Continue to Site

Welcome to MCAD Central

Join our MCAD Central community forums, the largest resource for MCAD (Mechanical Computer-Aided Design) professionals, including files, forums, jobs, articles, calendar, and more.

top-down design vs manufacturing

mdemers

Member
Hi y'all,


I have a sort of dilemma.


In our mechanical tem, there are designers like me who are convinced that top-down design (skeleton models) is the best way to use Pro/Engineer. Interfaces between parts are modeled in the skeleton model first and then in the parts,attaching the features to those of the skeleton model. This way, every interface has to bedimensionned once and all parts sharing it are modified accordingly, eliminating the risk of incoherence. Therefore, every feature is dimensionned to the nominal dimension. A shaft, for example, has a nominal diameter of 10 mm with a tolerance of +0, -0.1. The hub, also modeled to the 10 mm nominal value, has a tolerance of +0.1, +0.2. This has always worked very well for us as long as we had 2D drawings where the tolerances are displayed.


Unfortunately, there are some other peoplewho think more in terms of manufacturing and whosay that our 3D models should be modeled to the dimension in the middle of the tolerance range. Then, my shaft would have a diameter of 9.95
 
You can combine these 2 modes using the "Dim Bound" command (Setup -> Dim Bound). This will allow you to show "+-" tolerances in the drawing, while having the 3D geometry regenerated at the middlevalues. It worked like a charm for me just a few days ago. <?:namespace prefix = o ns = "urn:schemas-microsoft-com:eek:ffice:eek:ffice" />


Note that I mentioned "middle" values not "nominal" since middle and the statistical nominal values can differ.


Besides manufacturing this will help when doing FEA's, say when press-fits are involved.


On the other hand, the symmetric tolerancing/limit tolerancing works well if you have parts produced by cutting, say turning ormilling. When other methods are implied, such as casting, stamping, fine-blanking the symmetric/limits tolerancing doesn't suit the manufacturing process since the dies will wear allowing for asymmetric statistical distributions for some dimension.
 
Hi lbotez,


You're right, it does work really well, as long as your feature has dimensions. The problem here is that my feature is attached to the copied geometry of a skeleton model so there's no dimension to pick.


Is it a badpractice to attach my sketches to the surfaces of a skeleton model? I hope not, I learned to do that reading an exercise published by PTC.


Mat
 
Beats me, I guess you're stuck! Top-down design and manufacturing design sometimes just don't get along.
 
You could try making your part section reference the center (axis) from your skeleton but dimension the diameter so you have a part level dimension. You can write a relation between the part dimension and the skeleton dimension to make them the same or to add clearance. That should make both sides happy.
 
Here is what you do Create an Assembly just for the drawing. Assembly the skel and the part being dimensioned. now create a drawing of the Assembly and show the dims from the skel or the model. this also allows you to add features in the Assembly just for the purpose of a drawing.
 
But Ialso think that the best thing for your company would be tofind vendors that douse Pro-man. Its 2005 time touninstallthe 2d manufacturing software and use a true 5 axis plus manufacturing software. After all why would you want to do business with a vendor that is not smart enough to follow your lead. after all they should be able to turn most of your changes around just be hitting the regenerate button. that should save your company time and money.
 
Sorry for the late reply to your post Dr Gallup.


I tried that too (adding a relation of equal diameter instead of a constraint of equal radius)but Dim bound unfortunatelydoesn't work with dimensions that have relations. Too bad.


Thanks to all for your advices.


Mat
 
lbotez said:
You can combine these 2 modes using the "Dim Bound" command (Setup -> Dim Bound). This will allow you to show "+-" tolerances in the drawing, while having the 3D geometry regenerated at the middlevalues. It worked like a charm for me just a few days ago. <?:namespace prefix = o ns = "urn:schemas-microsoft-com:eek:ffice:eek:ffice" /><O:p></O:p>


Note that I mentioned "middle" values not "nominal" since middle and the statistical nominal values can differ.<O:p></O:p>


Besides manufacturing this will help when doing FEA's, say when press-fits are involved.<O:p></O:p>


On the other hand, the symmetric tolerancing/limit tolerancing works well if you have parts produced by cutting, say turning ormilling. When other methods are implied, such as casting, stamping, fine-blanking the symmetric/limits tolerancing doesn't suit the manufacturing process since the dies will wear allowing for asymmetric statistical distributions for some dimension.<O:p></O:p>


Thanks !!!!


