Lessons I have learned.....and am learning.

[Hilton Roberts]

Many of you know I am a strong advocate of some sort of process control on CMMs.

I have put together some of the lessons learned over the last 40 years and hope some of you will not experience some of the things I have when the machines don't deliver what you want or you question the results delivered.

* The machine may meet all the accuracy requirements of the manufacturer, and you may not be able to measure a part accurately.

* If the machine and lights in the lab are turned off at the end of the day, the machine will not be as repeatable as it would be if left on.

* The smartest programmers and the most experienced inspectors will not be able to convince manufacturing that good data is being provided to fix discrepant parts when features not being reworked do not repeat.

* Static repeat on the qualification sphere or cube has almost nothing in common with measuring a real part.

* Qualification of probes should be as automatic as possible.

* You need to have a test part to measure periodically to get a baseline on the machine so you will know when the machine begins to drift.

* You need to chart features that are representative of the features that you measure on production parts so that you can calculate control limits. This will help you to determine your measurement uncertainty.

* You cannot afford to be rejecting parts for being .0001 to .0002 in. out of tolerance if your measurement uncertainty at +/- 3 sigma is on the order of
.0005 to .0007 in.

* If your machine is not in control, the data you report is not as valuable as it could be in determining a course of corrective action.

* The little things will hurt you. If the temperature in the area where you use your cmm is not stable, computer compensation of that variable is of limited value.

* Drafts on the machine will destroy geometry.

* If drafts are present while qualifying tools, your results on every part measured with that tool qualification will be compromised.

* The beauty of process control is that charts will show that you have a problem.

* The beast of process control is that it will not always tell you WHERE the problem is.

That is why it is so important to have a part that you measure regularly that is NOT from the process stream.

If you measure a part from the process stream to validate whether you are in or out of control, the presence of out of control characterisitics can either be part of the manufacturing process or the measuring process.

If you do not know which is which, you may head down the yellow brick road to fix a problem in manufacturing that is in fact a problem in measuring.

Forty years of measuring things has taught me a lot. Chief among those things is the pride and ownership that manufacturing has on parts they produce. When those parts are rejected, for any reason, there is an ownership issue with the person or machine that present the part to us.

Sometimes it is an incorrect tool offset or wrong corner radius or cutter diameter that produces the discrepant part. These things occur in our line of work.

Those of us who are fortunate enough to work on these machines also have ownership of the process and pride that the results we report are valid and repeatable.

We owe it to our customers, both upstream and down, to report the true numbers of the features we measure and we must be certain enough to know the machines will repeat the numbers reported when we are tasked to re-measure a part.

I know this is long, but I really hope this helps.

Thanks for bearing with me on this.

Hilton

[DrBudman]

Nice

Very well said Hilton, printing out and posting for all to read.

Thanks

[\v/inston]

i dub this -> "HILTON'S LAW"

http://www.mmiclan.com/smilies/vb_congrats.gif

[tenndoc]

Here, Here - I second that emotion!

[B. Jacobs]

As for a newbie on CMM's, this has enlighted me. I didn't know there was that much to worry about. I do have one question though:

* If the machine and lights in the lab are turned off at the end of the day, the machine will not be as repeatable as it would be if left on.


Does this really make a difference. I turn off the computer and the machine daily when I leave. If that isn't a good idea, I will convince the higher ups to let me leave these things on.
Thanks

[JohnCTB]

I also shut my machine down at night and have never had a problem with it repeating. Why would it not repeat after being shut off?
JCTB

[Brian]

I read a previous post about your test part a couple years ago. When we got slow, I designed something that incorporated all the types of features we check on a 12x12 plate, cones, circles, spheres, lines, drafted walls etc...

We have been using this for over a year and a half now, and it is great to have, trend the results, it will tell you alot. You can look at the last years charts on one of our machines and you can tell the two months that the air conditioning went out to that room.

It is also useful to verify problems if you have the misfortune of crashing the head into your part which happened once.

I recommend everyone take your advice and come up with something like this.

Thanks for the advice
Brian

[STANLEY]

I have a master part that i check every morning, and i track the results on datapage. It keeps manufacturing off my back most of the time. I always tell the inspectors when your not sure of the results, check the master.

[kbotta]

Words of wisdom!
thanks for sharing.
Kev

[tking]

Thanks for taking the time to disseminate your wisdom.

T. King

[Hilton Roberts]

Radiant heating

I used to turn the lights off in the lab I had in Wisconsin. I was using a DEA 3305 that was 110 inches in X 60 inches in Y and 54 inches in Z. The frame on the Y minus side of the machine was a large steel weldment.

