Dealing With Variables

[Red_Right_Arm]

Okay, so I've come to terms with the fact that my company will just always use a bunch of variables for every job. Each new job brings with it a new combination of press, substrate, ink density, anilox cell count, anilox BCM, plate line screen resolution, screen angles, dot gain, dot gain adjustment curves, and various RIP settings that effect the dots in some way. That's just the way it is (for now at least).

Today we had a job that was a nightmare on press. We tried everything to see if it'll work out. With all of the trial and error and changing things, in the RIP and on press, I overlooked to set the correct line screen on one set of plates. This caused press downtime as the plates were remade. If we are going to have to keep up changing all these variables to get a job to print looking correctly, it seems that variables ARE the job, and the Prepress department just needs to roll with it. That being said, does anyone have any suggestions for how to handle a constantly changing set of parameters reliably and accurately?

 

[davarino]

You can't.

That is an overstatement, but it's near the truth.

You need a comprehensive baseline in order to know what to do next. If you have no baseline, you have no sure counsel for what to do to improve things; all you can do is "make changes" rather than make a change that you know ought to improve the situation.

In our shop, we refer to lack of a baseline as "shooting at a moving target". I refer to it more as "shooting at a morphing target with a different gun and ammo every shot".

__________

If you remember going through high school algebra, you may remember solving simultaneous equations.

It could try your patience to take 3 equations and try to solve for three variables, but it could be done if the three variables (and only the 3 variables) could be found in all three equations.

It was impossible to do it without guesswork if either​

• there were more variables than there were equations, or​
• there were missing variables in strategic spots​

(This is also true with more or less than three, I chose three because it's a good "for instance".)​

This is the same situation you run into if you have different substrate, different press, and different inks... and then add dot gain curves and RIP setting adjustments.

If you run only 4 kinds of substrates, have 2 presses, and use 2 significantly different inks, you have 16 possible scenarios. All else being equal.

So, let's say that you determine a repeatable output for those 16 kinds of jobs, with ABSOLUTELY no changes allowed to color curves or RIP adjustments. And no judgments about whether those outputs are completely optimal... they are simply "repeatable without any tweaks".

Now you know where you have to go, simply based on substrate, press, and ink.

It may be (it PROBABLY will be in most cases) that you have to then have to make further tests based on dot gain and RIP settings, etc., etc.

All this takes time. Most people don't want to use the time to do this because it is not "productive".

So how do you make it happen? You can do it empirically. It takes time, but it's invisible time.

• Make a list of every variable possible in the process.
• Whenever a job is run, record what the values of the variables were for an acceptable output. ("Ink at 4.5; plate tightened 0.4"... etc.)
• Also, if you can, record any unacceptable outputs' values of variables.
• As you build your data base, you can play "simultaneous equation" with it (spreadsheets are SO useful, especially if you use their goal seeker or solver tools)
• Eventually, you will probably see patterns.

This will not be quick, nor effortless, nor dramatic. But it will probably lead you in the right direction. And you will most likely not be recognized for your efforts.

Applying what you learn after you have an idea of what is what: Whenever there's a problem you should insist that the press, plates, ink, etc. all be brought to their baselines and then start solving the problem. Do not start from anything other than completely baseline: it will be a waste about a third of the time per change unless you know exactly what you're doing. (Can be shown mathematically)

 

[Red_Right_Arm]

Thanks. I'm trying to keep an open mind and tenacious attitude toward standardizing. There's A LOT of variables to attempt getting control of. I'll revisit this subject after I see what I can make stick.

 

[akalaray]

If you can't measure it, you can't control it!

 

[davarino]

If you can't measure it, you can't control it!

True to a point. Also quite untrue to a point. Not all of value is measurable.

 

[akalaray]

davarino, Would you say that would be subjective then?

 

[Deleted member 16349]

If you can't measure it, you can't control it!