This realy helped me a lot. Especially because we work on the ISO system of tolerancing.
 
Simple stated it is always a bad idea to aim for the middle of a tolerance. their is a reason the engineer has stated the tolerance the way it is. you must always aim for the nominal of the dim. If we do not tell them what the ideal designed dimensions are we will never get them. If the guys in manufacturing think of it in this fashion they will agree with it.


Example


when you setup a machine to grind a shaft you get heat and stress relief as the machine works. While it may by cut in spec as it sits it will shrink or warp. if youaim for9.95 it may end up being 9.89. you can always take more off but you can never put it back.


Sometimes the dim is the opposite of the growth or shrink of the part.The machinist needs to know what he is supposed to aim for not what would make his life easiest inan ideal world.


Besides this if you use a true Pro-E company to supply your parts they can copy all the data to a part just for manufacturing in about 2 seconds that they can shrink or grow to what ever size they need to meet your dimension requirements. when they do this it based on the idea part and will update with it when it changes.


AS for wsylvester positions it is so old school that it just hurts to think that even after most of the manufacturing in this country has gone to China peoplestill don't see that doing the same thing over and over will never get you different results.


while worrying about idle time may have been the main priority in the past. Fast paced turn around of the prototypes is the new priority. Its very simple really if you don't have state of the art equipment you wont be doing machining in this country very long. the Chinese can turn a profit on a part when all we have done is paid for the stock. besides who real wants to just pump out the same part for 2 years day in and day out. Let them have it we have been doing it for 40 years. Its time to move on and upward.
 
Just because you say re-posting is a waste of time doesn't make it so.


The facts are that you you old guys and your unwillingness to change or consider that just because you been doing something as long as you can remember dose not mean you are right in fact history has shown all of use the exact opposite to be true.


everything changes but change it self.


Now to get back to what i said before if the designer wants 10mm you don't aim for 9.95 you aim for 10mm. When something is finished being made it is to late to check it. In other words you set up a process that eliminates human error. letting the machinist dial it in is setting your self up for problems. by the time it gets to the machinist the program should already be proven out with vericut or some other program.If the program is wrong you will know it before hand and repost (no chips made yet or wasted machine time). when it comes time to check something you are not check to see if the program is good it is you are check to see if the machine is out of wak or something like that. also you are not checking it by hand you use a program just like you do machining the part . this is way faster and a hell of a lot more accurate then any guy reading a print. It will also out put a report that can be sent with the part.


The reason most of the shops in this country have shut their doors is because they have no clue period. They never bothered to update their machines, capabilities or the training for their personal. on top of this the machine shops in America constantly ignored the signs that change is coming. They milked the industry for all it was worth and now the whole country is paying for it.


can you name a law school, med, engineering, or even art. I bet you can.


Now how about a school to learn to be a machinist or trade schools of any kindthat offer a 4 year degrees. If they did would the machine shops in this country hire them or would they be to expensive.


Anyway the point I am trying to make is not that it was done on purpose but that the people in the industry such as your self have to bare the weight of responsibility for the fall in manufacturing not the government and dam sure not the Chinese. If you are going to out right blam some one or a group of people it would be the owners and employes of the manufacturing industry. If you want something to go on living you have to feed it.
 
I've read all the posts on this topic with interest, since I sit on both sides of this fence. In my current position, I do both engineering/drafting and CNC programming. So, what do I think is the best approach?


With my CNC hat on, sure, it would be much easier to have the model be at the middle, or at least within tolerance, not at the limit (high or low). Having said that, I think it's actually best to model parts to the nominal, with desired clearances built in, and tolerance the drawing appropriately. Sure, the CNC guy has to work a little harder, but it's not that painful. Besides, if you model a part in the middle of the tolerance, you're assuming that's what the machinist is aiming for. Perhaps yes, but not always. Sometimes one shoots for the low on an OD, to allow for plating, to use an example. How can the designer know what the machinist will want to do? The machinist's job is to make a part within tolerance, not hit a number exatly in the middle of the tolerance. So, Ithink the model should be to the designer's intent, and the CNCprogrammer will simply have to adjust to where within the tolerance he wants to hit.


