What's new
What's new

3D model modifications after CAM Programming

Jay Fleming

Hot Rolled
Joined
Aug 7, 2014
Location
Noble, OK
The entire title suggests it should be in the CAD/CAM forum but the question I have doesn't really relate to any particular software, so here goes.

Suppose you model a part, which is part of an assembly that may not physically change. As your machining the parts, it is determined additional material removal is required for a better fit, therefore you make a change to your CAM program. Do you then go back and make your model match what would be the output from the machining operation, then redo your CAM programming, for future use? Or just leave everything as is and change if an actual modification is called for down the road?
 
How much does your documentation matter? If it will never be seen again, why bother? Personal experience has taught me everything should be updated, because if you stay in business long enough that job you thought you would never see again will come back.
 
I have done both, for one off thing I build for myself it might be changed only in CAM. For a customer job I usually change the model. I just did a customer job today that had a change from 100 to 82 deg countersink, I did not change the CAD model. It is a prototype so if I get another order I am expecting an updated design/ new files anyway. My program let's me re-import a changed CAD model with all the same operations so I just need to recalculate and verify everything still works as planned. I suppose it only matters if it matters.
 
Thanks for the thoughts. These are in house parts. I'll plan to change the model the next time I run a batch.

Sent from my HTC One using Tapatalk
 
Just remember, sometimes depending on how much you need to cheat the program, if you change the model, radius values may need adjusted as well. I work for Boeing, and some model makers miss that.
 
The important thing is that the model is up to date, period. IF the job comes back the programmer should in theory grab the most recent cad model and reprogram at that point. At least that is what we do.
 
If I have to make a change like that I'll do it on the model, then reimport into CAM and update. Making modifications in CAM and not changing the model is akin to hand drafting revisions to a cad print.
 
The entire title suggests it should be in the CAD/CAM forum but the question I have doesn't really relate to any particular software, so here goes.

Suppose you model a part, which is part of an assembly that may not physically change. As your machining the parts, it is determined additional material removal is required for a better fit, therefore you make a change to your CAM program. Do you then go back and make your model match what would be the output from the machining operation, then redo your CAM programming, for future use? Or just leave everything as is and change if an actual modification is called for down the road?

I'm still trying to nail down the exact meaning of what you are driving at here ?

My interpretation of what you wrote is kinda like this.

Jay writes: "Suppose you model a part, which is part of an assembly that may not physically change. As your machining the parts, it is determined additional material removal is required for a better fit, therefore you make a change to your CAM program"



Apologies for length, (just wanted to nail down the situation... / bugging me/ knowing away at me lol).

My interpretation of what you wrote is:

1. You can't change the model for the part (that needs adjusting) because it's part of a CAD assembly representation. I.e. the part you are cutting in "computer/math space/CAD" directly fits with adjoins abuts or is fitted to other parts as a sub component in a larger CAD representation of potentially 9 or more other parts? [Well two or more at least]. So if you change the dimensions on one part then therefore one would have to change the dimensions and tolerances on all the other contacting components? OR are you saying it doesn't matter cuz the final dimensions are smaller / were oversized, just remove more material on the one part so it won't affect the other parts in the assembly (kind of thing?)

2. It sounds like the issue you may be having might stem from CAD errors of representation VS set up and machining errors... VS. hard coded CAM won't have CAD translation/geometric errors?

3. It's not clear if the changes you are having to make are from Errors in CAD, natural errors that might happen if you are trying to achieve really close or difficult to achieve tolerances? Without knowing the tolerances or reason or source of change / error it's hard to make an explicit call here, although I obviously agree with everything said so far.

4. Are the changes to make the thing "Work" just plane old design errors?

Maybe I'm way over thinking this but from the wording of your post ^^^ These things came immediately to mind.



The question you pose is interesting to me as most machinists seem to focus on absolutely busting their arse to deliver on that one component or part in the best way they can. But @Jay Fleming what your question poses here is more about how a machinist (perhaps in a prototype environment) might be responsible for multiple parts working together... That's very different. Kinda crosses the line between machinist and project engineer IMO.

