With all of the advancements made over the years, it seems as though the manufacturers have decided that everyone is interested in but one thing: speed. It almost appears as though we have overcomplicated some of the more simple, or seemingly so, tasks that are still in need. Over the past 45 days, I have done considerable research and it seems as though everyone wants me to spend many dollars to retrofit my Sheffield 1808 MEA just to do what I happen to think is a relatively simple process. All I want to be able to do is to manually scan (I’m not concerned about speed and how long it will take), the internal dimensions of a part so that I can fulfill my reverse engineering requirements. This part is relatively small, about 2' 2.5" diameter transitioning from a round part to a rectangular piece. I read about the "expensive" dedicated analogue scanning probes available but they cannot be used with a manual CMM. Would someone please enlighten me as to how we have come so far technologically as to complicate what should be simple. I say this because I worked with a gentleman about 10 years ago who was doing this very thing with a manual CMM. Unfortunately, he has since passed away and no one else seems to be able to tell me how to do this. Is there anyone able to assist me in what I feel is a relatively simple task without spending a minimum of $15K?
Larry
Send to friendSign In to get started!
Comments
epb 6/8/2004
Variable data obviously gives more information that could be used to control or see the process BUT can a variable gage tell you everything you need to know about the bolt-hole being acceptable or not-acceptable?
I would say NO.
A plug gage will tell you some things the bore gage will not tell you. The bolt hole may be correct on diameter but may have incorrect cylindricity. The bore gage would not tell you this while a plug gage would. The plug gage would perform a functional check of the bore to determine a drift in the diameter.
Conversely, the bore gage will find some things that the functional plug gage will not. The plug gage GO/NOGO could tell you that the bore is correct when in actuality it is only correct at the entry points but there may be some oversize condition between the end points.
In critical applications it is recommended to use BOTH variable and Functional gages on a bore to determine acceptance.
mkomarmy 4/27/2004
The previous replies all make good points. The measurement process can be done for a variety of reasons. Assuming the gauge is adequate for the job, you may be fortunate and have a capable process! If that is the case, you are measuring parts to insure the process has not changed and the quality is stable. Are you currently sampling or inspecting parts 100%? The goal of sampling should not be to just detect defects, but to identify process changes so corrections can be made prior to producing excessive defects. This means that some criteria other than specification limits are used to identify process changes. Some of the tools available are SPC control charts, precontrol, warning limits, etc. The method you use will depend on the sample size you can afford or are willing to take, and the capability of your process (determines the allowable shift of your process). Another factor is how well do you know your process? Process knowledge is key to a good control plan. If all of your 5 holes are machined with one tool on a CNC machine, the sampling requirements may be different than a machine which uses a multi tool head to machine them. As you understand the sources of variation and improve or eliminate them, the frequency of your checks can be reduced.
mkomarmy 4/27/2004
I almost forgot to answer the original question! If your goal is to monitor the process, attribute data is not very sensitive. For example, a 1.0 Cpk process has a defect rate that most producers do not accept (approximately 2700PPM). To monitor a defect rate such as this, you should select a sample size large enough to observe at least 2-3 defects per sample on average. For this defect rate the sample size would be over 1000 parts. This is not practical. With the proper use of variable data and statistical methods, the process can be monitored effectively with sample sizes as small as 1-5 parts.
qdigest 4/16/2004
To provide you with a good answer, there are a number of questions you need to look into.
1) Do you ever find parts out of spec during your in-process checks? If so, what is the percentage?
2) Do you experience customer complaints/returns due to hole size?
3) You state that the data fluctuates very little. Have you conducted a GR&R to quantify measurement sysytem error? The lack of differences in the data values could be due to lack of discrimination of the gage and/or measurement system. The minimum resolution of the gage should be 10% of the tolerance. (i.e. if the tolerance is .001, the gage should read in no more than .0001 increments).
4) Is the bore gage design adequate for the hole configuration? If the hole is oval a gage with 3-point contact is not suitable, or if the hole is 3-lobed a gage with 2-point contact is also not suitable. If this is the case, it could also explain lack of variation in data measurement values.
As a general rule, variables data provides much more information than attribute data. Many variables control charts use a subgroup size of 5, however for p and np charts the sample size is determined by np>5.
bobdoering 4/15/2004
Good point, what is he going to do with it? If he is using it to determine tool wear and tool change frequency or capability, then variable is the way to go. If you just want a process report card chart, then an attribute chart with go/no go will do. The charts should be used to make decisions. At this point, what is the decisions being made with the chart?
Bob Doering