I am new to the forum so please forgive me if this subject has been discussed in the past. I did a search and found some partial answers, but I am curious if anyone else has run into this situation. I have recently started working for a company makes components with turned and bored diameters, ID and OD, with tolerances as close as +/-.0003”. Most of the gaging used for tolerances up to +/-.0010” have a resolution of .0001”.


The manufacturing department has operators plor one or two Key characteristic data 100% on a Run chart. Most of the lots are small and/or short runs. In many cases the data is used for final inspection.

The quality department wants to use SPC charts to monitor process control. I have reviewed historical data and found that most of the data collected has very little variation (no more than +/-.0002” in most cases). My investigation into gage R&R’s finds that a significant portion of the variation is due to measure system error. It appears that our machining processes have less variation than the gages are capable of measuring. And no, the company will not purchase gages with better resolution.

This brings me to two questions.

Would you still use a run chart or record the actual results on an inspection sheet?

Does anyone have a suggestion for process control parameters for turning, CNC milling and grinding instead of the current method of selecting tight tolerance print dimensions to establish process control?

I am new to the forum so please forgive me if this subject has been discussed in the past. I did a search and found some partial answers, but I am curious if anyone else has run into this situation. I have recently started working for a company makes components with turned and bored diameters, ID and OD, with tolerances as close as +/-.0003”. Most of the gaging used for tolerances up to +/-.0010” have a resolution of .0001”.


The manufacturing department has operators plor one or two Key characteristic data 100% on a Run chart. Most of the lots are small and/or short runs. In many cases the data is used for final inspection.

The quality department wants to use SPC charts to monitor process control. I have reviewed historical data and found that most of the data collected has very little variation (no more than +/-.0002” in most cases). My investigation into gage R&R’s finds that a significant portion of the variation is due to measure system error. It appears that our machining processes have less variation than the gages are capable of measuring. And no, the company will not purchase gages with better resolution.

This brings me to two questions.

Would you still use a run chart or record the actual results on an inspection sheet?

Does anyone have a suggestion for process control parameters for turning, CNC milling and grinding instead of the current method of selecting tight tolerance print dimensions to establish process control?

Hello John and welcome to The Cove. :)

Quote: "My investigation into gage R&R’s finds that a significant portion of the variation is due to measure system error. It appears that our machining processes have less variation than the gages are capable of measuring."

A few comments: If this is indeed the case, then all measurement data would be highly suspicious. What would be the use of the continuation of data collection and charting? How did you determine that the process variation is less then +/- .0002"? Did you use the same measurement system? I guess you're faced with an interesting dilemma. The suggestion that I was going to make (regarding your measurement equipment) will not fly. No, I do not have suggestions. :nope:

I hope my Fellow Covers will comment.

Stijloor.

Hello, John! A big welcome to the Cove!:bigwave:

Stijloor has adequately captured the essence of this.

You state you are manufacturing products to +/-.0003, yet the best accuracy of your standards is +/-.0010. Something does not add up here.
How did you determine that .001 accuracy? What kind of standards are these?

The accuracy value of your standards is your best case accuracy. It will be difficult to measure accuracies better than that.
*******
Now, you may have the calibration accuracy of +/-.0010, but you instrument may produce a much tighter tolerance than that. Assume a bell curve, and .0010 is like 99.9% extreme. Thus, you might could assume your standards are like .0005 99% of the time. But... this is a really long, risky way around doing the job right the first time.
***********
I understand your organization may not want to buy new standards. But they have no basis to state they are making parts to +/-.0003.

I hate to sound as if I'm avoiding your topic. But the true error of your measurement system is an important essential component.

Brad,

Perhaps I did not state my problem correctly. Our standards are ring gages and gage blocks with accuracy measured in millionths. What I did say is that a lot of our parts have a print specification of +/-.0010" with some of the parts having a specification as tight as +/-.0003”.

An operator using a mechanical bore gage with an indicated resolution of .0001" has a very difficult time measuring variation on a bore with a +/-.0003” tolerance. We have limited availability (due to the cost) of air gages that have resolution to 50 millionths. When an operator checks the same parts using an air gage we find that the process variation less than +/-.0002”. A Gage R&R on the air gage produces excellent results, less than .0001” uncertainty. Given the findings using an air gage I feel confident in stating that the process variation is within .0002”.

As to Stijloor’s comment about the measurement data being highly suspicious, I agree. However mechanical bore gages are the “norm” in the short/small run environment of our industry. It is not economically possible to justify better gages. The reject rate, which is almost non-existent for tight tolerance bores, does not support the purchase of better gages.

Going back to my original questions, would you have an operator simply write the size on a data sheet or would you plot the data on a chart as a visual reference only.

As to my second question, I would like any suggestions for process control and improvement of precision machining operations on CNC lathes and CNC mills. Since these machines are set-up for small runs all the time, Is there any parameter(s) that you would suggest that I monitor?

Thanks Brad and Stijloor for your input

John

Brad,

...
An operator using a mechanical bore gage with an indicated resolution of .0001" has a very difficult time measuring variation on a bore with a +/-.0003” tolerance. We have limited availability (due to the cost) of air gages that have resolution to 50 millionths. When an operator checks the same parts using an air gage we find that the process variation less than +/-.0002”. A Gage R&R on the air gage produces excellent results, less than .0001” uncertainty. Given the findings using an air gage I feel confident in stating that the process variation is within .0002”.

As to Stijloor’s comment about the measurement data being highly suspicious, I agree. However mechanical bore gages are the “norm” in the short/small run environment of our industry. It is not economically possible to justify better gages. The reject rate, which is almost non-existent for tight tolerance bores, does not support the purchase of better gages.

Going back to my original questions, would you have an operator simply write the size on a data sheet or would you plot the data on a chart as a visual reference only.

As to my second question, I would like any suggestions for process control and improvement of precision machining operations on CNC lathes and CNC mills. Since these machines are set-up for small runs all the time, Is there any parameter(s) that you would suggest that I monitor?

Thanks Brad and Stijloor for your input

John

John,
What is the purpose of your charts? Remember, the measurement is part of the process. If the chart is to detect any change in a process which is capable and in control, I'm not sure you have a problem. If your parts have been in spec, and you haven't had customer complaints, then why do you need more precise measurements? I have used bore gages on similar parts, and yes, the gage variation may be a significant part of your observed process variation. So what? Just make sure that your operators do not adjust the process unless a definite out of control signal is received, and the process should be ok.

As to short runs, I have found that if you plot IDs and ODs as deviatin from nominal, instead of actual size, you can reliably continue charts across similar parts with different sizes, and detect process problems quite well.

Geoff Withnell

Geoff,

The purpose of the Run Charts is twofold. Manufacturing uses them to monitor the operators. I would like to use the data for process capability analysis.

I have found that lack of gage resolution produces a histogram with too few cell intervals. This usually causes the software that I use to issue a warning that there is a lack of curve fit. Also the software generates a Cpk for both a normal curve and a fitted curve. The resultant Cpk’s can be significantly different when data only varies by 3 or 4 gage increments. I don’t feel that these results are accurate.

I was thinking about using a higher resolution gage to calculate control limits and process capability. I would let the operator use the regular gage for “real time” monitoring, but use the control limits obtained with the higher resolution gage.

Any thoughts?


P.S. I would have the operators use an Individuals and moving range chart instead of the run chart.

Thanks,

John