Nominal Models Rant

[crazy^millman]

Why is it so hard for engineers to understand the concept of a nominal model? You call out a .01 edge break all around guess what when I break that edge and run it through Verification I have a collision. Call out a +.01/-.000 dimension on 600 holes and then model it to the smallest side of the tolerance and I machine it correctly to the middle to gain bonus on the Max material condition it is going to gouge. Guess what there are more drills out there than just the 118 drill point. Yes STI holes should be modeled to the correct nominal for an STI thread not the standard thread the STI is going to represent after it is assembled. Yes all STI threads require a 120 deg chamfer on the top of the threads. No I really didn't want to have to make up my own nominal model to get the Verification software to not freak out when I had the 3700 gouges in the because I programmed it correctly, but your model is not made for Manufacturability.

[civiceg]

The worst part is it is not just a few engineers. Its some of the largest companies. NG is famous for that crap, update a rev because parts don't fit together. Leave the model alone and change the profile tolerance to make it to one side. 

[Leon82]

I agree

The best is when you have to adjust a bunch of features and tangent points because of it.

[Colin Gilchrist]

I make it a standard practice to always copy the original 'Engineering Model' to a separate level, and make a 'Manufacturing Model', that I used to Toolpath the part. It takes time to re-model the features, but it makes the downstream programming and verification easier, since you don't have to check every depth and feature on-the-fly.

[crazy^millman]

26 minutes ago, Colin Gilchrist said:

I make it a standard practice to always copy the original 'Engineering Model' to a separate level, and make a 'Manufacturing Model', that I used to Toolpath the part. It takes time to re-model the features, but it makes the downstream programming and verification easier, since you don't have to check every depth and feature on-the-fly.

I have 5 copies of the model. I had a V24 model I started with and then got a V28 model and then the 3 copies for different things. 

[Colin Gilchrist]

20 minutes ago, crazy^millman said:

I have 5 copies of the model. I had a V24 model I started with and then got a V28 model and then the 3 copies for different things. 

They sure don't like to make it easy...

[Joe777]

On 10/10/2020 at 9:41 AM, crazy^millman said:

Recently, I had an argument with our customer that our drill points are not 118%. First of all, no one longer uses 118% drills bits. Secondly, they want +/- 0.005" on the drill depth. STI chamfers is a different story. In fact, I had a pleasure of installing a number of STI heli coils, I can attest that degree of a chamber does not make a difference. If the pitch is too tight,  you are not getting it in!

 

[Newbeeee™]

6 hours ago, Colin Gilchrist said:

They sure don't like to make it easy...

My experience is that most don't care.

I had one customer who fully understood and modelled everything on nominal though except threads (nominal drill size only).

Back in the day I had one customer model a manifold all on nominal and then updated the print throwing in a couple of -0/+ tolerances on there. We had to scrap the stock and WIP for that one...

[crazy^millman]

11 hours ago, Joe777 said:

 

The chamfer is also to eliminate the feather edge being left when the hole is threaded and in certain applications that becomes a FOD event and the last thing you want on a $40-$100 million rocket is any FOD(Foreign Object Debris). There are certain places where you need to watch cross material contamination. Just materials touching each other can be enough to have problems.

[jlw™]

This probably accounts for 80% or more of the geo I create.

[Leon82]

I had a model where they dimensioned radii in a pocket to the center point. They have the radius at .118 with a .120 floor radius.

I couldn't use a quarter inch ball endmill because it would have changed the center point of the radius. So i had to order 6mm ball endmills 

[huskermcdoogle]

On 10/10/2020 at 8:41 AM, crazy^millman said:

Why is it so hard for engineers to understand the concept of a nominal model? You call out a .01 edge break all around guess what when I break that edge and run it through Verification I have a collision. Call out a +.01/-.000 dimension on 600 holes and then model it to the smallest side of the tolerance and I machine it correctly to the middle to gain bonus on the Max material condition it is going to gouge. Guess what there are more drills out there than just the 118 drill point. Yes STI holes should be modeled to the correct nominal for an STI thread not the standard thread the STI is going to represent after it is assembled. Yes all STI threads require a 120 deg chamfer on the top of the threads. No I really didn't want to have to make up my own nominal model to get the Verification software to not freak out when I had the 3700 gouges in the because I programmed it correctly, but your model is not made for Manufacturability.

These types of things are precisely what keep us employed.....

