How much in Machine Inspection are you doing?

[rgrin]

About a year ago I was tasked with setting up a pallet changing 5 axis.  I was pretty green then and have learned a ton since then.  I have convinced management that it is time for a refresh on our processes with the end goal being to get an operator level person able to run this machine.  One thing I am keen on implementing (and have started doing so) is using the probe to ensure the correct stock was loaded for the given program.  We run a family of parts that vary from 10"x10" to 14"x14" and it would be easy to either select the wrong program or load the wrong material and cause a crash.

I am now exploring adding probing to at least verify the critical features at the end of the program.  The main thing is that we use the tight tolerance holes inherent to the part to locate on our 2nd op fixture.  I don't think its wise to replace a robust inspection plan with the machine probe, but it might be smart to use the probe to spot check here and there.

 

Thoughts?  Other other ideas to keep a machine like this running more are always appreciated. 

[Leon82]

We do it for in process datum pickups a lot.

Some for loading errors. I have it rejecting the pallet if it's mis loaded.

 

You could use if for inspection and either make a macro to recut it and re inspect or reject the pallet and stop the machine. You have to tell it to ignore the out of tolerance flag and run your macro instead. Maybe a spring pass first then you can give it a ratio to adjust cutter comp. The book should have an example 

 

[Simon Kausch]

the CIMCO probing  plugin makes it very easy so I use it quite often.

[Jake L]

Two reasons we shy away from in machine inspection.

1. Do you want your machine making chips or inspecting parts? We'll opt for chips over inspection everyday. 

2. Most of our parts have some sort of stress or spring to them when you take it off the fixture. Because of this, we would get different readings in vs out of the machine. Also, if the print doesn't specify that we can inspect the part "in restrained position" then the part must be off the fixture to inspect.

We do have some parts that we probe datums so we can hit tight true position tolerances.

[rgrin]

3 hours ago, Jake L said:

Two reasons we shy away from in machine inspection.

1. Do you want your machine making chips or inspecting parts? We'll opt for chips over inspection everyday. 

2. Most of our parts have some sort of stress or spring to them when you take it off the fixture. Because of this, we would get different readings in vs out of the machine. Also, if the print doesn't specify that we can inspect the part "in restrained position" then the part must be off the fixture to inspect.

We do have some parts that we probe datums so we can hit tight true position tolerances.

1. As of right now, I don't think I will ever be able to get this machine to run past 24 hours unattended. Due to size of work we run, we typically can only load 1 part per pallet and I only have 20 pallets. Average cycle time is 1-2 hours.  So, I think there is room to add some probing.  We have a heavy emphasis right now on making the first part right. We are no longer planning on ordering extra material with jobs and instead relying on having a small inventory of extra stock on hand in case something does go wrong.

2. That is something to consider.  Our parts are running on dovetails and this really helps prevent the parts from being overly distorted from workholding.

[Zoffen]

So many questions to give you some good feedback...

What kind of machine/Pallet Pool?

What type of work? 

What type of materials?

Cube size/Weight of parts?

Quantity of each part#?

How good are your operators?

How well can they follow instructions?

Can you do critical features with dedicated tooling/offsets so you know these features are dialed in and don't need to probe?

What is the avg material cost for these parts?

How many shifts do you run?

 

To really get a 5 axis cell to shine and pump out work takes a COMPREHENSIVE look at the entire process as a whole. This includes the business management side as well because when you really get the parts cranking 24/7 you run into a whole host of problems that have nothing to do with the actual machining...

Its really amazing what one spindle can churn out with the right support and encouragement!

[JParis]

2 hours ago, Zoffen said:

Its really amazing what one spindle can churn out with the right support and encouragement!

Yup... still trying to teach this aspect to an old dog...

[cncappsjames]

If "inspection" is a bottleneck (meaning it's holding up part shipments) then it's worth further serious examination. If available spindle hours is a limitation, in machine enspection is going to exacerbate the situation.

If you're referring to setting work offsets, checking stock size/location/orientation, critical features for re-machining, etc... then my answer is I do one or all nearly every time.