Just my 2 cents worth (that and a dollar will buy you a cup of coffee).
 
You spend so much time putting down the idea of moving away from drawings that it seems you have forgotten that at one time managers and company's felt the same way about CAD.


Please stop telling me about cutter comp if you are using cutter comp still then you just don't understand what real 3D CAM is. You can not compensate for the length of a cutter and still cut a good 3d surface because the tools just don't ware out evenly IE length and dia on a ball or bull nose cutter. also a up to date machine will have a auto tool setter.


What you doinstead of cutter comp is figure out ahead of time how long a finishing tool will last and still make good parts. you can calculate this with speeds feeds cutter type and material being cut.


Thewhole reason youthink scrap is exceptable is because youhave never know anything else.you keep saying I know nothing about you. WRONG. I know that you have stated over and over that your quick turnaround times are driving you process. if you ever real want control of something then you have to plan first and act second other wise you are just putting out fires.


As for the how old you I DONT CARE I was just repeating what you said the post before. As for you work ethic I DONT CARE. But if you think by using the same process over and over that you are going to get different results then you are insane. At least according to Benjamin Franklin.


To get back to what this post is about the man asked what would be better and you told him what you do now. I am just saying stop think what would be better. do you even have the ability to step back look at the whole thing and say if I set up shop to day how would I do it.


Do I just focus on CAM or do I look at the company's I will be doing work with or maybe I just start from scratch and design a system that I can use any kind of input and get the desired out come.


Do you even know what vericut is??????????


Or that the CAM software you listed earlier doesn't even do 5 axis milling?


Its 2005 If you are still using coolant to cut aluminum then to be honest you have not got a clue or at least not a machine that can handle true high speed milling so why would you know.


Anyway it sound like you are decent hard working American. with well intended advice that is to be frank not at all the best way.
 
Guys, please stop this thread as it is right now. Or at least move it to "Rant and Rave".


I think all "technical" aspects were covered and let one make his choice. If aluminum milling doesn't need coolant anymore might be true, but it is sure needed in this discussion.
 
Ibotez,


I agree, but I feel the ranting or at least the insults have been coming from only one direction
Edited by: wsylvester
 
I too sit on both sides of the fence. As a machinist, I'm going to aim
for the middle of the tolerance because the middle is the farthest from
either end of the tolerance zone, knowing full well there are variables
to consider: material I'm cutting, hwo sharp is my tool, etc., that
will affect the finished size. Assuming I don't misread the print, I'm
going to give the designer what he asked for; is he sure that's what he
wants? I'm not paid to second-guess the design, I'm paid to make chips.
The designer tells me what he wants, and I decide the best way to
produce it. It would annoy the snot out of me to have a pencil-pusher
tell me how to make a hole. But I digress...



As a designer, we model our parts to the mean dimension -- meaning if
the drawing says 1" plus 20 minus 0, it gets modeled to 1" plus 10.
Most parts that go to the shop are solid models that get assembled into
a workpiece and the model inside drives the toolpath. Modeling to the
mean dimension often saves the machinist having to futz around with
cutter comp and whatnot, which saves me, the designer, money, and makes
me, the machinist, happy because I like 'clean' models I don't have to
fix, and I don't get yelled at for taking too long making chips. In
other words, if I model my parts right, the machinist is happy and the
parts cost less.



In the past we had this "us and them" way of thinking. Designers
were dreaming up stuff that either couldn't be machined or would
be terribly expensive (Yes, I can give you a square corner in the
bottom of that hole -- are you sure
you want one?), and machinists were spending too much time (and
therefore money) trying to give them what they asked for. That doesn't
even touch the problem of scrap...



Fortunately both sides have been talking to each other, and
both agree that kind of 'old-school' view has to go. Even though the designer doesn't need to know how the hole is made, he does need to know his allowable design limits. The machinist doesn't need
to know what the part is for, but if he did know, he might be able to
suggest some minor design changes that would still produce the design
intent without costing as much.
 

Sponsor

Articles From 3DCAD World

Back
Top