[Mild anecdote: As a more naeive designer 25 years ago (still kinda wet behind the ears) I gave out institution's head machinist a very well executed assembly drawing, and he looked at me as if this was a major insult and I had given him a steaming turd. So he had to explain that he could ONLY work off single part drawings/CAD drawings etc. We became (eventually became) BFFs... But still underscores the incredible focus that a lot of machinist's have for the ONE part and I have to be honest that most designer's think in terms of assemblies and systems of parts working together].

_________________________________________________________________________________________________________________________________________________________________

Long story short: scratching my head seems like you have to attach or maintain the physical changes as a separate step (interim working attached annotated document) before things go back up stream so that the designers or engineers can decide how things need to be altered. If its really close/difficult tolerance stuff that is causing problems between CAD representations and physically cut surfaces it could take Three or more iterations for the CAD model and final machined surfaces to converge with each other YKWIM? Not to get super bureaucratic about it all but kinda depends on what the organization is and what procedures either constructive or obstructive that may exist to iron out all those kind of wrinkles.
 
Intergrated CAD/CAM

LOL...

Mysteriously after I made the above post I get an "Urgent" Quote from SOLIDCAM (no affiliation)... (they are good peeps).

But did give me a "Nudge" to think that maybe that is one advantage of an integrated CAD CAM solution. So for example such a design change that Jay is mentioning here could/maybe taken care of really easily as the CAM interface is built into Solidworks, so there is (I believe) the possibility of an upstream "ripple" of changes if you want it. So the CAD is adjusted accordingly... (maybe I am being overly naïve there but that's how it's supposed to work).

Not wanting to hijack the thread: but Integrated CAD CAM does sound great/amazing but in the case of Solidworks I'm tired of being fleeced and wish they would go to subscription model (as my CAD needs are sporadic, intense use for a few months and then don't use it for 8 months kind of thing). So my CAM end of things will or completely collapse on one front because of the higher end versions of Solid Works I initially purchased. I can't split the license into a lower end version of Solidworks and jettison the higher end product. I really don't want to get into maintaining multiple seats of a constellation of different versions of what is essentially one product. Also tired of the "infinite" budget expectation to make everything work...

So that's kinda why I decided (perhaps) for more complex work that maybe splitting CAD and Design software for CAM would be a good or better idea/more long term reliability? That's why I'm leaning more towards SolidEdge and Esprit. Good design tool and really good CAM tool... Not dependent (or interdependent) on each other and SolidEdge has subscription service so I can use it for $200 to $300 /month when and if I need it... That's a lot cheaper than a $10,000.00 Solidworks license renewal just so I can run an additional $15K upgrade to my existing CAM license... (To go from 3 axis to 5).
 
The entire title suggests it should be in the CAD/CAM forum but the question I have doesn't really relate to any particular software, so here goes.

Suppose you model a part, which is part of an assembly that may not physically change. As your machining the parts, it is determined additional material removal is required for a better fit, therefore you make a change to your CAM program. Do you then go back and make your model match what would be the output from the machining operation, then redo your CAM programming, for future use? Or just leave everything as is and change if an actual modification is called for down the road?
.
CAD model is rarely 100% accurate. often parts are shown at max od size and minimum id size and part wont fit together without some clearance.
.
had printing press giving a oscillating problem. Swiss designers could not say why. i check with optical alignment, indicators and precision levels and i cannot see misalignment.
.
then i bolt on a precision level and watch mechanism run while watching the level readings change. as the load go to end of a steel guide bar the guide bar was bending under the load which was effecting how mechanism traveled. fancy CAD model did not show steel bending from forces and bending more as force was traveling the guide bar.
.
i used CAD model in semi transparent mode to give Xray type vision to see the mechanical assembly and see through sections or parts and study it for weeks all while actual machine was running. like a lot of things CAD model was close but not perfect. it did not show stuff bending or changing shape as weight and parts push and pull on one another
 
