Dot gain / TVI variation...

[meddington]

We should be measuring what is being manufactured and delivered rather than something that will end up in the recycling bin.

Perhaps, but most systems are not currently set up to objectively measure live image data. Don't want to re-open that worm can necessarily, but wondering what your thoughts are on a theoretical dynamic colorbar of sorts, similar to EFI's Dynamic Wedge for proof verification where the imagery is sampled and colorbar patches derived from live image content, rather than typical solids and tints or gray balance. Of course with an analogue printing press we're still limited ink film thickness as the primary process control method.

 

[rapid3]

I had similar concerns on a SM74 a few years back. I would measure 1 sheet, then
measure again 200 sheets later and get different dot gains on the same color. No
doubling or slurring at all. So I began measuring 8 to 10 sheets and recording the
average TVI. This brought the measurments down to +/- 3%. I believe the main
culprit wast the stock. We sheet all of our stock off press using a Maxim sheeter with
multiple roll stands. In the most extreme case, I have seen density differences of .15
from sheet to sheet while running reflex blue. Like every other company out there, paper
companies are cutting costs anywhere they can and this reflects in their quality. Ah, but
that's another issue.

 

[meddington]

The standard deviation for a sheetfed press is +/- 2% dot gain.

By "standard deviation" do you mean a "typical" deviation that one might expect when running, a deviation based on a standard, say ISO 12647-2, or a value based on a statistical population.

If we go by the ISO 12647-2 standard for TVI tolerances for production printing, midtone deviation shall not exceed 4% and midtone spread (difference between CMY TVI) shall not exceed 5%. Of course, with a rather wide tolerance around the solid CIELab aims, even with tight TVI tolerances there can still be visual differences between proof and press

 

[Deleted member 16349]

I had similar concerns on a SM74 a few years back. I would measure 1 sheet, then
measure again 200 sheets later and get different dot gains on the same color. No
doubling or slurring at all. So I began measuring 8 to 10 sheets and recording the
average TVI. This brought the measurments down to +/- 3%. I believe the main
culprit wast the stock. We sheet all of our stock off press using a Maxim sheeter with
multiple roll stands. In the most extreme case, I have seen density differences of .15
from sheet to sheet while running reflex blue. Like every other company out there, paper
companies are cutting costs anywhere they can and this reflects in their quality. Ah, but
that's another issue.

Also there is the fact that presses do not print consistent ink films within an image.

There are periodic variations in ink film that can show up as mechanical ghosting in severe conditions or in less severe conditions that are not be so noticeable but can affect the inking of a screen area and affect the dot gain.

There are non periodic variations in ink film that randomly affect different areas of the print.

The periodic and non periodic variations in ink film are due to the image being printed and the roller train. Ink is applied to the plate leaving patterns of thinner ink film on the form rollers. These patterns interact with other rollers in the nips and get applied back to the plate.

There are also the variation in ink film due to the charge of new ink from the ductor and some say from the timing of the oscillators, but so far I don't understand that one. Still curious though.

These kinds of variations in ink film on the image will cause variations in dot gain. From print repeat to print repeat, there will be differences and these differences are mainly due to the design of the press roller train. One main advantage of the Anicolor and other "in register" single form roller press designs is the low level of ink variation in different parts of the image.

Now if one has different inking of the plate in different areas and on different rotations, how can one have a profile that is predictable? Predictable profiles for prepress require press designs that are capable of producing consistent inking on the plate in all areas of the plate for each impression.

This is what makes printing so interesting. There are so many different problems. For me that means many things to solve.

 

[Deleted member 16349]

By "standard deviation" do you mean a "typical" deviation that one might expect when running, a deviation based on a standard, say ISO 12647-2, or a value based on a statistical population.

If we go by the ISO 12647-2 standard for TVI tolerances for production printing, midtone deviation shall not exceed 4% and midtone spread (difference between CMY TVI) shall not exceed 5%. Of course, with a rather wide tolerance around the solid CIELab aims, even with tight TVI tolerances there can still be visual differences between proof and press

Standard Deviation normally means the statistical measurement (sigma) of a limited number of samples that then can determine, statistically (if the distribution is a normal distribution), the variation in all the samples.

A range of +/- 3 standard deviations statistically represent about 99% of all the samples one would expect. Therefore with a standard deviation of +/- 2%, that implies that the whole print run was experiencing a variation of +/- 6%. Does not sound so good to me.

One has to be careful when applying statistical methods. Also I am not sure that variations in density and dot gain follow the normal distribution model.

 

[Disappointed]

Thanks all.

I am now testing the press following a day with its guts all over the place

It looks promising, if it maintains the results i am getting so far, i will let you all know what the cause of my headache was.

Dave

 

[has-been]

Erik:

What they might be referring to:
Sometimes, when the roller oscillation takes place, a light scumming or slight directional dot change occurs at the point of the forme roller contact.

