 (ND4)

At first this seemed like a total fail but looking closer I noticed something interesting in the curved ceilings on the right side and on the pine tree (the #5 filter side). So I decided to halve the ND filter to a #2 and got the second result.

2700K Result #2 (ND2)

At first this seemed like a total fail but looking closer I noticed something interesting in the curved ceilings on the right side and on the pine tree (the #5 filter side). So I decided to halve the ND filter to a #2 and got the second result. This result is honestly very surprising and interesting. The 2700K seems to respond well to the higher contrast filters but reveal more information than the previous bulbs.

Conclusions of Experiment #1

The results of the 2700K build were completely unexpected for me. I'd love to hear some comments on what you observe but I think this opens up a whole new of possibilities. For example I have a beach photo which I haven't been able to get quite right and I wonder of playing around with the 2700K could help me tackle that one.

Anyway, I will continue to play around with different LED bulbs, and share the results here. It's very exciting to see that LED light sources could provide a completely new tool set at our disposal and the results are really encouraging.

Hoping to hear your comments !!

--
Alex

 

[albada]

Alex, first of all, welcome to Photrio!

Papers and contrast filters were designed for the light-spectrum emitted by tungsten bulbs. To me, the spectrum of your 2700K looks closest to tungsten (much less blue than green), so it will probably best match tungsten.

Another thing to be aware of is that an enlarger's condenser system projects an image of the bulb's surface onto the paper. Thus, it's prudent to check uniformity of illumination. Here's a way to check uniformity: With no negative in the enlarger, you can expose a sheet of paper to middle gray using the #5 (highest contrast) filter. If it's decently uniform, I suggest buying a few more of those bulbs as spares, because you found a good match.

Mark

 

[koraks]

Yes, welcome to Photrio, and thanks for taking steps to figure out this relevant issue. Many people are interested in replacing their tungsten enlarger bulbs with LED devices. Testing like yours can help to shed light onto the matter.

Here I tried to overlap the curves on somewhat the same scale to get a rough idea of the best matching bulbs.

To be frank, this isn't as relevant as it may seem. The overlap between the spectrum of the bulb and the sensitivity of the paper may tell you something about the speed matching of a set of filters. For instance, the Ilford filters are speed-matched across a couple of contrast grades, but this assumes a typical halogen light source. If you install a LED bulb that emits a relatively large amount of blue light (i.e. higher color temperature), then this may/will break the speed matching of the filter set. This is not a problem per se if you're OK with doing a new set of test strips when switching to a different paper grade. Some may find this annoying, however.

What the plot analysis will not give insight into, is the contrast behavior of the filters. Multigrade/VC filters essentially pass through a certain amount of blue and/or green light. In terms of a light bulb, what's relevant mostly is whether it emits a usable amount of light around the peak wavelengths of these filters. Hence, a more meaningful comparison would be between the bulb emission spectrum and the multigrade filter transmission spectrum. The latter can be found by googling around a bit; this is what I managed to pull up:

This suggests that the multigrade filters have notches ar around 420nm (royal blue) and at 520nm (green) + 560nm (green-lime).

As you can see in the LED bulb spectra, they're all virtually dark at around 420nm, but fortunately, the MG filters are rather unspecific and there's a large slope between 420nm and ca. 460nm where the LEDs will have decent output. The fact that this doesn't result in much of a problem is also because tungsten/halogen light sources are horribly inefficient in the blue region as well. This is taken into account in designing the high-speed paper emulsion. See e.g. here:

A typical enlarger bulb will be a halogen type and as you can see, they behave also don't have much oomph in the sub-450nm region.

The green peaks are not much of a concern either way, because these are nicely within the high-output part of both the tungsten and white LED curves. There's always plenty of output to work with, there.

A final area of concern is the bit around 490-500nm, which is used apparently in MG filters to differentiate between high and very high contrast; see the filter curves for grades 4, 4.5 and 5.

So taken together, the discussion IMO should revolve around the relationship between bulb spectrum and filter spectrum, not bulb vs. paper sensitivity. If paper sensitivity spectrum is taken into account, it should be differentiated by the different emulsion components; i.e. the high- (blue), medium (cyan) and low(green)-speed emulsions.