.
CAD model is rarely 100% accurate. often parts are shown at max od size and minimum id size and part wont fit together without some clearance.
.
had printing press giving a oscillating problem. Swiss designers could not say why. i check with optical alignment, indicators and precision levels and i cannot see misalignment.
.
then i bolt on a precision level and watch mechanism run while watching the level readings change. as the load go to end of a steel guide bar the guide bar was bending under the load which was effecting how mechanism traveled. fancy CAD model did not show steel bending from forces and bending more as force was traveling the guide bar.
.
i used CAD model in semi transparent mode to give Xray type vision to see the mechanical assembly and see through sections or parts and study it for weeks all while actual machine was running. like a lot of things CAD model was close but not perfect. it did not show stuff bending or changing shape as weight and parts push and pull on one another

@DMF_TomB that's a great example you cite there...

Really that is a design fault/engineering fault.

I can't speak for other designers but I typically do several months of FEA/simulation work on the critical functional components (early stage design work/sanity checks), before the designs start to become more refined... For things I can't realistically simulate, (i.e. too real world and complex) I do build physical test mock ups...

And it is quite Ironic that the example you cite (@DMF-TomB) is the reason why MY SOLIDWORKS license for CAD / CAM is so expensive is because it has FEA dynamic analysis for assemblies... LOL so it might have been that SoildCAM inside of Solidworks with FEA for assemblies would have picked up on that ("unforeseen" design problem); (The irony is not lost on me there).

I Think its boils down to being a "small fish" (rather than a large institution), that we use certain capabilities of CAD/CAM and FEA at different times of the year or project phase and not continuously... Like I don't have three design engineers all doing basic FEA or basic design work 24/7. Year in year out.

But having said all that @ DMF_TomB it is true that CAD is not always perfect description. FEA can only model quite crude things IMO and things become much more uncertain when you have rolling element bearings and various linkages and connections as contact surface conditions and loads and pre-loads seem quite difficult to model accurately unless you have a NASA level super computer...(and a lot of time), Cheaper to build and test (at least for our little applications).


@DMF_TonB I hear you about complex oscillations as sometimes additive wave functions and harmonics are not always easy to predict or model and even NASA has had troubles with that in that past YKWIM? (lol)...
 
@DMF_TomB that's a great example you cite there...

Really that is a design fault/engineering fault.

I can't speak for other designers but I typically do several months of FEA/simulation work on the critical functional components (early stage design work/sanity checks), before the designs start to become more refined... For things I can't realistically simulate, (i.e. too real world and complex) I do build physical test mock ups...

And it is quite Ironic that would cite your example as the reason why MY SOLIDWORKS license for CAD / CAM is so expensive is because it has FEA dynamic analysis for assemblies... LOL so it might have been that SoildCAM inside of Solidworks with FEA for assemblies would have picked up on that lol (the irony is not lost on me here).

I Think boil down to being a "small fish" that we use certain capabilities of CAD/CAM and FEA at different times and not continuously... Like I don't have three design engineers all doing basic FEA or basic design work 24/7.

But having said all that @ DMF_TomB it is true that CAD is not always perfect description. FEA can only model quite crude things IMO and things become much more uncertain when you have rolling element bearings and various linkages and connections as contact surface conditions and loads and pre-loads seem quite difficult to model accurately unless you have a NASA level super computer... Cheaper to build and test (at least for our little applications).

.
also difficult to calculate temperature changes and different alloys changing shape. i aligned a linear rail to a concrete floor to measure roundness of 18 foot dia 50 ton wheel next to it, next day linear rail is bowed
.
under a skylight window. steel rail warmer than concrete floor. longer rail bowed and was straight when cloudy on the next day.
.
i once had trouble leveling many sole plates 100 feet apart. readings not repeating. i zeroed on outside building column so i put another zero reference on inside building column. i measure outside building column change height .030" over 6 hours from the effect of sunlight on outside walls.
.
had another job i see optical level change as fork truck drives by literally making the floor go down like being on a trampoline
.
i have often measured buildings and large machines move. had a laser setup and when 200lb man stepped on sole plate for 10 ton machine sole plate went down .0001", man got off it went up .0001" tried it many times and it repeated
.
CAD models dont show nylon plastic get bigger when wet and when dried out od is smaller and id bores are bigger.
 