Anyhow, if the person is explaing the situation properly, it would appear to be mechanical, i.e. roller settings. But somehow I think that information is lacking re: is this problem occuring in during the same run or is this problem occuring job to job? What are the run-lengths, (short-run I would suspect? Exactly what type of press?

Best regards:
has-been

 

[has-been]

dissapointed:

What is the blanket wash formula.

has-been

 

[Bret Hesler]

Now we're getting somewhere, thanks.

... rollers and duct are temp controlled, room is temp controlled.

Have you measured the temperature of the ink trains? Sometimes the temperature control readout can be incorrect.

Bret

 

[Bret Hesler]

Standard Deviation normally means the statistical measurement (sigma) of a limited number of samples that then can determine, statistically (if the distribution is a normal distribution), the variation in all the samples.

A range of +/- 3 standard deviations statistically represent about 99% of all the samples one would expect. Therefore with a standard deviation of +/- 2%, that implies that the whole print run was experiencing a variation of +/- 6%. Does not sound so good to me.

One has to be careful when applying statistical methods. Also I am not sure that variations in density and dot gain follow the normal distribution model.

Here is an excerpt from ISO 12647-2 regarding dot gain variation:

"For production printing, the average mid-tone value shall be within 4 % of the specified aim value. The
statistical standard deviation of the tone values shall not exceed, and should not exceed one half of, the
variation tolerance specified in Table 5." (which is 4%)

So from a purely mathematical standpoint 68% of the job must be within +/- 4%, 95% within +/- 8% and 99% within +/- 12%. The +/- 2% is a "Should" (recommended) not a "Shall" (required)

Bret

 

[prepmaster]

We also run a Ryobi DI and have periodically experienced similar problems. We switched from Toyo to Taniguchi and have practically eliminated them. Martin from Taniguchi came over and took some temperature measurements from unit to unit and designed special ink formulas for each color with different tac for each unit. He also increased the pigment count in the ink. The result was wonderful. Our jobs are more stable, dry faster, easily repeatable, we use less ink, the jobs register much better and a big bonus was that there is no more of the annoying toning in the yellow. Our roller temp is is 21c and our room temp stays around 74-75 f. We do not however wash up inks between jobs, just the blankets.

 

[Deleted member 16349]

Here is an excerpt from ISO 12647-2 regarding dot gain variation:

"For production printing, the average mid-tone value shall be within 4 % of the specified aim value. The
statistical standard deviation of the tone values shall not exceed, and should not exceed one half of, the
variation tolerance specified in Table 5." (which is 4%)

So from a purely mathematical standpoint 68% of the job must be within +/- 4%, 95% within +/- 8% and 99% within +/- 12%. The +/- 2% is a "Should" (recommended) not a "Shall" (required)

Bret

Bret,

I am not sure I follow your reasoning which results in +/- 12% for 99% of the job. +/- 12% means a range of 24%. That seems very large.

Anyhow I was talking about the definition of standard deviation and which is also related to Capability. Capability is related to the tolerance limits. For a capable process the tolerance limit would have to be: the absolute value of (average value minus the target value) plus 3 standard deviations. That would define the Capability of the process.

Often I see in printing the capability of the process used as a way to determine the tolerances, which is in my view backwards. Tolerances IMO should be based on the desired goal, which for printing is some allowable limits for visual consistency.

The big problem in printing is that one has more than one channel. This means that one should evaluate the allowable tolerances based on one channel being up while another channel is down and compare that with the first channel being down and the second channel being up. This kind of evaluation then determines the tolerances per channel that provide visual consistency for the pair of channels that is within reasonable levels.

With those tolerances, than one would work back to determine what the standard deviation needs to be for the process to be statistically capable.

This is the problem that the printing industry likes to avoid thinking about because if one does this analysis one finds that the process is NOT capable of providing visual consistency. Since the industry does not know what to do to deal with this capability problem they continue to talk about other issues as if those issues are somehow going to miraculously correct some inherent process problems. The process has to be changed to change the inherent variability and capability levels.

 

[Bret Hesler]

I am not sure I follow your reasoning which results in +/- 12% for 99% of the job. +/- 12% means a range of 24%. That seems very large.

A range of +/- 3 standard deviations statistically represent about 99% of all the samples one would expect. Therefore with a standard deviation of +/- 2%, that implies that the whole print run was experiencing a variation of +/- 6%. Does not sound so good to me.

Same reasoning as your example. The standard specifies that the standard deviation of the run be no more than 4%, so, three standard deviations of +/-4% equals +/-12%

Bret

 

[Deleted member 16349]

Same reasoning as your example. The standard specifies that the standard deviation of the run be no more than 4%, so, three standard deviations of +/-4% equals +/-12%

Bret

Bret,

OK, that is something very strange. It just shows how useless standards can be. I am also quite negative about standards. Standards etc. that have no relationship to the desired goal are useless.

All these standards that deal with structures of the process such as dot gain, Lab values of inks etc. are useless and what is even more damaging to the industry is that they lead people in the wrong direction. They damage the potential development of new technologies for the industry.

