FNO - Fading in black/cyan/"photo" dye based inks

Grandad35

Printer Master
Platinum Printer Member
Joined
Feb 24, 2005
Messages
1,669
Reaction score
182
Points
223
Location
North of Boston, USA
Printer Model
Canon i9900 (plus 5 spares)
This post discusses some aspects of fading of dye based inks, and uses information posted in (http://www.nifty-stuff.com/forum/viewtopic.php?id=498).

It has been noted that when black inks fade that they often take on a red color cast. The attached image shows the spectra of several colors swatches that may help to explain why this happens.

The first set of data compares Formulabs dye and pigment black inks, showing several points of interest:
1. The pigment ink is darker across the entire spectrum, giving a darker color. The pigment ink is very effective at all wavelengths.
2. The dye ink loses effectiveness at the longer wavelengths (far reds), but this is at the far limits of our vision and may not even be seen.
3. The dye ink is also slightly less effective at the shorter wavelengths (violet/cyan/blue), and this gives a blue cast to the dye black.

The second set of data shows cyan and PC. For now, just look at the far right of the cyan spectrum - it is far better at blocking the reds than the black. For a test, a mixture of 2/3 black&1/3 cyan was painted and measured. The third set of data shows that the ability of the blend to block the reds is better than the black alone, but at the expense of reducing the blocking ability at the shorter wavelengths. I am not suggesting that cyan should be added to black ink, but there may be other dyes that could extend black's blocking ability further into the reds without reducing its effectiveness at the short wavelengths. The reason that this is important is because of what happens when the black fades. It is unknown whether an ink's blocking ability is uniformly reduced at all wavelengths as it fades or not, but (since we have no real information on this) let's assume that a faded ink acts like a diluted ink so that we can guess what might happen when the black fades.

The bottom set of data shows the spectra for black when the ink is at full strength and diluted to 50% and 25%. Just like we saw with the diluted inks in the previous post, the effect of dilution (and therefore probably fading) is that the transition points are shifted sideways so that (in this case) shorter wavelengths aren't affected and only light at the long wavelengths (reds) will be increased. Obviously, more red light reflecting from the paper will give a red color cast to the image.
BlackFading.jpg


Another color that is often mentioned as having fading problems is cyan. It was previously shown that PC has about the same color as a 25% mixture of cyan ink and 75% water, so comparing cyan and PC will give a good comparison of the spectra of cyan and diluted (faded) cyan. The cyan and PC have "Lab" color values and corresponding "HSB" values of:
L a b H S B
Cyan 39 -18 66 208 100 77
PC 65 -45 -53 197 100 98

Note that cyan's "Brightness" was greatly increased when the ink was diluted. When yellow and red were diluted, they didn't change very much in brightness, but became less saturated. Magenta changed in brightness from 75 to 87 and also became less saturated. On the other hand, the cyan lost almost all of its light blocking ability (77-->98 on brightness). It may not be so much that only cyan fades, but that we are more sensitive to a change in brightness than either saturation or hue. Another factor that may also be at play is that cyan inks are less likely to block UV than yellow (but about the same as magenta). Is it possible that yellow's ability to block the short wavelengths extends into the UV spectrum and that yellow may help to protect against UV fading by partially blocking the UV?

The same reasoning extends to "lighter" (photo) inks (PM and PC) and why they are also reported to fade faster than the "darker" inks (cyan and magenta). These photo inks were necessary to get a high quality print when only large drop sizes were available, but is it possible that a 1 or 2 pl drop size has made the photo inks unnecessary? This would reduce ink usage by 1/3 (http://www.nifty-stuff.com/forum/viewtopic.php?id=466) and increase the life of our prints at the same time. Has anyone done a direct comparison between one of the new 4 color (plus pigment black) 1 pl Canon printers with an i8500 or i9900 to see if the lack of photo inks is compensated for by the 1 pl drop size?
 

JV

Getting Fingers Dirty
Joined
Apr 23, 2005
Messages
91
Reaction score
0
Points
29
I am attempting an ink fade test of OEM, BIJC#1, BIJC#2 and OA100 inks similar Renard DellaFave's tests in: http://forums.dpreview.com/forums/readflat.asp?forum=1003&message=12198835

Using a 2 cm x 2 cm plastic template and painting ink with Costco Q-tips, I can apply the ink about as evenly as DellaFave. I would like to apply the same amount of ink in each square.