 

[Alexo]

ERRATA on previous posts: Everywhere 6000K is mentioned, it is actually 6500K

To get a better understanding on what is going on with these experiments, here is a few more charts as reference. Multigrade papers work by filtering the amount of green and blue light that hits the paper during burning. The bluer the light, the more contrast and the greener the light, the less. The detailed explanation is here: https://www.ilfordphoto.com/wp/wp-content/uploads/2017/03/Contrast-control-for-Ilford-Multigrade.pdf

The ILFORD Multigrade Filters use different shades of yellow and magenta to vary the proportion of blue and green light that hits the paper. A magenta filter absorbs green light and transmits blue; a yellow filter absorbs blue light and transmits green.

Per the initial post. Here is the ILFORD MG response curve:

And here is the visible light spectrum:

If we overlap one on the other we get this:

Overlap with 4000K we get this:

Overlap with 6500K we get this:

Overlap with 2700K we get this:

Now the 2700 results make more sense!

In my first overlap images it seemed that the 4000K and 6500K curves better matched the ILFORD sensitivity curves. But on closer inspection, and especially when mapped on the visible spectrum, you can clearly see the both 4000K and 6500K have an extreme peak at 450nm which is mostly blue light, which is dis-proportionate to the rest of the emitted spectrum. In retrospect, the 2700K is much more similar to the MG paper's curve and to tungsten in that range.

Verdict and Conclusions (so far)

I have read from several people the 4000K LEDs are the perfect replacement for the 3000K incandescent opals such as the PH211 (which use tungsten filaments). But based on everything I have learned from this experiment, it seems that 2700K is a better choice, as it avoids the 450nm peak. Because the blue-sensitive coatings burn much faster than the green ones on ILFORD MG papers, it seems that the 450nm peak is an undesirable trait of LED bulbs of 4000K and above (for normal, day-to-day printing).

YMMV and maybe not all brands have the same curves, I honestly don't know (yet). Maybe there exists commercial 3000K LED E27s that may work even better, but looking at the 2700K curves against the paper curves, it seems the 2700K is the best choice. The small peak in the 450nm is cancelled by the reduced sensitivity of the paper in that wavelength, and the rest of the curve is almost identical to the paper, so all in all, I think the 2700K is a better choice than 4000K. If I stumble upon a 3000K, or bulbs with radically different curves, I will repeat the experiment and follow up with the result.

This doesn't mean that the 4000K and 6500K are useless. On the contrary, I believe they can be used to push the filters and the paper in unexpected ways, maybe by combining yellow and magenta filters to correct some stubborn negatives, or to get some desired effect.

Please take all my graphs with a grain of salt because the scales are force-matched on GIMP and most surely inexact. But they do give a general idea of what is going on. If you find curves for other bulbs, please attach the to this thread.

Regarding lumens, I think the 75W equivalents (1055 lm) burn too fast for the Durst M300 with a 50mm lens at f8. Which forces the use of ND filters (4 stops or more), which depending on the quality may or may not affect the final contrast and/or overall quality of the print. Also, the power up of LEDs is slow (1/2 to 1s), so for 1-2 second burns, will give unpredictable results. The 60W (806 lm) are probably a much better choice, giving you more burn time with less ND, which IMHO is always better.


--
Alex

 

[koraks]

But on closer inspection, and especially when mapped on the visible spectrum, you can clearly see the both 4000K and 6500K have an extreme peak at 450nm which is mostly blue light, which is dis-proportionate to the rest of the emitted spectrum. In retrospect, the 2700K is much more similar to the MG paper's curve and to tungsten in that range.

Please see my post above. It's not so much the paper's response curve you should be looking at, but the filter transmission curves.

As to the 450nm peak: all white LEDs have this. A white LED these days is basically just a 450nm blue LED with a phosphorous coating on it. This coating converts 450nm blue light into a broader spectrum. White balance is adjusted by tailoring the phosphorous and balancing the phosphorous with the native 450 blue wavelength. If you look closely, you still see the 450nm peak in the 2700K LEDs. Since the multigrade filters select especially for blue wavelengths significantly below 450nm, the peak is not really a problem when printing, with the caveat of speed matching between filters; again, see my post above.

