We are considering a shop floor CMM, any experience?

We are an aerospace job shop of about 25 machines mostly 4x and 5x mills and 8 lathes. We make small and large parts but these would be for the small of course. Mix of AL and steel. Our big CMM is a real bottle neck. We are looking at a couple small shop floor CMM's. Any of you use these machines? If so, are they operator friendly? I am unfamiliar with that type of programming. However, I run Mastercam daily, and I'm generally capable in several programing languages. Once we are trained up, how long would it take for a guy to walk up and get a part run through?

We have a hexagon sf 7-10-7. software is pcdmis. If the part is already programmed, then it's pretty quick. Load program. Change probe. Qualify probe. Run.. Maybe 10 min tops? Probably less with a probe rack.

That's assuming the "guy" knows the machine.

Walk up to it with no program, then it's going to be awhile....



Sent from my SM-G900R4 using Tapatalk

We have several on the shop floor. In the high volume production areas, they are programmed by the inspection department since these are series production parts. In another department, more like a job shop, they may only run a few pieces of any particular design. The operators are trained in programming them and they are pretty efficient at it. Depending on the complexity of the part and features being measured, 5 minutes to an hour to write a program. All of ours are Ziess with Calypso software. Calypso is pretty quick and intuitive and you can import your 3 models to it to do the programming from.

Second everything Tony said. Depending on if you only need 3 axis or full 5+, you can save money and increase shop floor durability by going with something like a duramaxx from Zeiss, that's what they're made for.

For us there is one manual CMM on the floor and 2 CNC CMM's in QC. For us, there aren't any button pushers in the shop, there are Machinists that are greener than others is all. Meaning if this isn't a new employee's career path, we don't invest much and they usually leave, but that's a different topic. The only reason I say it is that we run mostly low volume, and the "Machinists" only need to verify one or two features of a whole part on a CMM, before they feel comfortable submitting them to QC. So verifying Position or something like that. But it works well for our needs. With that said, we also have a Fab department and a Grinding department so air quality is a factor to question when having a CNC CMM outside QC.

R

TeachMePlease said:

Second everything Tony said. Depending on if you only need 3 axis or full 5+, you can save money and increase shop floor durability by going with something like a duramaxx from Zeiss, that's what they're made for.

Click to expand...

Yes, we have several duramaxx machines and they are very good for use on the shop floor. Ours have been on the floor for several years with no issues.

Some people advocate online inspection on the CNC machine itself using probes.
Some of the arguments are:
Inspection takes much less time, compared to that done on a separate CMM because there is no transportation/setup time.
Inspection results are immediately available. Hence, corrective measures can be taken without scrapping more parts.
If rework is needed, it has better chances of success in the same setup.

On the other hand some people argue that one should not inspect a part on the same machine that made it, and inspecting on the machine wastes production time. Environmental issues are also there.

Looks like most people prefer offline inspection on CMM over online inspection on the machine.

Is a probe in a spindle generally accepted as less accurate than a cmm? Of course the mechanics are different, how about the software? We have had repeat issues in our cmm room. Same part measured twice with slightly different reports.

If these are repetitive or 'stock' parts, then you might look at Renishaw. They have something new out. It is sort of like a CMM. Its a hexpod with a probe. It basically compares a part to a sample part. The price is a fraction of a CMM.

lowCountryCamo said:

Is a probe in a spindle generally accepted as less accurate than a cmm? Of course the mechanics are different, how about the software?

Click to expand...

If the machine has repeatability error while machining, isn;t it safe to ass-u-me that the in machine inspection is going to have the same error? Personally it isn't my luxury to agree or disagree whew.

lowCountryCamo said:

We have had repeat issues in our cmm room. Same part measured twice with slightly different reports.

Click to expand...

Welcome to the wild world off CMM inspection, a really good CMM operator knows that a CMM is NOT infallible. And the opposite can be said. A CMM is a very useful tool!! But they are not perfect. IMO if you can get a CMM to repeat reports on the same part, with the same probing approach within .0001" that is amazing!! .0003" is normal for an identical repeat. .0005" something wrong with the equipment.

R

One way to check the repeatability of the machine is to periodically probe a standard part and watch the deviation.

lowCountryCamo said:

Is a probe in a spindle generally accepted as less accurate than a cmm? Of course the mechanics are different, how about the software? We have had repeat issues in our cmm room. Same part measured twice with slightly different reports.

Click to expand...

As noted by others, using the same device to manufacture and measure will incur the same liabilities with respect to errors present in the device. That is a rationale for not using the VMC as an inspection tool on the part just made. Obviously, this is a judgment call based on your parts and the tolerances you need to work within for inspection purposes.

CMMs have the same issues as any other machine tool when it comes to setup anomalies and thermal differences, although generally at a somewhat lower level. My experience with a Zeiss Eclipse machine (late 90's) is that even under DCC operation, there can always be variations in the values being determined for features when you are resolving to 0.00001" increments.

The expectation that a CMM is "perfection in measurement" down to 0.0001" resolution is a fantasy in practical terms; once you start getting to a need for <.0003" repeatability (as Litlerob1 notes), you need to be quite rigorous in measurement protocols and temperature controls, as well as regular self-checking and calibration of the machine. The Zeiss I mentioned above did an excellent job as I saw it, and the results were so often dependent on factors other than the machine itself, not least of which is thought on the part of the operator, as well as several measurements to confirm and average the results on a high-precision measurement.

The problem with the repeatability on the 0.00002 scale doesn't have as much to do with the repeatability of the machine as much as it has to do with time. That difference is basically 0.3 msec at 100 mm/min probing feed. The CMM's computer bus interrupt timing can cause that much variation in your measurement.

