MSA for a self contained controller

Regarding a device is used to measure AND control a critical characteristic in a process.
Example 1
A heat chamber with a control to select the temperature and a built-in controller to maintain the temperature at the manufacturer's specified tolerance.
The temperature in the chamber is identified as a critical process characteristic on the control plan.
The manufacturer's specified tolerance is much less than the specified process tolerance.
Q1 Is an MSA study required on the temp measuring system of the heat chamber because it's measuring a process characteristic on the control plan?

There must be plenty of situations where purchased equipment includes gages or control devices for key process characteristics.
Q2 Are we expected to do MSA (GR,etc) on these imbedded measuring systems? (Surely not? Right?…)

Example 2
A temperature controller device is added to a heat chamber with only a heating element.
The temperature in the chamber is identified as a critical process characteristic on the control plan.
The manufacturer's specified tolerance for the controller is much less than the specified process tolerance.
Q3 For this one I won't ask IF an MSA study is required, but what kind of statistical evaluation is appropriate?

I present the following considerations and my proposed approach.

For systems used for process control, the key points in MSA (bias, discrimination,repeatability, stability,etc) are all focused on whether the system is suitable for use in controlling the process.

My contention is that for a system which IS a process controller, the key points are:
1.Can it control the process
If the process is in control, then it must be capable of controlling the process which is the intended purpose of the system
2. Is it accurate
When calibrated the individual measurements have a known relationship to a standard therefore the measurements can be relied upon to correctly indicate the process condition

3. Is it stable
If the process stays in control and on target (within limits) then the system must be stable

My proposed MSA approach is to control chart the indicated measurement from the controller showing the process is in control and certify it's calibration.
The rational is that if the process is in control, stable, and on target, then the system must have the necessary statistical properties as detailed in the MSA manual

Q4 Will this approach fly?
Q5 Are there any suggestions other than just treating these as individual measurement systems?
Q6 Should I go back to flipping burgers?
Q7 What does Bobby Knight have to do with this?

I appreciate any feedback.
mikeb



[This message has been edited by mbruner (edited 04 May 2000).]

[Jerry Eldred]

I'll preface my response that I am a metrology engineer with extensive background in measurement systems, calibration methods and techniques, etc. BUT... I am not a MSA guru. So take the MSA requirements portion of my answer in that context.

My understanding of 'measurement tools' used in a process is that any measurement tool used to measure any quantitative characteristic of your end product must comply with all the requirements of an MSA per QS9000, etc.

However, what is making me scratch my head is calling a temperature chamber a 'measurement tool'. If you call it a measurement tool, that implies that it is a final gage of some product parameter. There is a difference between tools used to measure product parameters, and production equipment used to produce the product.

Most certainly a temperature chamber used to bake or cure or perform a critical process step must be properly maintained, including some version of calibration (some like to lean toward the side of doing a PM - I favor leaning on the side of calibration). And there is also an implied need to profile the chamber. MIL-STD-883 had a good method. But your quality engineering people need to be involved in whatever profiling method you use (should you decide to go that route).

But as for an MSA. You need to correspond upper and lower control limits of a process parameter with upper and lower temp setting limits. If there is a measurement tool of some sort that measures the parameter created by the process in your temperature chamber, there should certainly be an MSA on that measurement tool. And an adequate calibration procedure (including chamber profiling in my view), should be created and followed to periodically calibrate the chamber, at appropriate interval. But I struggle with doing MSA on production equipment.

If the temp conroller specs on the temperature chamber are not good enough, I recommend something such as a data logger with higher accuracy thermoocuple inputs to constantly monitor the chamber temperature. Or a strip chart recorder with adequate accuracy to monitor and document that the chamber stayed within acceptable limits for the duration of the process. Connect the data logger or strip chart recorder to your computer network, or to a stand alone alarm (which ever is most applicable), to indicate out of spec chamber conditions.

What is the output parameter of the temperature chamber? Does the temperature chamber make a measurement of the product? Product specified parameters, and process parameters are two different things. Process parameters are controlled. Product specified parameters are measured. The temperature chamber is a process tool, not a product measurement tool (unless there is something here that I am missing - very possible).

It is common to install a network of thermocouple probes at the corners of the temp chamber, and one near the center as external monitors. Monitor and use them to assure that the temp chamber stays in specs.

