Printalyzer UV/VIS Densitometer

[dkonigs]

Hello again everyone!

I spent a bit of time talking about this project of mine near the end of my previous densitometer thread:

Printalyzer Densitometer - A compact budget-friendly densitometer project

Hi @dkonigs 😀 I just got my Printalyzer Densitometer and after first test I loved it! 🥰 Thank you and I'm waiting for Printalyzer Enlarging Timer!

www.photrio.com

However, a lot has happened since then and I figured it was time to give this upcoming product a thread of its own.

The UV/VIS Densitometer is most likely going to be my next released product. I expect to have pre-release units ready to send out in the coming weeks, with full production getting underway sometime this summer. Hopefully that means full availability by this fall.

The initial product page is now up on my company website, with a bunch of bullet-point details:

Printalyzer UV/VIS Densitometer — Dektronics

www.dektronics.com

Okay, so with that out of the way, I might as well talk about what this thing is.
The Printalyzer UV/VIS Densitometer is going to be my next transmission/reflection densitometer product. On the exterior, it looks exactly like the current Printalyzer Densitometer. But on the inside, there are a lot of changes. New sensors, new LEDs, new optical path, new calibration methods.

The main headline feature is the addition of a UV transmission mode (using a 385nm UV LED light source and a UV-A sensitive light sensor), which should be able to give measurement results matching the old X-Rite 361T. It can probably maintain sufficiently accurate and repeatable measurements to higher UV densities, however, because of its light source.

For the "VIS" mode, I'm now using 3000K full-spectrum LEDs that more closely resemble the spectrum in the ISO 5 standards, combined with a light sensor that has a built-in Photopic filter. This may not make much of a difference versus the older product under most B&W measurement conditions, but it might help in some corner cases.

The optical path itself also has a number of improvements. Below the film, there is now a flashed opal diffuer (instead of matte acrylic) positioned to be in direct contact with the film. Above the film/paper, there is now a focusing lens which improves the precision of the measurement spot. This should make it easier to measure smaller patches on step wedges.

Finally, I've included a temperature sensor inside the optical sensor head, and am actually going to be doing thermal calibration and correction! Now I'll admit that the actual effect of ambient temperature on VIS measurements is miniscule, but for UV measurements its significant enough that you might notice a small change if you go between a cold house and a hot darkroom (or vice versa).

I've also been taking a good hard look at my options for calibration. While my current product is tracable to the Stouffer T5100C and R550C, for this one I wanted to take things a step further. As such, my current calibration plans for this new UV/VIS densitometer are:

• VIS Transmission - Tracable to NIST 38120C (formerly known as SRM 1008)
• UV Transmission - Tracable to a relatively new X-Rite 361-68 reference
• VIS Reflection - Most likely going to be tracable to a properly calibrated set of Spectralon references and/or BCRA tiles, falling back to an X-Rite 302-21 if that doesn't work out

A lot of research (and comparing data across many densitometer references, reference strip measurements, etc) went into arriving at these decisions, as nearly everything has its tradeoffs. Even seemingly minor differences in the measurement setup can make noticable difference in what values you get. (I also evaluated new T5100C and X-Rite 301-27 references as part of this.)

That should be enough to get this thread started. Hopefully I'm not opening a whole can of worms here, but I know I probably am.

 

[koraks]

Hopefully I'm not opening a whole can of worms here

May I? Hehe.

Great work as always @dkonigs, and what comes next is kind of particular, but I want to mention it anyway.

I understand your desired to adhere to existing calibration standards, if only to create a common baseline with existing/legacy products out there. On the other hand, I doubt many people who buy your devices will actually use them side-by-side, let alone interchangeably with such equipment. So I wonder how hard you should be on yourself/your developments in trying to adhere to these standards - although I'll happily admit that from a marketing perspective, it looks good.

The reason I'm bringing it up is the wavelength issue. Perhaps even more so than the VIS domain, wavelength makes a massive difference in terms of UV transmission density of materials. Combined with the proliferation of LED technology, which is fairly narrow-bandwidth by nature, this can create rather massive differences between measurements done at e.g. 365nm vs. 395nm. In a bit of a niche application (although seemingly gaining popularity), it turns out that for DAS-sensitized carbon transfer, a combination of exposures at different wavelengths is beneficial in optimizing shadow and highlight behavior in the same print. See e.g. Calvin Grier's work in this domain, and my own experiments and present printing methods are in line with this.

Long story short, I think there's a use case (but perhaps a relatively small market potential) for differentiation in wavelength. I wonder to what extent this present development of yours could still be augmented to accommodate this kind of functionality.

Good luck with this development, to the extent it is still underway!

 

[Carnie Bob]

Have you ever considered making a UV meter to help pt pd workers controlling density outside?

 

[Nicholas Lindan]

Have you ever considered making a UV meter to help pt pd workers controlling density outside?

What you need is a portable UV integrator. There are a few on the market. Picked at random: https://www.linshangtech.com/tech/tech405.html

Integrators pop up on ebay. Prices seem rather fanciful for the 30+ year old nuArc and Olec boxes - I suppose they qualify as 'vintage,' but I would put a fair price at $50. If you get an old one be sure it comes with the measuring probe and filter.

 

[koraks]

Maybe we should not go too far off course in this thread. The question of a portable UV integrator is already being discussed in this dedicated thread: https://www.photrio.com/forum/posts/2901637/
Let's focus here on @dkonigs new development.