 

[Alexo]

Yes, welcome to Photrio, and thanks for taking steps to figure out this relevant issue. Many people are interested in replacing their tungsten enlarger bulbs with LED devices. Testing like yours can help to shed light onto the matter.


To be frank, this isn't as relevant as it may seem. The overlap between the spectrum of the bulb and the sensitivity of the paper may tell you something about the speed matching of a set of filters. For instance, the Ilford filters are speed-matched across a couple of contrast grades, but this assumes a typical halogen light source. If you install a LED bulb that emits a relatively large amount of blue light (i.e. higher color

Yes! I realized this after more careful examination and research. I later mapped the visible spectrum against the MG paper curve, plus the LED curves and got a completely different perspective!! I posted this as a follow up thread that much better explains the results I got.

I agree with you that the 2700K is the closest to the incandescent tungsten bulbs such as the PH211 which is the standard bulb to be used in old and basic Durst 3xx enlargers. I also think that higher K LEDs can be used by combining yellow and magenta filters. It may offers some interesting results. In any case, I think that for replacing the PH211, the 2700K seems like a pretty good fit.

 

[koraks]

I later mapped the visible spectrum against the MG paper curve, plus the LED curves and got a completely different perspective!!

That's not what I meant, though. It's not just about the paper curve. It's a about the filter transmission curve.

I also think that higher K LEDs can be used by combining yellow and magenta filters.

They'll work with multigrade filters, too, but speed matching between the grades may be off. This will likely be off with the 2700K LED as well since it's still a little different from a halogen bulb. Whether this is a problem, is a matter of personal preference.

 

[Alexo]

Please see my post above. It's not so much the paper's response curve you should be looking at, but the filter transmission curves.

Yes, you are correct. I did not think of looking up the MD filter transmission curves and it offers a much better perspective.

Anyway, I think the objective of the thread was accomplished, which was to open up a more scientific discussion as opposed to "I had great results with this or this other bulb", etc. I had no idea of any of this, even though I have been developing on-and-off for decades!

I wonder what it would take to combine several LEDs to replicate the tungsten and/or halogen curves. This is very typical in grow-lights today. Maybe it's already being done..

 

[koraks]

Anyway, I think the objective of the thread was accomplished, which was to open up a more scientific discussion as opposed to "I had great results with this or this other bulb", etc.

That's certainly very valuable. In that vein, I'd certainly recommend repeating your tests with a step wedge instead of a photograph. A photo is of course pretty, but more difficult to interpret than a step wedge. Something like a Stouffer 21-step is suitable for this, but there are other options as well.

I wonder what it would take to combine several LEDs to replicate the tungsten and/or halogen curves.

I don't think it's very feasible to approximate tungsten all that much better than a decent white LED bulb already does, and the question presents itself how useful it would be in the first place. It would mostly serve to somehow replicate a tungsten bulb for use with multigrade filters, but we already know for decades that exposing these papers with pure blue and green light also works extremely well, and it's the basis for all 'multigrade' light sources such as the Ilford 500 series as well as modern blue/green LED solutions.

 

[Alexo]

I don't think it's very feasible to approximate tungsten all that much better than a decent white LED bulb already does, and the question presents itself how useful it would be in the first place. It would mostly serve to somehow replicate a tungsten bulb for use with multigrade filters, but we already know for decades that exposing these papers with pure blue and green light also works extremely well, and it's the basis for all 'multigrade' light sources such as the Ilford 500 series as well as modern blue/green LED solutions.

LOL, of course, and certainly more energy efficient and precise. I would suspect that for many hobbyists, there's a certain mystique, ritual, nostalgia and artistry in the use of clunky old photographic equipment and doing things the hard way ;-)

As some countries have either banned incandescent and halogen light bulbs (or stores have stopped selling them because there is no demand) it is always good to find ways to minimally modernize some of this equipment, just enough to keep them working, but without killing their character. Taking old enlargers and migrating them to LED seems like the right kind of upgrade, all whilst reducing their energy footprint in about 90%!

--
Alex

 

[Alexo]

That's certainly very valuable. In that vein, I'd certainly recommend repeating your tests with a step wedge instead of a photograph. A photo is of course pretty, but more difficult to interpret than a step wedge. Something like a Stouffer 21-step is suitable for this, but there are other options as well.