.
also difficult to calculate temperature changes and different alloys changing shape. i aligned a linear rail to a concrete floor to measure roundness of 18 foot dia 50 ton wheel next to it, next day linear rail is bowed
.
under a skylight window. steel rail warmer than concrete floor. longer rail bowed and was straight when cloudy on the next day.
.
i once had trouble leveling many sole plates 100 feet apart. readings not repeating. i zeroed on outside building column so i put another zero reference on inside building column. i measure outside building column change height .030" over 6 hours from the effect of sunlight on outside walls.
.
had another job i see optical level change as fork truck drives by literally making the floor go down like being on a trampoline
.
i have often measured buildings and large machines move. had a laser setup and when 200lb man stepped on sole plate for 10 ton machine sole plate went down .0001", man got off it went up .0001" tried it many times and it repeated
.
CAD models dont show nylon plastic get bigger when wet and when dried out od is smaller and id bores are bigger.

Not to go off topic too much but again excellent examples you point to. Yup very real/does happen and can be super important to take into account/consideration.

In CAD and FEA you can simulate thermal expansion/ contraction. Shape and movement and movements even down to 1 micron kind of thing with similar and dissimilar metals...

[Off Topic]:

My bag is supposedly precision optical/digital systems so I have to think constantly about thermal expansions, symmetry and effect of dissimilar metals +corrosion effects too/long term use. In my minds eye before hand, I try to imagine everything is made of soft rubber to predict how stuff might move or deform. I have to admit that CAD/FEA is not very good for near micron type simulation of what can happen at various contact surfaces. (That's relevant to the stuff I'm trying to roll out) the kind of thing where spotting compound would be used on mating critical high precision surfaces that CAD simply can't simulate yet(as far as I am aware) [Simple example fretting/spotting compound used to check contact surfaces on a spindle taper/tapered tool]. Even reasonably accurate FEA simulation of something like spherical roller bearing is in fact very challenging, even the most expensive and complex option from NX Nastran (THE FEA guys), can barely get a good handle on that. So you always end up (instead) deriving simple physical models and constraints from real world testing (Sometimes tuff is just too big and expensive to physically build and test 1:1 scale all the time).

Totally agree about Nylon, swelling etc... Not being taken into consideration.

______________________________________________________________________________________________________________________________________________________________


[On Topic]:

Does lend credence to the CAM solutions that are very focused on sophisticated and real world surfaces and machining processes... Like Hyper mill, Open mind, Esprit and the like...


Trying to find a good analogy here... I think Gregormarwick said it quite well... " If I have to make a change like that I'll do it on the model, then reimport into CAM and update. Making modifications in CAM and not changing the model is akin to hand drafting revisions to a cad print. "

The CAD model is only a representation... Seems the way that Software CAM packages like Esprit and Hypermill and others "Dissect" that import/model and what they do subsequently is much more meaningful to the final result / product. So I think of it as a "split reality and duality" of CAD Representation (which is a bit of a "wish list" if you will (in three dimensions)) and then the actual CAM model and associated ops and processes as being closer to "What makes it so". And then of course what actually happens on the machine + control / code and operator and machinist + INSPECTION etc. etc. ... That's super important too. Not 100% sure that there are tools that make that ALL converge perfectly in one go... But I'm kind of Newb at some of this/ A lot of this so I'm sure there folks out there that have waaay better or more masterful strict procedures to formally deal with these kinds of things rather than folks like me that winging-it to achieve a final result (or very narrow set of goals). My inclination would be to go around the mulberry bush many times before I would make a formal change to a top down CAD DESIGN model. The distinction being a CAM-model vs a CA-Design-model [IMO they are not the same thing]. But as I say there are pros, that have thousands of procedures and contingencies and formal work flows for this kind of thing (I would suppose)?
 
Holy hell!! Next time I need a dissertation on why 3D models are different than functional parts, I'll know where to look.