[Leon82]

If you can cut a lot of features with 1 tool you can probe the last feature cut. With CMM verification of the first part it should be a reliable fast check that won't eat too much time .

[rgrin]

  Machine is a Yasda H40i. Currently setup with a 20 pallet pool. We removed 4 of the pallets so we can run larger parts for our customers on our bottom row of pallets.

  We are a job shop. We are using this machine to fulfill our largest customer demands and are focusing on getting as much repeat work into this machine while using the same fixturing. (Dovetails primarily and vises for stock prep). We machined a bunch of subplates and use ball locks to be able to do faster setup changes in the event we need to do that, say if our customer base changes.

  We run 90% alu thru this machine and some 17-4. This will likely change to almost 100% alu for the near future just based on customer demands.

  The main part family I have already made programs for 40 unique part numbers and I know they have a ton more made in Germany, but they won't give us more until we get our lead times down. Hence why we are running them in the Yasda.  Typical job qty is 1-3.

  I am programmer and operator right now.  But I am trying to put pressure on management to let me take over process changes for this part family so that I can make adjustments to it as needed.  We have alot of old paperwork that is way out of sync for how we make these parts.  We are extremely bad at keeping things up to date since our engineering team is more worried about quoting new work as opposed to updating old work. 

  My personal goal is to not be running this machine by end of year and to have enough work for someone to be able to keep it running.  Currently, inspection is our bottleneck. We have 1 inspector and 1 CMM that services 10 machinists all making one off parts for the most part.  I could see us getting another CMM for the shop floor this year just to inspect our repeat work.

[Kyle F]

2 hours ago, rgrin said:

I am programmer and operator right now.  But I am trying to put pressure on management to let me take over process changes for this part family so that I can make adjustments to it as needed.  We have alot of old paperwork that is way out of sync for how we make these parts.  We are extremely bad at keeping things up to date since our engineering team is more worried about quoting new work as opposed to updating old work. 

  My personal goal is to not be running this machine by end of year and to have enough work for someone to be able to keep it running.  Currently, inspection is our bottleneck. We have 1 inspector and 1 CMM that services 10 machinists all making one off parts for the most part.  I could see us getting another CMM for the shop floor this year just to inspect our repeat work.

Really nice machine you got there, by the way.

I am also very interested in the turnout of this thread, as I am my shops only 5axis programmer/set-up guy, and we just acquired 2 very nice 5axis pallet mills, so I can foresee a lot of similar issues for myself. Someone else is going to have to step it up lol!

 

Only thing I may have to add, since you say your bottleneck is Q.C.

assuming you're as9100/iso, our inspector lady told me she doesn't mind in-process probing for feature verification, but it does not take the place of any Q.C. checking like a FAI. So it works great for keeping the machine running and having confidence in parts being nominal, but it may not be enough to alleviate the q.c.  bottleneck. She basically put it as "you can't have the machine that made the parts, also check the parts"

[cncappsjames]

8 minutes ago, Kyle F said:

...She basically put it as "you can't have the machine that made the parts, also check the parts"

This is always the fun discussion with QC.

If 1) one maintains the same coordinate system the parts were made in, or, 2) there's a requirement the parts me measured in the unrestrained condition, then that is absolutely correct. However, if there is no specification for 2), then creating a new coordinate system, then measuring, provided the machine has glass scales, you "can" yield the same results as a CMM. And if you have a Renishaw SPRINT scanning probe, you can even check freeform profiles.

I had a QC manager tell me I was "Full of $#!+..." so I asked for their CMM stylus size, point position and cycle measurements. The individual indulged me (to my surprise). The machine produced results within 5µm of the CMM report. No features were closer than .010µm. I suspect machine envelope/coolant temperature was the culprit for the differences as the part were aluminum.

Now, because they were so specific in their ISO/AS manual, they could not add CNC Inspection to their procedures, but the point was made, the results spoke for themselves, and the QC Manager said "I stand corrected, but, I am not allowed to deviate from the ISO/AS procedures, however, they would discuss adding CNC Measurement for 'certain' part buyoffs so long as reports can be generated from the CNC directly (they can).