On a more serious note.  Most of the things they are doing are purely based in best CAD practices, or basic CAD defaults.  We have similar things on the CAM side that we could care less about on the CAD side, and vise versa.  There has to be a happy medium somewhere.  That said, when I model fixture parts I will model to nominal size (not necessarily center of tolerance), as it makes for a much easier time to make your assemblies go together properly without major mistakes, it also makes drawing creation far faster and at least how my mind works, that is when I start to think about determining final fits and whatnot.  Once the drawings are done, I used to sometimes  go back and make a new configuration with everything in the middle of the tolerance.

A lot of this depends on if you use top down or bottom up assembly modeling techniques.  I am typically disorganized and end up with a hodgepodge mixture of both, but mainly bottom up design as I go.  I have taken some training on the top down approach, but some habits die hard and usually I'm not doing things complicated enough to where top down becomes a requirement.

It would be cool if when create a drawing, you could have the drawing tool automatically kick out an associative model, that has the surfaces which were tolerance color coded based on if it is unilateral, bilateral, or whatnot.  Then we would know to check and select tools or comp toolpaths accordingly.  I supposed if more people used annotated 3d  models as the standard it wouldn't be needed.

[Rstewart]

Just finished a Big project with garbage drawings/models.  85% single sided tolerance, 15% +/- tolerances.  Some were all Minus, some were all Plus.

 

"pocket feature 1.25" in "X"  +.02 - 0.0  Guess where the model was.....  It was more complicated than just leaving negative stock in places, if you went too far you would blow the tolerance on the Other side of a wall or overall thickness.

I feel like it's also my responsibility to have an official engineer check my "machining" model that I created.

[cncappsjames]

Ever make those parts where you have unilateral tolerances that erase ALL the tolerance because everything is RFS?

And QC says you need to hold it instead of picking up the phone?

Those Engineers need to be publicly flogged. 

3 hours ago, Rstewart said:

I feel like it's also my responsibility to have an official engineer check my "machining" model that I created.

This is why companies NEED Manufacturing Engineers that have ACTUAL manufacturing experience instead of the toilet paper they have hanging on the wall.

[JB7280]

16 hours ago, huskermcdoogle said:

ues.  I am typically disorganized and end up with a hodgepodge mixture of both, but mainly bottom up design as I go.  I have taken some training on the top down approach, but some habits die hard and usually I'm not doing things complicated enough to where top down becomes a requirement.

I'm a pretty novice solidworks user.  I can do what I need, and that's about it, but I have a good friend who is a pretty smart mech. engineer who has always told me "DO NOT use top down methods"  He says it will just get me in trouble.  I'm sure it has it's place.  But he uses all day everyday, so I usually take his advice when it's modeling related, and he listens to mine when it's manufacturing related.  

 

On 10/10/2020 at 9:41 AM, crazy^millman said:

Why is it so hard for engineers to understand the concept of a nominal model? You call out a .01 edge break all around guess what when I break that edge and run it through Verification I have a collision. Call out a +.01/-.000 dimension on 600 holes and then model it to the smallest side of the tolerance and I machine it correctly to the middle to gain bonus on the Max material condition it is going to gouge. Guess what there are more drills out there than just the 118 drill point. Yes STI holes should be modeled to the correct nominal for an STI thread not the standard thread the STI is going to represent after it is assembled. Yes all STI threads require a 120 deg chamfer on the top of the threads. No I really didn't want to have to make up my own nominal model to get the Verification software to not freak out when I had the 3700 gouges in the because I programmed it correctly, but your model is not made for Manufacturability.

 We have a semiconductor customer who almost never uses symmetrical tolerances.  Everything is unilateral, and it's always way more time consuming than it needs to be!

[huskermcdoogle]

2 hours ago, JB7280 said:

"DO NOT use top down methods"  He says it will just get me in trouble.  I'm sure it has it's place.

It can get you in trouble really fast.  Hence why I don't use it a lot of the time.  But if you do it properly, it pays off in spades.  The problem is, most people like to just start sketching, which is a no no in true top down.  The other thing with top down methods is if you don't have a robust integrated pdm / teamcenter system, multiuser on the same assembly or part is a total nightmare.  Top down enables people to work on things in teams.  Of course you need a robust CAD software to do this well.  NX, Creo, Catia, are first to come to mind for that at an enterprise level.  Solidworks isn't bad, but I think has some limitations if not implemented properly.  All CAD systems require a certain level of rules / standards and strict adherence to them is the only thing that make it work long term without major timeline setbacks.