I'd look into a romer portable...really maximizes usage.

Tonytn36 said:

The problem with the repeatability on the 0.00002 scale doesn't have as much to do with the repeatability of the machine as much as it has to do with time. That difference is basically 0.3 msec at 100 mm/min probing feed. The CMM's computer bus interrupt timing can cause that much variation in your measurement.

Click to expand...

Tony could you explain that for my stupid head please? Are you saying the amount of time it takes the probe to make the contact approach to the actual part causes that much error? So if the part alignment isn't perfect, and you are inspecting a diameter (for example) the probe goes to one side and it takes it XX time and the other side takes XXXX time, that creates deviation?

lowCountryCamo said:

Is a probe in a spindle generally accepted as less accurate than a cmm? Of course the mechanics are different, how about the software? We have had repeat issues in our cmm room. Same part measured twice with slightly different reports.

Click to expand...

As a rule of thumb I'd say a probe in a spindle is 5-10X less accurate than a cmm.
Depends on your machine tool and your cmm.

Do not confuse accuracy and repeatability. Different animals.
And then linear accuracy and volumetric accuracy, one typically 3-9x worse than the other.
It is possible to make a probe in spindle repeat as well as a cmm with the same feedback resolution but you have to do your homework.
It will not be accurate and machine tools use backlash comp which screws all this measuring stuff up if you probe from two directions.
Even with glass scales mean ole mister Abbe comes to visit.

The repeatability Tony refers to is a latency issue (although some mechanics in the probe are involved).
You touch a point and break a switch while moving. This signal runs at the speed of light through your wire to the computer, some lag here.
Then the computer has to see your signal and quickly read the number off of the moving scale. More lag here.
So by the time you read the scale the probe is past the position where the touch occurred.
If you do this with an interrupt the amount of time to service this request is variable so you get bouncy numbers.
Sometimes the cpu gets there fast, other times not so fast. Jitter in the real time world.
Often in a CMM design you will hardwire a position latch to the probe so the scale reading is captured and held by hardware circuity but even this can only repeat with +/- 2 counts. Perhaps worse if you are using sub-dividing circuitry in between (variable lag depending on clock synchronization) and most do use this circuit in between to get the resolution users want.

Like a part draw in cad it all looks easy and perfect from the outside. Not so much when you have to build it and make it work.
From the floor side it gets stranger still if you have 20 cmms.
Some say good, others say no and operators know which ones agree with their floor gauges.

All that said I'm a big fan of cmms on the floor at every machine group.
Programming is programming, not hard once you learn it and should be quick.
Time to check depends on what you need to measure.
You may have some arguments between your floor machine and the big guy and have to figure out how to deal with this dispute.
Ziess is sort of the gold standard as they do build some of the best high end but many options just as good in the workhorse side.

The nice thing about a cmm or other automated gage is that it has no bias towards a good vs bad part. You eliminate this built in human tendency from your measurements.
We all want to make good parts and consciously or subconsciously we will all bias our numbers towards zero, computers do not care.
Bob

litlerob1 said:

Tony could you explain that for my stupid head please? Are you saying the amount of time it takes the probe to make the contact approach to the actual part causes that much error? So if the part alignment isn't perfect, and you are inspecting a diameter (for example) the probe goes to one side and it takes it XX time and the other side takes XXXX time, that creates deviation?

Click to expand...

Read Bob's post..... he explains it quite well. What I was saying is that even if you do a repeat measurement without moving the part, you can get differences in the readout based on the lag time of the circuitry and computer actually doing the measurement. The computer can't be doing two things at exactly the same time. It is doing all of the work to actually move the stylus and tracking position, etc and then this signal comes along that says "stop and read the scale". Even if the lag in the circuitry is very repeatable (not so on this scale as mentioned above by Bob), the CPU might be doing something else important at that instant - like checking position and it takes another clock tick or two for it to actually "see" the stop command. During that clock tick or two delay, the axis is still moving, so when it reads the scale it's "wrong" by a couple crystal vibrations. The faster you run the probe to the part, the worse the repeatability will be because you are covering more ground per crystal vibration.

Now, modern computers run at quite the frequency and this is a very small amount of lag, but when you add it to the circuitry lag and variation, the software and operating system lag and variation and the mechanical lag and variation, then you get measurements that seemingly don't repeat.

And to explain what an Interrupt is..... (very, very simplified). If you write a simple computer program, it will execute from top to bottom. During the time this software is executing, that is all the CPU is doing is executing this code block. That is it's job. If you want it to pause executing this block of code to do something else that is more important, you need an "interrupt" to break (pause) the code execution, basically a signal that the CPU interprets as "wait a sec, gotta to do this for a min, it's more important, be right back".

CarbideBob said:

It will not be accurate and machine tools use backlash comp which screws all this measuring stuff up if you probe from two directions.

Click to expand...

Why don't we have similar issue with CMMs? The mechanism might be different.

I played with a Keyence XM near infrared based CMM and was impressed with the simple user interface. Worth a free in house demo to see if this meets your needs. Seemed better than a Romer or Faro arm.

I would also throw in a vote for a Renishaw Equator. The are very inexpensive compared to a full-blown CMM. They are very fast, and they are designed to be used on a shop floor. Some applications even run closed-loop with the CNC machine to fine-tune tool paths/positions to keep parts in spec.

The one thing to realize is that the Equator is really a "comparative" measuring machine, not an "absolute" measuring machine like a CMM. Once you wrap your head around how they work and what they can do, you might find this to be a really good solution for the shop floor.

PM