Once you break the work coordinate system link (G54, G55, etc...), the part and CNC are no longer connected in the traditional sense. If you maintain the same coordinate system then the CNC and part are connected.

[Jake L]

27 minutes ago, cncappsjames said:

Now, because they were so specific in their ISO/AS manual, they could not add CNC Inspection to their procedures

Gonna hijack the thread real quick. Is it true for ISO that each company writes their own rules? If so, why not be super vague to allow the most flexibility?

[rgrin]

21 minutes ago, Jake L said:

Gonna hijack the thread real quick. Is it true for ISO that each company writes their own rules? If so, why not be super vague to allow the most flexibility?

I think this is partially true. Depending on the type of work you want to do, might change what you put in there though.  Our Quality Statement won't allow machine verification to be the final word on if a part is good. We do this because of our ITAR/DoD work we do and they really like to see green squares on a CMM report.

My current plan is to verify locating features and the random tight tolerance features that are tricky to check by hand.  Just to give me (and future operators) the confidence that they are making a good part.

1 hour ago, Kyle F said:

Really nice machine you got there, by the way.

 

I like it, but it has some quirks that really drive me nuts.  They saddled it with third party software for scheduling pallets and the tool magazine.  The implementation is sub par IMO.  It also has the LONGEST tool break detection I have ever seen. 30 seconds for a fast tool break with the laser and over a minute if I want to use the touch sensor. (When the tool is already inline with the sensor. It's even longer when it has to position itself)  I'm almost at the point of making my own.

It also has a major problem with flushing chips away from the pallet interface.  I am constantly dealing with chips getting underneath and am trying to find a better way than the single coolant nozzle they set it up with.

Edited December 27, 2023 by rgrin
(Clarified tool break time)

[JParis]

21 minutes ago, Jake L said:

Gonna hijack the thread real quick. Is it true for ISO that each company writes their own rules? If so, why not be super vague to allow the most flexibility?

Yes...you have to be clear enough so as to create a process that can be audited BUT you need to walk a line that allows you to adapt to situations that might be out of your norm...

I have seen companies attempt to kill themselves by documenting everything to the 'nth degree...they always get in trouble and have to rewrite most everything to accommodate some flexibility in their process.

For a loose definition....

ISO = Say what you do, Do what you say and have the documentation to prove it.

 

 

[Kyle F]

1 hour ago, rgrin said:

It also has a major problem with flushing chips away from the pallet interface.  I am constantly dealing with chips getting underneath and am trying to find a better way than the single coolant nozzle they set it up with.

ooof that does not sound like fun. I wonder if an extra line or two can be retrofitted? 

[Zoffen]

5 hours ago, rgrin said:

 We are extremely bad at keeping things up to date since our engineering team is more worried about quoting new work as opposed to updating old work.

Without thorough documentation your journey is all but Doomed.

DOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOMMMMEEED

 

I say.

 

[rgrin]

4 minutes ago, Zoffen said:

Without thorough documentation your journey is all but Doomed.

 

Too True.  It's why I am putting pressure on them to let me take over managing it.  I'm sure they would want to check my work as I go since I'm just a machinist and not an engineer.  But how hard could it be to type correct work instructions into an ERP and make sure it's actually reflective of how we do stuff

[cncappsjames]

2 hours ago, rgrin said:

I think this is partially true. Depending on the type of work you want to do, might change what you put in there though.  Our Quality Statement won't allow machine verification to be the final word on if a part is good. We do this because of our ITAR/DoD work we do and they really like to see green squares on a CMM report.

Properly formatted data from a CNC can be imported into reporting software.

"Machine Verification" is merely a mental hurdle not a physical or mechanical limitation until the decimal places move beyond 4 places metric and 5 places inch.

A properly calibrated, and volumetric error mapped CNC is nearly as capable as the highest of high-end CMM's. Elect to get the Renishaw SPRINT scanning probe and limits become even fewer.

Don't get me wrong, I'm not saying a CNC should replace a CMM. I'm just saying that a CNC CAN be used to take some of the load off the CMM department successfully in many situations, but because of mental hurdles that people cannot or have a REALLY hard time overcoming it gets dismissed. 