This phrase is used often but it is meaningless. It is an over simplification of what can be complicated control issues. Each control problem has to be evaluated on its own specific merits.

The problem with the statement implies to the general public, that a closed loop feedback solution of some kind is the best way to control variation in a process or a problem. This is not the case. Closed loop control is used when one can not find a better way to solve the problem or if the tolerance is so small that even a stable process still needs help.

An open loop control method, using proper calibration on a process that is made inherently consistent will out perform a closed loop control system used on a process that is not consistent. So I say again, as I have for many years on this forum, printing processes need to be made consistent by designing out the fundamental causes of variation.

There are many examples of controlled processes that work very well without continuous measurement due to the control method used.

Say you have a wooden or metal part and you want to control its fixed position. What is done? Not an closed loop control but an open one. You nail it or bolt it or glue it down.

All kinds of pick and place automation does not use closed loop control to move a part from one location to another such as in installing micro chips on circuit boards etc. The movements are preprogrammed to determine distance and speed of moving the pick and place device on rails. Fast and highly accurate placement is then possible.

AC motors run at an accurate speed and do not require closed loop feedback. Your wrist watch does not need closed loop feedback from you to ensure it runs accurately for long periods of time. Your TV set or computer monitor does not have any device to look at its picture output to measure it and control it.

If one thinks that the only way to control something is to measure it, then one will not bother to think of the better solutions that are possible to solve problems.

One can not calibrate a process if it is not consistent first. One will not be able to have a predictable process unless it is consistent and calibrated. Inherent consistency in a process is the prerequisite for the control needed to have high capacity, low waste and high quality.

Also, are we actually solving a problem or are we just dealing with the symptoms?

Closed loop colour control is addressing the symptom of variation in colour output but it is not addressing the fundamental cause of colour variation. It works but it is not the best solution.

Just as an example. Someone who has cancer, probably will have pain at some point in their illness. Pain is a symptom. Giving strong pain killers may help with the symptoms of pain but it is no cure for the fundamental cause of the cancer. In the printing industry there is only the interest in methods to address the symptoms but not the cause. So even with the measurement methods continually pushed by the experts in the industry, the cancer of variability still remains in the processes.

 

[davarino]

Indeed, Erik is right: The system dictates the variation.

Without changing the system, the best you can do is kludge inside the system to get the best possible performance.

As a rule, a bright person on the shop floor has to resign him/herself to the fact that he/she cannot make substantial capital improvements, cannot hire and fire, cannot "make" others act in certain ways.

There are blunders that can be made at the moment of this realization: hopelessness or gullibility.

Hopelessness is obvious. You see it in a lot of older journeymen and foremen who have resigned themselves to a 45-55 hour a week flurry of craziness.

The gullibility blunder is thinking that there is a solution all boxed up and ready to use.

Let's face it, there are a lot of things about our craft(s) that are not utterly clear, nor are they measurable.

For example, three people will see colors in three different ways, especially if they get to see those colors in various environments.

There is no way to measure this. None. All we can do is try to standardize the viewing conditions: but the "too green" will still be incomprehensible to many.

And then, to communicate the need for color change to a pressman and thence to an ink train that is already set up... ah!

Add to this the well known fact that a color bar tells you nothing about people's faces or (God forbid!) whether a logo's color is right enough that an agency won't reject the printed piece.
​

The first trick in improving any system is to reduce variability.

But do not confuse that with insuring quality, it only helps ensure consistency.

For example, you may develop a system that produces consistently mediocre plates. Usable, but nothing to write home about.

Is mediocrity better than getting the occasional brilliantly produced plate? I say yes. Simply because the cost of that one wonderful plate is that there has probably been at least one horrific plate (with an extra hour or two of makeready) as its sibling.

Did I used to believe this? No. That, I would say, is the difference between apprentice and journeyman, and possibly between journeyman and master.

(We are here to do our best with what we have and what we are. And that changes.)