How exactly ? You mean using it as the negative ? My enlarger's film carrier is only for 35mm and smaller so I would have to do it in steps? In other words, one exposure per as many frames of the wedge fit lengthwise ? Or perhaps it can be done by contact printing with a glass? I'm willing to give it a go but no idea how to use step wedges in enlarger printing.

 

[koraks]

Or perhaps it can be done by contact printing with a glass?

That's it.

 

[FotoD]

The 4000K LED seemed to work pretty well. Maybe I should have dialled back the burn to 3 seconds to get it closer to the baseline image, but the results, overall seem tu suggest that the contrast levels work as expected.

I've used 4000K LEDs with Foma Variant papers. It works great.

 

[mshchem]

How does this correspond to dichro head sources?

 

[mshchem]

I should say compare.

 

[koraks]

How does this correspond to dichro head sources?

Take the transmission spectra of the M and Y filters, plot against bulb transmission spectrum and see if you can make sense of it.

Better yet, plug in a LED bulb into a dichroic head and print a step wedge. Repeat with incandescent bulb. Compare results.

 

[mshchem]

Yeah that sounds like a lot of work.
My equipment doesn't really take LEDs. At least not yet. I have a beseler head setup for VC, green and blue dichro channels with a separate white light for graded. I have spares, but one of these days I will need to do something (assuming I outlast the light source). I also have a couple Zone VI enlargers with green and blue coldlight tubes, these work fine now, but again, the clock is ticking.
The Beseler units will work with condensers so that would be a way to use LEDs.
Best Mike.

 

[koraks]

Converting to LED when using condensers may or may not be a straightforward task. It all depends...I'm no optics guy so I'm somewhat out of my depth here, but I do know that while a LED is theoretically pretty much a point source, once you start scaling up power, you're usually dealing with 2-dimensional arrays. And that can create challenges with an existing condenser stack that assumes a (quasi) point source. I guess everything is OK as long as you replace a frosted bulb with a similarly opaque LED one. But if you're replacing stuff like those smaller halogen lamps found in many enlargers, I think there will be challenges ahead.

The cold lights are fairly easy to replace with LED arrays paired with a diffusor.

 

[kingbuzzie]

I have a durst 606, and I just ordered this now that I've set up my darkroom again: https://store.waveformlighting.com/...t-full-spectrum-flicker-free-a21-15w-led-bulb

It has a CRI of 95, they seem to cater to a lot of pro lighting work, so I think they are trustworthy. I choose the 3000k version. I have an advantage in my enlarger due to the fact I still have the opal filter (it was designed to take a clear t8 bulb as well as a opal bulb).

 

[Alexo]

I have a durst 606, and I just ordered this now that I've set up my darkroom again: https://store.waveformlighting.com/...t-full-spectrum-flicker-free-a21-15w-led-bulb

It has a CRI of 95, they seem to cater to a lot of pro lighting work, so I think they are trustworthy. I choose the 3000k version. I have an advantage in my enlarger due to the fact I still have the opal filter (it was designed to take a clear t8 bulb as well as a opal bulb).

I followed the link but there was no 3000K version.

I have been using the 2700K OTS bulb with pretty good results. The only thing is that the startup time is too slow so I have to use ND filters and extend burn times to get consistent results.
I believe 3000K is the sweet spot but I haven't found any OTS E26 that are 3000K and instant start.

 

[kingbuzzie]

I followed the link but there was no 3000K version.

I have been using the 2700K OTS bulb with pretty good results. The only thing is that the startup time is too slow so I have to use ND filters and extend burn times to get consistent results.
I believe 3000K is the sweet spot but I haven't found any OTS E26 that are 3000K and instant start.

Sorry, I linked to the wrong one: https://store.waveformlighting.com/products/centric-home-flicker-free-a21-15w-led-bulb

 

[ic-racer]

Seems like a lot of trouble. Are the incandescent lamps no longer available?

 

[Alexo]

Nice. Thank you !

The specs don't show the startup time.

 

[Alexo]

Seems like a lot of trouble. Are the incandescent lamps no longer available?

They're becoming scarce and expensive in some places.