Do an MSA of whatever measurement tool measures the output parameter of the chamber. My question keeps coming back to - how do you know if your product is in or out of manufacturing control limits (not the process, the product).

Feel free to email me if I can help any further. Hope this was of some help.

------------------

Jerry,
Thanks for the feedback. Sorry I was so long in responding.

This was a hypothetical system presented to simplify the issues related to a number of different complex sub-systems in our manufacturing plant.
This is referring to production equipment and process parameters NOT product parameters. The temperature chamber does not make a measurement of the product.

Your statement
"But I struggle with doing MSA on production equipment."
indicates that you don't normally do MSA on production equipment.Is this true?

Regarding your statement:
"Product specified parameters, and process parameters are two different things. Process parameters are controlled. Product specified parameters are measured. The temperature chamber is a process tool, not a product measurement tool ..."

Our understanding of QS requirements is that MSA must be done on production (process) equipment if that equipment is used to measure a process parameter which has been judged to be a "key" process chracteristic. That is, one that has a significant impact on product quality. We routinely do MSA on production equip used to measure such process parameters.

I was hoping to make a distinction between evaluating how well a system measures a value(GRR) versus evaluating how well a system controls to a value (calibration?).

I don't want to have to do MSA (including GRR,stability,linearity,etc) on sub-systems in our process equipment.

[Jerry Eldred]

My understanding within my industry (semiconductors) is that there is a dividing line between equipment used to manufacture the product and equipment used to measure the product.

To be even more confusing, there is yet another distinction between those production equipment which perform a process step for which there either are, or are not measurable parameters at the output of the process step.

You demonstrate the ability to adequately control the product parameters to within acceptable cpK, etc through MSA. If in a given circumstance, there is a production manufacturing process step that does 'X' to the product, and if there is a measurement tool to measure 'X'. The tool that measures 'X' is the one that requires MSA. The production tool that performed 'X' (to my understanding) does not require 'X'. You perform maintenance and/or calibrations to specified procedures, etc. to Production tool that performs 'X', but not MSA. You demonstrate the ability to assure that parameter is controlled in your production process not be MSA on the production tool, but through MSA on the measurement tool.

For example, you have a stamping machine (forgive me, I don't work in mechanical mfg environment). The stamping machine is set to a certain pressure, downward speed, X, Y, and Z coordinates, and has a fixture on it that chops out a part of X x Y dimensions, and it makes what we will call a YUTZ.

You verify that samping machine is chopping your YUTZ to accurate repeatable X x Y dimensions by measuring those dimensions on a big micrometer (hypohtetical). That micrometer must be of adequate repeatability and reproducibility to be sure that the YUTZ meets its X x Y dimensional specs.

In semiocnductors, we maintain the actual production tools through pretty rigorous PM and calibrations. We have metrology tools that measure things like film thickness, line widths, step heights, etc. We do MSAs on all those measurement tools. The MSAs done on the metrology tools give us acceptable confidence levels (cPK) that our process is maintained in proper control.

We have many, many temp controllers in many, many applications. Applications for example such as controlling diffusion furnace temperatures. We don't do MSAs on those temp controllers. They are maintained by trained maintenance techs and engineers.

But when the product comes out of the furnace, there is a measure done of the output parameter of the process step. And in theory, I understand that what QS9000 wants is for us to be able to prove that we have an acceptable degree of confidence that the product meets its specs (and within acceptable limits of error).

My theoretical mind looks at doing an MSA of a production tool. Using the diffusion furnace example. If I attempted to do an MSA on the furnace, it wouldn't tell me that the product met its specs. It would tell me that the furnace met its specs. So I would still be left needing to determine how well the product met its specs as well. Meaning that I would end up doing two MSAs for the same parameter, where only one was needed.

Coming back (after all of my rambling) to your original details. The distinction in question is process control parameter versus product specified measurable parameter. In the end, it is the product measurable parameter that matters. The ability of a production tool is in the finality of things, the ability to assure that the product meets its parameter.

If you do a good MSA on the measurmeent tool that measures the product parameter at the output of a process step, that tells you only how adequate your measurement tool and/or method is to monitor the porcess.

Subsequently, knowing how 'capable' that measurement tool is to measure the product parameter, and using it to monitor your production process step, assures that the product parameter is in control limits, and subsequently, that the production tool is performing its process step capably enough.

I don't know if I've muddied the waters further, or helped.

------------------