NC Costs?

About 5K per machine plus a seat of PCDMIS and a seat of NC.

You shouldnt measure on machine tools... (spoken by Arnold)

About 5K per machine plus a seat of PCDMIS and a seat of NC.

You shouldnt measure on machine tools... (spoken by Arnold)

What does that mean? $5,000 + the cost of PC/DMIS + the cost of NC?

Or is NC about 5K?

Do I need a separate seat of PC/DMIS even though I have PC/DMIS on my CMM computer? Can I put the NC module on the CMM computer and output the g-codes to a file to be downloaded to the NC machine?

You should check with a sales person. But...

It was around 5K per Machine tool that you were going to run it on.
Plus around 5K to add it to a seat of PCDMIS
But to get the benefit, you should buy an additional seat of PCDMIS instead of loading on top of your already busy CMM. I think that you post the data downstream and evaluate it on the seat of PCDMIS.

To me the only reason to do this would be if you have parts that you need to check but dont want to break the setup yet. Maybe some intermediate checks(not final checks)

You shouldn't measure on machine tools... (spoken by Arnold)

Dream on brother.... You are so far from the truth here.... Sorry...

I agree somewhat, though: be VERY careful before you start to think about NC. Are your machines calibrated? Are you willing to spend money on artifacts? There are many costs to consider.

But I use it every day. I have 2 machines up and running. End of the year it must be 6. We intend to run more than a dozen within 2 years.

Yes, PC-DMIS converts its code to Gcode. And it does it quite reliably. Only problem is something to do with coordinate rotations. They have not been able to explain that one to me yet. I added a piece of sample code for your enjoyment (for ethernet). It is generated totally automatically. Nothing you have to do about it...

After the program is finished, Server sucks in the data and Executes, just like as if you were running PC-DMIS and executing a program. And you get a true measuring report as a result, with all powerful features from PC-DMIS.

The biggest advantage is that NC offers you the possibility of bringing SPC right to the machine tool, while the part is still fixtured. So fixing issues at that point is a breeze. No time delays; instant gratification. But make sure you use accurate probes...

The way it works is that you need to run a product called "PC-DMIS NC Server". It wants to run on its own computer. So you'll need at least 1 additional off-line license. Server can serve many machine (for additional cost of course).

You can run using RS232 lines. Better is Ethernet. I am upgrading my machines to Ethernet because it allows for bi-directional parameter upgrade.

Contrary to popular believe, you do NOT need 1 PC-DMIS license and computer per station. You do need enough licenses on server though....

Also, again, you'll need an ACCURATE probe on your machine. I use the MP700. Lots of good luck with it.


If you want more info on it, send me a PM and we'll discuss off-line.



Jan.

jan,
what were you referring to? I actually qualified that statement.

I still think as a general rule, you should not measure on machines tools for a variety of reasons, except for what I pointed out earlier.

What is your justification for checking parts on the machine tool?
Parts too big for CMM?
Intermediate Checks?

It seems like lost production and independence of the measurement is a big reason not to. You also dont access to the datums or the variety of probing.

What are the limitations of using the machine tool to check parts. When would you not do it?

jan,
what were you referring to?

Many people say you can not measure on machines. In principle they may be right. In practice, I have found them to be wrong.

You are right, it is mostly for parts that I do not want to break the setup yet. Intermediate runs are good. But also final runs. You can NEVER replace a CMM, but you may find that if you get good correlation between CMM and machine tool, you may go over to the machine tool and not to the CMM anymore (except for first artical and spot checks). Yes, I buy hardware off of the machine. But my machines are all calibrated and we use artifact to check for drifting.

Our justification is that we run extreme close tolerance parts. They are very valuable and we want them to be right the first time. I can track tool wear, tool wander, offset correction etc. This all while the part is still ON the machine. After the operator is done machining, he cleans the part and runs the program. Takes all of 1 additional hour (it takes them about 20 hours to make manufacture the part in the first place). After that run, I have instant feedback. If the part is good, it immediately moves on to the next operation. No queue time in CMM land. And there you have the justification. Our queue time in CMM land was 2 to 3 days. That has evaporated to nothing on most parts. On top of that, we get better and faster SPC feedback. Also, we were looking at buying more CMM's. Still need one addional machine, but not 3, as originally planned.