As long as there have been Engineers that don't actually make parts, this has been an ongoing topic. Long, long, long time before CAD.

The real question is, do you go back to the source for accurate information; The answer is YES, always. No point in having missing parts of an information chain.
 
Holy hell!! Next time I need a dissertation on why 3D models are different than functional parts, I'll know where to look.

As long as there have been Engineers that don't actually make parts, this has been an ongoing topic. Long, long, long time before CAD.

The real question is, do you go back to the source for accurate information; The answer is YES, always. No point in having missing parts of an information chain.

That's hilarious :-)

Very useful point..! RE: "ongoing topic. Long, long, long time before CAD."



Also (an aside) I've had too much coffee this morning... kinda hyper and I'm going to step away from this thread and go to the gym before I bury myself (and this thread) with too many f*cking worlds lol.

I'm a bit of an excited "Puppy" this morning (for a number of reasons) so I'm goin to step AWAY from the computer while I still can...


:cheers:

_____________________________________________________________________________________________________________________________________________________________

Taking "Old school" Iv'e always thought of it as Design drawings Vs. Working drawings... The working drawings being the ones that get scribbled all over many times until in metal/material that the component is proven/inspected etc.. It could be many weeks and months before a formal Design drawing is altered and yet we would have gone through scores and scores / stacks of working drawings..? So the CAD model is the design model and the CAM model is closer to the working version /working drawings... (But folks here probably have very different names and ways looking at things and doing stuff etc.)
 
The CAD model is only a representation... Seems the way that Software CAM packages like Esprit and Hypermill and others "Dissect" that import/model and what they do subsequently is much more meaningful to the final result / product. So I think of it as a "split reality and duality" of CAD Representation (which is a bit of a "wish list" if you will (in three dimensions)) and then the actual CAM model and associated ops and processes as being closer to "What makes it so". And then of course what actually happens on the machine + control / code and operator and machinist + INSPECTION etc. etc. ... That's super important too. Not 100% sure that there are tools that make that ALL converge perfectly in one go... But I'm kind of Newb at some of this/ A lot of this so I'm sure there folks out there that have waaay better or more masterful strict procedures to formally deal with these kinds of things rather than folks like me that winging-it to achieve a final result (or very narrow set of goals). My inclination would be to go around the mulberry bush many times before I would make a formal change to a top down CAD DESIGN model. The distinction being a CAM-model vs a CA-Design-model [IMO they are not the same thing]. But as I say there are pros, that have thousands of procedures and contingencies and formal work flows for this kind of thing (I would suppose)?

I am sure I have not been in the game as long as you but I have yet to run into a case where the CAD model is not realistic or a case where my CAM software does not interpret the CAD model any way other than exactly what is defined in the CAD model.

Any decent CAD or CAM software should be able to model/read NURBS shapes/surfaces and it was my understanding that NURBS is an exact mathmatical representation of shapes/surfaces (only limited to significant digits as computed by the software/computer). The only times parts don't come out exact is due to variables like tool runout and size, tool deflection, or machine backlash or controller limitations. These (sub 5 thou tolorances) are the only thing I'll change in the CAM and never the CAD because I know the CAD is accurate.
 
I am sure I have not been in the game as long as you but I have yet to run into a case where the CAD model is not realistic or a case where my CAM software does not interpret the CAD model any way other than exactly what is defined in the CAD model.

Any decent CAD or CAM software should be able to model/read NURBS shapes/surfaces and it was my understanding that NURBS is an exact mathmatical representation of shapes/surfaces (only limited to significant digits as computed by the software/computer). The only times parts don't come out exact is due to variables like tool runout and size, tool deflection, or machine backlash or controller limitations. These (sub 5 thou tolorances) are the only thing I'll change in the CAM and never the CAD because I know the CAD is accurate.