In the Topic "Basics of combining ink colors" Grandad applies 8 drops of ink on a fresh swab (Q-tip) for each swatch (square) dropped from a blunt #18 needle. I could try that with a needle or a pipette, but I am still thinking of alternatives. Any ideas?

How good would the samples have to be for a Spectrophotometer?

JV
 

Grandad35

Printer Master
Platinum Printer Member
Joined
Feb 24, 2005
Messages
1,669
Reaction score
182
Points
223
Location
North of Boston, USA
Printer Model
Canon i9900 (plus 5 spares)
JV,

FNO warning

It's amazing that you wrote your post at the same time that we were working on the same problem.

I generated color swatches with 5 different ink sets that I have, and Rob and Fotofreek sent swatches made with 6 of their inks. The goal of testing these color swatches was to start a catalog of CIELab colors for known inks so that we could (hopefully) identify an unknown ink by matching its colors to a known ink. It might also be possible to individually select each ink color from the supplier with the closest match to the OEM inks for those who don't want to deal with profiles and were willing to run a "mixed supplier" inkset. At the very least, we would have a scientific basis for deciding which inks caused which color shift.

Rob used swabs and Fotofreek used an artist's brush to apply their inks, while I used the "8 drops/swab" technique. I thought that applying a consistent amount of ink to the swab would give consistent color, but was I ever wrong. When I measured the various colors, it quickly became obvious that the method used to apply the ink affected the color readings (mainly the "Lightness"). For example, the brush painted samples gave the most consistent color across the swatch, but they were darker than the swabbed samples.

A new term needs to be defined - "Delta-E" (or dE). This term was used by alchemist in (http://www.nifty-stuff.com/forum/viewtopic.php?pid=2998#p2998). When two colors are compared, dE gives a measure of the total color difference. As a point of reference, 10 readings on a uniform gray surface gave an average dE between any two readings of 0.3, with the highest difference being 0.6. Reading the same point on the surface 6 times in a row gave only one reading with a dE of 0.1 - the rest were 0.0, showing that the spectro itself is very repeatable. Two more reference points - the dE between magenta and PM is typically in the range of 30-40 and the dE between cyan and PC is 40-50.

Some basic math. From a previous post on ink usage, we know that an average 8x10 print uses about 0.5 to 0.6 CC of ink (excluding cleaning cycles). This works out to an average ink thickness of about 10 microns. Based on how long it takes my i9900 to print an 8x10, estimating the time that is spent actually printing and using 6000 drops/second x 2 pl drops from each of 768 nozzles (for one color), you get a maximum ink lay down of 0.35 CC from a single color (7 microns of a single ink). Using a (crudely estimated) ink drop diameter from the syringe of 1.5 mm and 8 drops of ink spread uniformly over a 1" square (my swatch size), there is enough ink on the swab to lay down 22 microns of ink if it was all transferred from the swab to the paper. Actually, getting a 7 micron layer of ink from this setup is reasonable, since only part of the ink is transferred when the swatch is painted. In any case, a series of 6 swatches of the same magenta ink were painted on the same sheet of paper using the same 8 drops/swab. The average dE on these samples was about 4. Since the average dE on the 11 color swatches of magenta from different suppliers was only slightly higher than this, it is clear that the swab painting technique has too much variation to generate the color samples for our tests. The spectra for the 6 swatches of the same ink all showed almost exactly the same shape, but looked like the ink had a different strength (was diluted differently) for the various samples. The spectra for the 11 magenta samples had different shapes as well as apparent strengths.

When I tried using a brush, there was a glossy layer of ink on the paper indicating that I had applied more ink than the paper could rapidly absorb. This can't be good, since it never happens in actual printing because it will cause pooling of the ink, as can happen on swellable paper when the ink is laid down too fast. It also means that the color of the swatch will be affected by how much ink the paper can quickly absorb. It will be difficult to get uniform results if variations in the paper can have an effect on the colors. In any case, this technique also gave an unacceptable dE between "identical" samples.

I tried "doctoring" a puddle of ink across the paper's surface and anything else that I could think of, but nothing gave consistent results. For example, doctoring the ink across a swellable paper gave a very much lighter color than a nanoporous paper. At this point, I am of the opinion that the only practical "home" solution may be to buy a used printer that uses BCI-6 carts and which prints solid blocks of each individual ink as part of some test routine to generate the samples. Maybe someone has access to an anilox roll (http://www.harperimage.com/anilox-specify.asp) that could be used to meter the ink (by doctoring the ink onto the roll, then pressing the inkjet paper against the roll to transfer the ink to the paper) to generate consistent samples. As a point of interest, note that this link recommends a 3 micron ink thickness for a single color.