It would be very easy to make a standard for printing. One could say that the printed image and the target image, should not exceed a Delta E value of greater than X for every comparable point in the two images. Something like that.

How you get the match between the printed image and the target image is a matter of having the right knowledge and technology. Don't standardize the methods.

 

[meddington]

Same reasoning as your example. The standard specifies that the standard deviation of the run be no more than 4%, so, three standard deviations of +/-4% equals +/-12%

Actually, its the average that shall be 4%, and, as you wrote earlier...

The statistical standard deviation of the tone values shall not exceed, and should not exceed one half of, the
variation tolerance specified in Table 5." (which is 4%)

So the STDEV shall/should be half of 4%...or 2%, and 3sigma would then be +/-6%...though Erik's point about process capability is still valid either way.

 

[gordo]

It would be very easy to make a standard for printing. One could say that the printed image and the target image, should not exceed a Delta E value of greater than X for every comparable point in the two images. Something like that.

How you get the match between the printed image and the target image is a matter of having the right knowledge and technology. Don't standardize the methods.

The technology to do that has been around for many years. One benefit is that it would eliminate the need for a color bar. The Heidelberg CPC 2000 was on the right track by scanning the press sheet - but stopped short by using the data collected from the color bar. Also, the data did not reference the plate image.

AFAIK, this line of development in color control appears dead.

BTW, Although I agree with you, standardizing the methods generates a revenue stream for standards organizations - so I don't think you'll see that dropped anytime soon.

best, gordon p

 

[Disappointed]

Hi all, getting technical now!

Anyways, th eblanket wash is an impregnated cloth roll so no variation as such. The roller temps are around 25c which is nicely in the CTT of the ink.

The variation seems to be continuous i.e. during the run.

I am hoping to explain more as soon as i get my results finished. It looks like the cure/cause may have been found.

We do run to the ISO standard (or try to!)

Dave

 

[Bret Hesler]

Hi all, getting technical now!

Anyways, th eblanket wash is an impregnated cloth roll so no variation as such. The roller temps are around 25c which is nicely in the CTT of the ink.

The variation seems to be continuous i.e. during the run.

I am hoping to explain more as soon as i get my results finished. It looks like the cure/cause may have been found.

We do run to the ISO standard (or try to!)

Dave

That's my take on the non adoption of standards. They need to be technical to cover all the bases, but then they are too complicated for the average user to understand.

Anyway, back to your original issue. Is there any relation to ink coverage in the variation? ie light coverage = high gain or heavy coverage = low gain

Is there a trend in the variation, or is it random?

Bret

 

[Deleted member 16349]

The technology to do that has been around for many years. One benefit is that it would eliminate the need for a color bar. The Heidelberg CPC 2000 was on the right track by scanning the press sheet - but stopped short by using the data collected from the color bar. Also, the data did not reference the plate image.

AFAIK, this line of development in color control appears dead.

BTW, Although I agree with you, standardizing the methods generates a revenue stream for standards organizations - so I don't think you'll see that dropped anytime soon.

best, gordon p

Gordon Gordon Gordon,

After all these years, you still don't understand my philosophy on these issues.
Very sad. :-(

Sure one could have technology to inspect and compare all these points in the image relative to the target. Yes it is already here. I am not advocating using such technology. Basically I do not want to have to measure for control. Measure for calibration but not so much for control. Make the process consistent and predictable and then there would be no need to measure for control.

Just because one would have a standard that asks for the matching of points in the images that does not mean that one has to have active control via measurements. If a capable process, that is calibrated, demonstrates that it continually meets a standard of performance, then one does not have to measure it to confirm it each time. It will become accepted by the customer.

Heidelberg's use of image control at the press was not a great thing at all. It is a technology to compensate for not being able to print consistently within the image. Heidelberg's system could not change an area in the print in the image independently of other parts of the image. You pick an area of concern, the system would look at it and compare it with a reference. Then ink keys would be adjusted to satisfy the match in that area but at the expense of other parts of the image in line with that area.

They would never consider putting such a system on the Anicolor press since there is more consistency with the inking of the plate in all areas. Also this consistent inking of the plate means that the profile for the Anicolor press is more predictable.

I would say the standards organizations are damaging the industry for exactly the reason you stated. It is in their interest to do what they do. Now they also seem to be pulling RIT into the mix. Not a good trend IMO. The universities should be independent so that they can have an independent voice. Although I am not sure they are imaginative enough to know what is wrong. I don't think they have done such a great job in this area. Much of the time they act like users of technology and not fundamental researchers of knowledge.

 

[Disappointed]

Anyway, back to your original issue. Is there any relation to ink coverage in the variation? ie light coverage = high gain or heavy coverage = low gain

Is there a trend in the variation, or is it random?

Bret

Most of our jobs fall in the 5-15% coverage range, it was less noticeable on a heavy job. The TVI would go up and down throught the job, quite randomly.

I will know more today after i fully test the engineers work.

Dave