 

[dkonigs]

I understand your desired to adhere to existing calibration standards, if only to create a common baseline with existing/legacy products out there. On the other hand, I doubt many people who buy your devices will actually use them side-by-side, let alone interchangeably with such equipment. So I wonder how hard you should be on yourself/your developments in trying to adhere to these standards - although I'll happily admit that from a marketing perspective, it looks good.

I think a lot of it comes out of simple frustration with how most of the "easy options" for getting calibration standards come across as a bit wishy-washy as to their traceability, but more importantly that you can't actually get off-the-shelf calibration references anymore for a lot of density measurements.

What led me down this rabbit hole was finding a metrology lab to actually do spectral scans of a T2120 strip, as a way of having enough data to basically calculate values for any density standard I like. I then discovered just how big of a difference emulsion orientation and diffuser configuration can make, and how no two standards ever seem to actually match.

Through this exercise, however, I did learn a lot about what density-vs-wavelength looks like across one of these step wedges.

At "patch 1," which is the film base, it looks like this:

But once you get to "patch 15", which has a visual density of ~3D, it starts looking like this:

(I've tried to match the scale ranges of both.)

The reason I'm bringing it up is the wavelength issue. Perhaps even more so than the VIS domain, wavelength makes a massive difference in terms of UV transmission density of materials. Combined with the proliferation of LED technology, which is fairly narrow-bandwidth by nature, this can create rather massive differences between measurements done at e.g. 365nm vs. 395nm. In a bit of a niche application (although seemingly gaining popularity), it turns out that for DAS-sensitized carbon transfer, a combination of exposures at different wavelengths is beneficial in optimizing shadow and highlight behavior in the same print. See e.g. Calvin Grier's work in this domain, and my own experiments and present printing methods are in line with this.

As far as UV wavelength, my first driving factor was attempting to get as close as possible to something else people might familiar with and/or have used before. That would be the X-Rite 361T. Now there's nothing official out there actually stating what "X-Rite UV" actually means. But I have found a few clues. First, I tried shining light into the sensor head while sweeping its wavelength with a monochromator. I found that readings peaked in the ballpark of 390nm. Second, I found some X-Rite marketing sheets which mentioned the key-word "diazo". In ISO 5-3:2009, that suggests "Type 1 Printing Density", which is around 400nm.

Of course my second driving factor was really practicality of constructing the optical path, and practicality of actually calibrating to something. The farther into the UV range you get, the wonkier the transmission spectrum of film becomes. Also, the more likely that your various light handling materials (lenses, diffusers, etc) will attenuate the UV and make it hard to get good signal.

Long story short, I think there's a use case (but perhaps a relatively small market potential) for differentiation in wavelength. I wonder to what extent this present development of yours could still be augmented to accommodate this kind of functionality.

In practice, it wouldn't be too hard to make a version with a 365nm UV LED. The problem is that the 365nm UV LEDs aren't as bright, and I'd basically need to invent my own calibration standard (likely based off that spectral data) to handle it. And it would be an either/or type of decision, since having both in the same unit would be too major of a change.

Plus, we're now getting into a niche within a niche within a niche, so it likely doesn't make sense to even go there unless people really start begging for it.

 

[koraks]

@dkonigs copy that; thanks for posting those transmission plots - they're really interesting!
And yes, it's a niche within a niche within a niche. Even worse - you fix it for one target group, you'll break it for another. And the users involved will likely never even realize it, either.

The farther into the UV range you get, the wonkier the transmission spectrum of film becomes.

Basically *everything* gets wonkier, that's the issue. As you've noticed, LED efficiency goes down, but sensor sensitivity (as well as chemical process sensitivity) tends to go the other way around and in any case tends to be extremely non-linear.

I understand your compromise at 385nm and if you have to choose any single wavelength, 385 makes good sense IMO.

 

[dkonigs]

So one feature of this device I'd like to share a few details about is temperature compensation.

This is something that I suspect many around this site think is important, because whatever sensors they were using 30+ years ago were susceptible to temperature variation. I think I've even seen some ancient DIY densitometer schematics that include some form of temperature sensor.

I honestly have no idea how many current densitometers actually include any sort of temperature correction. Data sheets for modern ambient light sensors sometimes admit there could be temperature variation, but rarely quantify it as a useful thing to pay attention to.

In my previous product, after extensive testing, I concluded that the effect of temperature on measurements was too small to care about. Thus, it has no temperature compensation.

This time around, I did the same testing with slightly different conclusions. For the "Visual" channel, the temperature effect was measurable but small enough that I could get away with ignoring it. But for the "UV" channel, the effect was still relatively small, but large enough that users might actually notice. So I decided to include a temperature sensor next to the light sensor, and put together a calibration-and-correction process.

With all the corrections in place, I decided to collect some "real world" data to show its effects. I measured a T5100 step wedge at room temperature, then moved the device to an unairconditioned closet room that heats up across the day. So I effectively collected data to answer the question of "What if you zeroed the device at 25C, then took measurements at various ambient temperatures up to 38C?"

I then plotted the change in density from one temperature to the next, both before and after correction.
(On these graphs, "P0" shows the change in an open or "zero" measurement, while P1-5 are the 5 patches of the T5100.)

Of course there are a few important factors to keep in mind:

• I've intentionally kept all these graphs at the exact same scale
• Any error that rounds to less than 0.01D will never be noticed by an end user
• Many densitometers only claim +/- 0.02D at the higher densities
• For each line, I didn't average across many measurements, so don't nitpick too much into any specific line wonkiness

 

[Bill Burk]

This is really cool!

 

[koraks]

So I decided to include a temperature sensor next to the light sensor, and put together a calibration-and-correction process.

To good effect, clearly. The results are pretty tight indeed!