Access to datums is an issue. We therefore work with simulated datum structures on the fixture. And you are right, this is a weak point. But we have overcome that issue with experience. If you have a part that has a LOT of springback, from restrained to un-restrained, that's a problem of course. But we have found that if you have a good fixture, that issue just goes away.

There must ALWAYS be independance in measuring, so CMM measurements will never go away. However, how many do you really need to do? In our case we have to do full inspection. By moving it to the machine, we have eliminated lots of queue time plus the need to buy more CMM's is lower now. Sure Hexagon wants to hear that one...


Jan.

Hi Jan,

You mentioned problems with coordinate rotations. Could you be more specific. Also, in another thread you reference Inspection+. Is that part of PCDMIS or some other software?

Thanks

Be aware that NC is a different animal from regular PC-DMIS on a CMM.

Regular PC-DMIS executes PC-DMIS code line-by-line. Therefore, when it encounters an alignment, it will rotate the actual coordinate system for the machine and causes the machine to move (and report) in the "part coordinate system".

This does NOT work like that in NC. In NC you create a "static" G-code program and the data coming back is in that coordinate system. You can NOT cause the machine to move in the new part coordiante system in one program. This causes quite a bit of confusion with typical CMM programmers.

This is not a real problem. You just have to be aware of it. Obviously, I was not, in the beginning...

Inpsection+ and MP700 are Renishaw products, see www renishaw.com. You can (but do not have to) use Inpsection+ in tandem with NC. Since I already have MP700s and Inspection+ on all my machines, this was an obvious course of action.


Jan.

Hey Jan, are C axis part of the calibration? I am thinking of backlash issues that I have seen on machines that are not supposed to exhibit backlash. Or is the C axis not part of the measurement routine? I know nothing of PCDMIS NC. I am in J's camp where I will not lend any credibility to a machine tool's measurements for the most part. But I can see a value for in process monitoring especially on large parts, long run times, situations where breaking the setup is timely/costly, I'm sure the reasons add up. I know you are doing well by it but do you have any pitfalls to share. Things that you have to work around perhaps?

Craig

OMG is not only for large parts.

Hey Jan, are C axis part of the calibration? I am thinking of backlash issues that I have seen on machines that are not supposed to exhibit backlash. Or is the C axis not part of the measurement routine? I know nothing of PCDMIS NC. I am in J's camp where I will not lend any credibility to a machine tool's measurements for the most part. But I can see a value for in process monitoring especially on large parts, long run times, situations where breaking the setup is timely/costly, I'm sure the reasons add up. I know you are doing well by it but do you have any pitfalls to share. Things that you have to work around perhaps?

Craig

We are measuring extremely close tolerance parts on very very good machines. I have correlation with a Zeiss UPMC Karat to within 0.0002". I used to be in the camp of the doubters, until I actually started to do the work.

The key is to have very very good machines. That means dimensionally stable and with limited backlash. And you need to calibrate them, just like you do your CMM and yes, this includes the C axis. By the way, typically, you can only correct for angular errors. After you know how your machine performs as a CMM, then have artifacts that make sure that the machine remains accurate: in other words, start to treat your machine like your CMM. Our most accurate machines are sitting in our CMM room, side to side with the Zeiss's.

We have rotary tables that have encoders mounted directly to the rotary axis shaft. We do NOT monitor the motor encoder position. Therefore backlash is a minimal issue for us. Our biggest challenge on the rotary is actually the wobble. We spend an awful lot of money to keep that to a minimum too.

I use PC-DMIS with the rotary option on my most challenging part. I found that most features can be measured without re-datuming, except at one angle. We found that that specific angular position is the area where the wobble is the highest (with respect to the starting position). For that position I had to re-datum in order to be able to measure surface profiles to a 0.0004" FCF.

I am finding that my best machines can measure with very close correlation to my best CMM's. It is a mistake to think that OMG is only for large parts.

The savings are reduced load on the CMM's (I will never state that CMM's will be eliminated, because that is simply not true), improved process flow and improved quality ownership by the operator (because he has immediate on-the-machine, in-the-fixture results).



Jan.