I liken this discussion to the "SPC" discussion. There's a TON of misconceptions out there, and the misconceptions have been passed down for so long nobody questions them anymore and the misconceptions and waste persist.

2 hours ago, JParis said:

Yes...you have to be clear enough so as to create a process that can be audited BUT you need to walk a line that allows you to adapt to situations that might be out of your norm...

I have seen companies attempt to kill themselves by documenting everything to the 'nth degree...they always get in trouble and have to rewrite most everything to accommodate some flexibility in their process.

For a loose definition....

ISO = Say what you do, Do what you say and have the documentation to prove it.

#FACTS

[CEMENTHEAD]

We did this once. on our VG5000's a few years back,  Machine inspection send back was good but comparing to CMM was off a bit. Machine error will reflect on inspection in the machine. We were able to duplicate it. a few tenths off.  We can't trust that.

All inspections are now still done in inspection. You can make a lot of scrap checking a part with a machine that is out. We ball bar ours twice a year, but yeah if your gauges  are off, and inspection uses them to check a part, it's still going to be off. 

IMHO don't do that.

[JParis]

Generally speaking the inspection equipment should be 3x's better that your tolerance....

Check a ±.005, ok, Check a ±.002, should be OK.....anything tighter than that on a CNC machine should not be trusted for inspection purposes.

I know people try...some might be somewhat successful but it is ALL TOO EASY to lose control of a process doing that. 

[cncappsjames]

21 minutes ago, JParis said:

Generally speaking the inspection equipment should be 3x's better that your tolerance....

Check a ±.005, ok, Check a ±.002, should be OK.....anything tighter than that on a CNC machine should not be trusted for inspection purposes.

I know people try...some might be somewhat successful but it is ALL TOO EASY to lose control of a process doing that. 

That's HEAVILY dependent on the machine, environment, calibration, etc... Some machines I'd not trust for tolerances less than ±.0100"  - my world is a little different than most.

We helped a customer set up some inspection (on a Matsuura LF-160). The customer did some testing. They found  the sweet spot in their shop (climate controlled 24-7 and temps held to between 68-72° F and humidity was measured at the very least) was ±.010mm (.0004" rounded) and above. Below that environmental factors just played too big a role thus necessitating sitting in QC lab for the 24 hour time period specified in their QC documentation. As stated before, volumetric error mapping, regular machine calibration, coolant temperature management, etc... . A ball-bar test can be telling, but it doesn't tell the full picture and I would DEFINITELY not be buying parts off from a machine that had only been ball-bar'd. There's WAY more to it than that.

JM2CFWIW

[JParis]

29 minutes ago, cncappsjames said:

That's HEAVILY dependent on the machine, environment, calibration, etc... Some machines I'd not trust for tolerances less than ±.0100"  - my world is a little different than most.

and that's what I stated at the beginning....  "Generally"

[Leon82]

11 hours ago, rgrin said:

I think this is partially true. Depending on the type of work you want to do, might change what you put in there though.  Our Quality Statement won't allow machine verification to be the final word on if a part is good. We do this because of our ITAR/DoD work we do and they really like to see green squares on a CMM report.

My current plan is to verify locating features and the random tight tolerance features that are tricky to check by hand.  Just to give me (and future operators) the confidence that they are making a good part.

I like it, but it has some quirks that really drive me nuts.  They saddled it with third party software for scheduling pallets and the tool magazine.  The implementation is sub par IMO.  It also has the LONGEST tool break detection I have ever seen. 30 seconds for a fast tool break with the laser and over a minute if I want to use the touch sensor. (When the tool is already inline with the sensor. It's even longer when it has to position itself)  I'm almost at the point of making my own.

It also has a major problem with flushing chips away from the pallet interface.  I am constantly dealing with chips getting underneath and am trying to find a better way than the single coolant nozzle they set it up with.

I made a sub program to go to b-90on the mam and blast it with thru spindle while rotating. The 30 seconds beats a night lost to a gap check alarm 

[Leon82]

Back in the day Moore Tool had a jig-bore model for inspection if I recall