Well I think you hit the nail on the head (in some respects) ... you said "Any decent CAD or CAM software should be able to model/read NURBS* shapes/surfaces"... I did a Ph.D (seems like along time ago now), in a CAD laboratory so being on the computer end of things I have a some vague insights (i.m being humble here not snarky) as to how those models are derived and how those surfaces (not thinking Nurbs here) are executed on a machine control with a G-code representation? To my way of thinking there is or still can be a pretty big gulf between the two (in some instances) I'm sure Nurbs on a M-560V in the right hands can do a good job, similarly Makino F-3 for mold work, but would be a bit "Sketch" on a Makino PS-95 kind of thing ? AND we are not even talking about production oriented issues such as just RAW motion control and how various manufacturer's handle tool motion. Hurco Ultimotion as being one example of proprietary "Fiddle" that does not strictly obey your G-code... But achieves a hopefully better result in terms of dynamic ACC and DEC... Even SolidCam (HSS tool paths) + Ultimotion (Hurco) don't step on eachother's toes yet do quite similar things... So I don't reckon that the result is 100% predictable by any means... So much can happen between CAD and final part...

So that's why some of the really expensive top drawer CAM packages do indeed a really good job in executing raw geometry (for a specific machine that has a really good post that has had the "snot" smashed out of it for over a 15 year or more period to converge on a really good solution for a particular machine) and that's also why some of the more budget-ish (still) not cheap stuff can kinda create more unexpected results.

My kind of work has low precision "Curvy" geometry but the high precision stuff is more orthogonal cuts/ prismatic/positional. SO I have not yet had to prove out complex curves just really high precision and accuracy on screwy angles in some cases.

Maybe 20 years now I get a better handle on this whole question LOL.

@Goldebfab great point and question... I'm going to run away whuile I still can...

________________________________________________________________________________________________________________________________________________________

*If you want to talk about Nurbs + 5 axis there's a hell of lot that can go wrong between CAD model and final cut part... There's a lot of skill there to pull that one off nicely... There are folks on here that face that challenge every day (day in day out) and by their skill knowledge and wit really do an amazing job of closing that very basic gap. It never ceases to amaze me how still today basic gauging procedures (and the like) on the machine and part are still really important to getting the part you want or is being asked for. There's still a lot of "Artfulness", insight and know how to get the "Part" you want from a CAD description. I certainly don't think I am being overly "Romantic" in that idea... I have seen various operators pulled off various machines and swapped around for good reason...
 
I am sure I have not been in the game as long as you but I have yet to run into a case where the CAD model is not realistic or a case where my CAM software does not interpret the CAD model any way other than exactly what is defined in the CAD model.

Any decent CAD or CAM software should be able to model/read NURBS shapes/surfaces and it was my understanding that NURBS is an exact mathmatical representation of shapes/surfaces (only limited to significant digits as computed by the software/computer). The only times parts don't come out exact is due to variables like tool runout and size, tool deflection, or machine backlash or controller limitations. These (sub 5 thou tolorances) are the only thing I'll change in the CAM and never the CAD because I know the CAD is accurate.

"Any decent..." is relative and up for debate.
"The only times...." holy shit are you on a different planet?
"I know the CAD is accurate"----How? Is it qualified, quantified, regularly calibrated and tested?

Jockeys :rolleyes5:
 
"Any decent..." is relative and up for debate.
"The only times...." holy shit are you on a different planet?
I think debating CAD software is beyond the scope of this thread but by all means..
Maybe so, or maybe you are, that is up for debate.

"I know the CAD is accurate"----How? Is it qualified, quantified, regularly calibrated and tested?
The part where I can open up a .STEP file with a text editor and see coordinate points with 20 significant digits and the part where the header of the file says ISO 10303. I know its not just this simple as CAD is some magical thing and its always right. I was just saying in my limited experience I have yet to have problem and in my experience when I have had issues with parts not coming out like they should the problems were traced to the things I mentioned and not the CAD file or CAM software. I'm not doubting things are less trivial for 5-axis work or sub thou tolerances but even then I doubt the problem is with the CAD model. But for those cases I would be interested in learning how those problems are solved.


So what of it? I am here to learn and have to start somewhere. Did you have something to say to contribute to the conversation or are you here just to pass insults?
 








 
Back
Top