Closed Loop Control Limits

[mglouis]

Anyone care to share where you are setting your action limits for closed loop control for sheetfed?

Heres why I ask: I have mine set at 0.03, meaning when this limit is exceeded, ink control compensation kicks in. Recently I started scoring press operators based on SID % out keys and now suddenly, the troops want this limit raised to 0.05. This is not a big difference but I hesitate to change when from an ink control perspective, our process is not broken, however from a reporting perspective, the tolerance seems harsh. Xrite's "keys out" reporting on ATS/ITX is based on the action limit (yellow line) and coincidentally this is typically the same value for controlling ink control. Ink control may optionally be set to the control limit (red line), but I have no interest in going there since all that would accomplish is loosening closed loop and would have no affect on reporting one way or the other.

Thanks,
Matt Louis

 

[rich apollo]

Hiya', Matt. I think that .03D is awfully tight. However, there are a few variables to be considered when I say that.

1) How long are the press runs? .03D might not be an unreasonable expectation if your average run length is 1,000 sheets. But, that wouldn't give the operator, or the system, the opportunity to make any changes anyway. Would be more a consideration of ink-key presetting.

2) Is the system looking at consistency during the run, or is it comparing back to a standard/baseline?

3) Is the system calculating for every box in the colorbar, or is it averaging for each sheet measured? Achieving .03D variability across a sheet requires a small miracle and a sacrifice to the appropriate deities.


All kidding aside, you need to figure out the normal variability of the system. If you try to run tighter than that, then the pressmen, and the system, will begin to work against themselves. You can wind up in a rhythm of adjustment that creates greater variation.

 

[mglouis]

CLosed Loop Control Limits

CLosed Loop Control Limits

Hey there Rich.

1. Press runs are very short. When the operator has a “good” make-ready sheet, he tags it as “production”, starts the counter and runs the job.

2. Keys % out refers to the baseline. Consistency is also reported via standard deviation. Xrite reports evaluate keys that are turned on for sheets marked as production. Make-ready sheets and "off" keys are discarded. Pressmen are encouraged to turn off meaningless keys for which closed loop is not required.

3. Each patch is pass/fail on it's own merit.

Our newer equipment does very well, with % keys out on average at 10% with the older equipment at 49%. The average standard deviation for all presses ranges between 0.03 ~ 0.05, which tells me the SIDs on average are not too far out.

I've had 0.03 SID tolerances in place over a year and ping-pong overshooting does not jump out as an obvious problem.

I do not expect SID 0.03 across the sheet. I'm happy with 0.10 for KCM and 0.06 for yellow. My goal is to aim high so I'll like where we are when we fall short. The goal is to minimize the min/max range and maintain a low standard deviation, which defines how close all data is to the average.

I suppose I will get my answer to whether or not 0.03 is too tight by making a change to 0.05 and comparing the data back to the period before the change. I’ll compare SID % out but also standard deviation. I'll test my worst press and then my best press. The answer may depend on the press.

Thanks for your help.

 

[Cornishpastythighs]

We set our software for no closed loop adjustments less than +/-5 density points. Over correcting by the pressman was a big issue, pull a sheet-adjust, pull a sheet-adjust, since we installed our scanner and software we have less variation. Pull a sheet-scan-leave alone

 

[rich apollo]

Okay, Matt. I'm confused.

"Keys % out refers to the baseline." What is the baseline value? How is it determined?

"3. Each patch is pass/fail on it's own merit." So, each measurement is only compared to other measurements of the same patch?

"I do not expect SID 0.03 across the sheet. I'm happy with 0.10 for KCM and 0.06 for yellow." I don't understand how you can set your tolerance over a run lower than your tolerance for a single sheet.


The other thing to watch is your waste factor. From your description, you can expect that your waste factors will go up as you tighten that report tolerance. More sheets/time spent on make ready. This will also have a negative impact on your average production speed. More time spent on make ready is less time producing. You'll have to find that balance.

And yes, I agree, you may find discrepencies between presses. That would be an interesting study to conduct.

 

[mglouis]

Closed Loop Control Limits

Rich, I confused you on purpose because I'm lonely and want someone to talk to

1.
Targets and tolerances are set within target/tolerance libraries and the operator, if given access, may change these on the fly with each job. X-rite reports unfortunately do not indicate targets. The reports contain [Total Keys Keys Out Percent Out Mean Value Min Value Max Value Range Std Dev] for each computation type such as [density, dE, trap, etc.].

2.
If a job has one production sheet with 30 100K patches and 10 are turned off, then 20 patches will be reported for 100K-SID. If the target is 1.70 with an action limit of 0.03, and if 10 keys are between 1.68 ~ 1.73, the % out would be 50% (10 of 20). The report is for one job, containing many keys, which all share the same target.

With %-out, patches are compared to the target. With standard deviation patches are compared to the average.

3.
I miscommunicated. You can't set a tolerance for a sheet lower than for the job or vice versa. Changing the tolerance in the middle of a job would cause all sheets to be recalculated based on the current tolerance so sheet and job tolerancing are one and the same.
What I should have said is yellow in our shop is very stable so I have higher expectations for how tight we can hold it. We do not tolerance it differently than KCM but it performs 3~6 X's better in terms of standard deviation depending on the press. IOW, +/- 0.03 SID with a std. dev. of 0.02 actually happens with yellow fairly often. Ranges are higher with other inks.

Matt Louis