I realize that you asked for the time and I told you how to build a watch, but I wanted to describe the problems clearly so that you didn't waste your time covering the same ground. Given your printing experience, I was hoping that you might know of a way to generate uniform color samples. How do ink suppliers and printers verify an ink's color? Surely they don't just "print and pray". There must be a simple way of applying a consistent, known thickness of ink to small samples of paper.

JV said:
How good would the samples have to be for a Spectrophotometer?
If your goal is to run tests similar to DellaFave's and you are just interested in the fading results, the swab application should be acceptable since you will be comparing scans of the same paper sample before and after fading, so minor variations in the ink's initial color won't matter. On the other hand, if your goal is to compare the colors of inks from different suppliers, the variations in how the ink was applied can severely affect your results.
 

JV

Getting Fingers Dirty
Joined
Apr 23, 2005
Messages
91
Reaction score
0
Points
29
Grandad,

Thank you very much for responding. I tried the following:

1. Printed the Nozzle Check three times on the same sheet of Kirkland paper. It gives the individual colors, but they look considerably lighter than colors of ink swabs or colors of photos.

2. Printed from PSCS using color numbers for RGB of:
255, 255, 0 Yellow,
255, 0, 255 Magenta
0, 255, 255 Cyan
0, 0, 0 Black
255, 0, 0 Red
0, 255, 0 Green
with Grandad's printer profile for BIJC#2 (i9900_Costco_InkGrab_3-2-05_1). Printed colors differed from colors of ink swabs.

When using "Assign profile": .. InkGrab_3-2-05_1, color numbers did not change and printed colors looked like the ink. I do not know if each color printed is only the one ink color or the result of mixing colors.

3. In reading the Ink Patents and Patent applications in http://www.uspto.gov/index.html there is much discussion in comparative testing of inks. In some of the patents the authors refer to a printer they used. I will re-read the Patents and look for how the ink specimens were printed or made.

JV
 

JV

Getting Fingers Dirty
Joined
Apr 23, 2005
Messages
91
Reaction score
0
Points
29
Please see the revised Post #5

JV
 

Grandad35

Printer Master
Platinum Printer Member
Joined
Feb 24, 2005
Messages
1,669
Reaction score
182
Points
223
Location
North of Boston, USA
Printer Model
Canon i9900 (plus 5 spares)
JV,

The nozzle check print is a good idea, and I let it print 8 times to darken the ink. Obviously, the nozzle check puts down a LOT more black ink over the vertical bars, as they started to smear after 3 prints. The black bars were wiped with a paper towel after each succeeding pass (but I forgot about the "2" above the right set of patterns).

The colors were very uniform on each set of 5 readings/color, with at least 4 of 5 readings having a dE of 0.3 or less. Unfortunately, the colors were still too light. The print was scanned at 2400 ppi and 600 pixel (1/4") samples of the black and cyan patterns are given below - the other colors all had similar patterns. I was absolutely amazed that all 8 sets of dots were almost perfectly aligned, given that any differences in how the paper fed into the printer would affect the alignment. This pattern explains why the colors were so light - less than 25% of the paper's surface had any ink.
NozzleCheckPrint.jpg


Because the internal RIP in the printer converts RGB values to CMYK in a way that is optimized for Canon inks, there is no way that I know of to print a single ink using a RGB value. Also, the RGB--> CMYK conversion doesn't always act as you might think. For example, open the "Color Picker" in PS and select RGB as 255/0/255 (for magenta). Note that the values in CMYK are 11/79/0/0. The values for 255/255/0 and 0/255/255 are 7/0/100/0 and 87/0/20 in CMYK. I gave up on trying to figure out this conversion process a long time ago.

Taking your idea one step further, I also tried the "Extended Nozzle Check" print. It prints a narrow band of solid ink colors that is just large enough to cover the opening in the spectro, and it could probably be used to generate the color tests. For some reason, this test routine only prints 7 of the 8 colors, skipping red. However, I am not really thrilled about using my main printer to generate these test patches, as it is working very well and I am not ready to risk clogging the head by printing with numerous "unknown" inks. Is there a cheap 4 color printer using BCI-6 carts and which prints solid individual color blocks as part of its test routines? The standard nozzle check patterns and the "staircase" patterns are not acceptable for this - we need a solid block of color.
 