This room also had air conditioning put in by a residential/business contractor and was not designed for lab use. The air conditioning vents ran around the ceiling of the lab just below the lights. The air return for the lab was on the Y minus side of the machine and we were in Wisconsin. We had halogen lights in the lab. The air supply for the lab was located on the Y minus side of the lab. In the winter, when the AC called for heat, the warmest air came out of the Y minus air vents and cooled off by the time it got to the Y plus side.

In the summer, the system called for cold, and the coldest air came out of the Y minus side and and warmed up by the time it got to the Y plus side of the machine.

The DEA, by design had the drive motor for X axis on the Y plus side of the machine and the Y minus side was "free."

In the winter, we had 75 degree air coming out of the vent and in the summer, we probably had 40 degree air coming out of the Y minus vent all in the attempt to control the lab to 68 degrees.

What that means is the Z axis ram was seeing about 35 degrees of temperature variation times the ram length of 54 inches times the coefficient of expansion of steel of 6.5 microinches per degree. This amounts to around
.014 inch. The Y plus side saw significant change but less in magnitude.

What I did was to take a two inch cylindrical ring gage and place it in a fixed position at the X plus Y minus end of the machine surface plate and at the end of the day, I found and set X/Y zero in the center of the ring gage.

I turned the lights off and went home and left the machine on.

The next morning ( 5:30 AM), I went in and remeasured the the ring gage before turning the lights on, and found the center had changed by .0008 in. in X and about .0013 in. in Y.

I turned the lights on, and began measuring the ring gage over and over again.

As soon as the lights came on, the AC began calling for cold. As I plotted the X/Y coordinates of the ring gage, I could see the results of the AC overshooting the set point and begin calling for heat. The change was immediate and the first chartable point occurred within one minute of turning the lights on!

As the AC continued to overshoot and call for cold and overshoot and call for heat, the X/Y coordinates of the ring gage described a sine wave of decreasing amplitude until around 10 am. The machine then settled within a tenth or two of the previous day. What was happening was the changing the heat load was raising the bridge and rotating about the pivot point on the drive side and by raising the bridge, the Y coordinate was changing. The X changed also but not as much.

We checked very large diesel engine heads there. Those things weighed about 2500 lbs each.

The staging area for the lab was on the X plus end of the machine. The castings were often out in the open until a kid or two of them were moved into the DEA room for measurement. I reasoned that if the airconditioning could do that to the machine, what would happen if a couple of skids of heads that had been sitting in the cold, were brought into the lab and placed in front of the machine. Sure enough, those heads cooled the X plus end of the bridge enough to bend the bridge toward the X minus end of the machine and if one remeasured the 2 inch ring gage, the center coordinates of the ring shifted along X axis.

We finally got to the point we would not allow parts to be brought into the lab until we had established the center of the ring gage. We would then bring the parts in and allow them to soak and we would measure them until the cmm had restabilized and the parts were normalized.

We had to soak the parts in the lab as the heat in the shop in the winter was next to non-existent.

It is truly amazing what you will see with a few simple charts when you are trying to find out what really goes on when you measure parts.

Radiant heat is a significant variable and the more you can reduce variation, the better off you will be. If the DEA was so sensitive to changes in the heat load with a steel bridge, that is something one should consider with a machine with an aluminum bridge.

It is this kind of basic approach to measuring that will enhance the performance of your machine. I know of no reputable metrology lab that EVER turns off the lights and AC on the weekends or after work. It takes too long to get things back in synch.

Be very careful of placing lights under your bridge so you can see your parts better.....the heat....and there IS heat there, will cause you problems.

If you do not chart anything, you may not find out just how much your machine is changing until you have a downstream customer in your face because something is not right with the parts you blessed.

Hope this helps.

You can give me a call anytime at 480-891-7263

Hilton

[tlnzz]

"Hilton's Law". I like that.

Thank you for sharing your insight Hilton.

[randy z]

Excellent

[dfiola66]

Great advice Hilton....a question to all:

To those that have designed a master, would you consider sharing the solid model file with anyone? I'd love machine one and use it. I could have our cad dept. do it, but that'd take FFFOOOOORRREEEEVVVVEEERRRRR!

thanks.

[NZ CMM Girl]

...And those that livest by Hiltons Law flourished... for the law was good...

Seriously tho Hilton... I really appreciate the benefit of your experience - because nothing beats just doing it and learning from experience. Thanks for sharing yours.

Lenice