JV

Getting Fingers Dirty
Joined
Apr 23, 2005
Messages
91
Reaction score
0
Points
29
Grandad,

1. To print Nozzle check on Canon ixxxx and iPxxxx printers : Press and Hold the Resume/Cancel Button (the upper button). When Power Lamp blinks two times, Release the Button.
To print the Extended Nozzle Check: ?? No information on Canon web site.

In local stores I performed the Nozzle check on the iP6000D and MP780 printers. Both have the light (25% coverage) printout.

2. In reviewing ink Patents, they always use an older inkjet printer (Epson 660, Epson 3000, Canon S520, Canon BJ-F870, Mutoh 4100,...) to print test ink patches. Printing is usually at 25, 50, 75, and 100% coverage. There is no explanation how they print the inks directly. The the printers are older because the patents were submitted some time ago.

In Patent Application 0050140763, Inkjet printing and appatatus by Du Pont : "Two Epson 850 printheads were mounted above a rotating drum......drop size could be varied by altering the piezoelectric element drive signal...."

3. Has anyone tried spraying ink with an Airbrush?

JV
 

Grandad35

Printer Master
Platinum Printer Member
Joined
Feb 24, 2005
Messages
1,669
Reaction score
182
Points
223
Location
North of Boston, USA
Printer Model
Canon i9900 (plus 5 spares)
JV,

To print the extended nozzle check pattern, see:
http://www.nifty-stuff.com/forum/viewtopic.php?pid=869#p869

I also thought about using a spray system like an airbrush, but the problem of how to apply exactly the same amount of ink for each sample remains.

I found a few devices that appear to be designed for this purpose, but they (1) are too expensive for our use and (2) probably won't work on inkjet paper, where the paper absorbs the ink very quickly.
http://www.uvprocess.com/Products/C...PEEK INK PROOF TEST.DRAWDOWN L/DRAWDOWN L.ASP

I am sure that I could make the patterns from the "Extended nozzle check" from my i9900 work (the relevant portion is shown below), but I would rather find an inexpensive printer that could print these patterns, even if it only printed 4 colors at a time.
ExtNozChk.jpg
 

JV

Getting Fingers Dirty
Joined
Apr 23, 2005
Messages
91
Reaction score
0
Points
29
Grandad,

Printing the Extended nozzle check pattern with the iP8500 gave samples only for the Cyan and Photo Cyan inks so far. I will keep trying with the iP8500, i860, i950 and store printers.

Inks were applied by draw-down to plain paper in Patent Application 20050155516 "Inkjet ink composition" by Du Pont.

Ink could be weighed (0.17 +-.01 g for 4 x 6 sample) but would probably be difficult to spray uniformly.

JV
 

JV

Getting Fingers Dirty
Joined
Apr 23, 2005
Messages
91
Reaction score
0
Points
29
Grandad and others,

We wanted to print ink patches for inks commonly in use to compare color and fading properties versus OEM inks in order to select the best inks. Grandad printed such ink patches with his i9900 printer in Post #9 using the "Extended nozzle check pattern".

I printed with all the printers available to me and found that the i9900 was the only printer that could print the ink patches at 100% coverage for the spectrophotometer.

It is possible that the Canon S seris printers, not available to me, could do the task.

To print the "Extended nozzle check" see http://www.nifty-stuff.com/forum/viewtopic.php?pid=869#p869 or the source in http://www.fixyourownprinter.com/forums/inkjet/16073#1 All printers starting with Canon S series use the same "secret code". Printing can be done from any store printer with ink and power.

My results for the following printers were:

Extended nozzle check Nozzle check Head alignment

i860 (1) (4) (4)
i950 (2) (4) (4)
iP3000 (1)
iP4000 (1)
iP5000 (1)
iP6000D (3) (4)
iP85000 C,PC ink patches (5) (4) (4)
i9900 R,PC,PM,Bk,C,M,Y ink patches

(1) http://www.nifty-stuff.com/gallery/inkjet-refill-1/canon_extended_nozzle_check_pattern The cyan patch looks less than 100% coverage.
(2) Same as (1) except black patches in place of the cyan patch.
(3) Same as (1) except a purple patch in place of the cyan patch.
(4) The ink patches printed are less than 100% coverage (~25% coverage on the iP8500 and i9900) see Post #7
(5) Response from Canon: "After checking we do not have any information on a nozzle test with 100% coverage" - Jack, Technical Support.

Grandad, I could mail you the OEM ink patches I printed with the i9900 at our local store. I could also print the ink patches for BIJC#2 on my iP8500 by printing two colors at a time using the appropreate number of cleaning cycles between the cart changes.

JV
 
Top