Inventor 2019 Axis-to-Axis Mate Constraint Default To Forced "Opposed" Solution

[jletcher]

I understand you added it from someones idea, but why could you not have left the default selection the old way? This way it does not screw up other peoples workflows?

When on the fly this thing is now a nightmare. I don't understand why it was even needed but now more clicks more wasted time for me and others.

Make it so users can default to old style please, these are things you should be thinking when making changes. I understand you wanted to please someone but don't do it at the expense of others. You can leave it just change default default to old style, I will never have a need for this new way.

This is going to drive me nuts and many users have already called asking if there is a way to default to old.

This should be easy to fix.

kelly.young has edited your subject line for clarity: 2019 New mate constraint nightmare

[Xun.Zhang]

Hi Andrew,

Thanks for sharing the point of view here, that's one option of four which was discussed internally.

We are in the same page. 

Thank you so much!

[mcgyvr]

---

@andrewiv wrote:

Maybe I'm stating the obvious here, but if Inventor just remembered the last option that was used wouldn't that solve the issue and make everyone happy?  There are a lot of commands that I wished Inventor would just remember the last selection that I used instead of reverting back to default.

---

Sadly you aren't stating the obvious..

IMO these wouldn't be issues if Autodesk was to either remember last option or allow the user to setup defaults how they want..

When Autodesk chooses one option over another they are ALWAYS leaving someone out which is a lousy decision on their part..

[andrewiv]

Since this topic is about an issue that came from idea station, here is a solution on idea station.

https://forums.autodesk.com/t5/inventor-ideas/remember-last-used-options/idi-p/8167465 

[jletcher]

Ok let me try this.

Appreciate for these valuable comments! 

No, we respect you indeed no matter who speak out opinion, please don't think in that way. We respect every opinion and everyone can share the opinion for sure.

OK, Let's talk one by one.

1) When placing parts in assembly I have the option to rotate my part on insert so the direction of my part can be done at that time if I wanted to. Undirected axis constraint will not flip the part when placed.

This is true, because Undirected just match the minimal rotation of position for both components. However, before that, you have to free rotate at least one of component to the right direction firstly. Am I right?

No you are not right, why would I have to do that when my 2nd constraint would do that?

If so, why not to leverage directional way instead of the pre-rotation of component? 

See above.

2) The new options does not control the motion along the axis. Thus making a 2nd constraint a must to control that motion, at that time the direction of the part can be placed.

Sorry, it is not that clear for me. I suspect you are talking about the constraint preview.

If you have one constraint between these two shafts such as a mate for face to face, then, at least one of the directional axis constraint (opposed/aligned) may detect the confliction. In Inventor, if a penitential confliction was defected, the preview is not working and it is the way to show the penitential confliction. In this case, if the preview is not working, please switch to another direction to avoid the upcoming penitential confliction. In fact, it is the general Inventor constraint behavior for all kind of constraint preview.

Let us leave preview out of this I do not use it, and thankful I can turn this off, so lets move paste that.

Ok, Let's start with some pros and cons for Undirected and directional.

Undirected - Minimal rotation when preview; the result can flip if the component free rotation angle > 90 degree; Additional face mate (direction protection) is required to persist the result.

Directional - Not minimal rotation due to direction option when preview; the result is persist no matter how to rotate the component; additional face mate (direction protection) is not required to persist the result.

Do you agree above assessment?

No I don't agree with it at all. Undirected or Directional you still need that additional face mate. Rather it is to make the direction or not it is needed to fully constrain the part.

Do you agree a fully constrain part is the most stable for a assembly?

If yes then you just learned why the new options are worthless and a waste off time. Because my additional constraint I put in will do the new "Directional options"  at the same time of fully constraining my part.

If no then I will never hire you because your assemblies will explode when animated.

if so, how about -

Directional - Not minimal rotation when preview by pre-determine the direction (opposed or aligned); the result is persist no matter how to rotate the component; additional face mate (direction protection) is not required to persist the result.

 You all need to stop looking at the additional face mate a direction protection, it is needed even with the worthless "Direction options" to fully constrain the part so why can't you see that the additional face mate is also the direction option all in one and the new "Direction options" really do nothing to save time but waste it.

All I want is when you add new options to ANYTHING leave the old way the default or give the option for users to control what they want.

Does that clear it up?

[johnsonshiue]

Hi Guys,

I think there is a little bit confusion among the opinions here. I want to go back to technical facts. For Assembly Constraint dialog, it does not remember the last used option. It is because the options are presented based on selection. It would be quite messy to have control over control.

It is true that the Opposed and Aligned options restrict the flipping behavior. Indeed, it can be viewed negatively. But, it was added to reduce DOF. Constraining is meant to restrict freedom. We have seen unstable solving behaviors caused by the flipping for many years. When you applied the Face-Face Mate constraint with pre-existing Axis-Axis Mate in the past and it flipped. However, it can also flip because of other related constraints leading to Face-Face constraint to fail. Without having the options, the user will need to add a zero angular constraint (Explicit Reference Vector) just to keep the direction. Please keep in mind that constraints are solved bi-directionally. Generally, there isn't a concept of master and slave between constraint participants. James and Curtis showed a very good example of desirable and predictable flipping behaviors. However, the truth is the flipping may not always be desirable and predictable.

I understand that the any change in a well-established workflow can impact some users. I am sorry for the disruption. The extra click can be perceived negatively. But, if we could turn back the clock, the Mate constraint should have these three options to begin with. The legacy Mate was sort of incomplete. The unstable flipping behaviors lead to issues that we could not resolve. I just want to provide some background on why the options are added. It is not that we are introducing new workflows for the sake of being new.

Many thanks!

[Anonymous]

In the world of constraints, if you align the axis of 2 parts there is no difference in degrees of freedom between vectors aligned or vectors opposed. One part can freely move along the axis of the other one and will continue to do so, wether you choose one direction or the other. In fact the orientation at that point is implicit, not explicit. You need a second relation that forces the orientation. Along with this second relation (orientation) you set a distance between 2 points. The ony DOF left is that the parts can rotate around the axis.

To make life easier on users it would have been sufficient to add an option "flip". Pretty much the same way as when you are redefining coordinates or create a half-section view in model environment.

It may be an impression but every time I want to put an axial part into the hole of the other part Inventor will flip it the wrong way for me. So I'm pretty close to a habit of first constraining faces or workplanes before applying an axial mate. It also keeps part from disappearing inside of the assembly.

Alex

[WHolzwarth]

Johnson and Xun, I've seen mate axes flipping by themselves often in flexible assemblies. Is the 2019 axis mate an improvement in these cases?

I think, I can live with 2019 axis mate. If only I would be able to change direction, after misplacing it at first attempt. Deleting and re-apply is worse

[SBix26]

@WHolzwarth I am able to change the direction of an axis mate from aligned to opposed and vice versa.  Simply right click on the constraint in the browser and choose Edit.  Is this not what you see?

---

Sam B
Inventor Pro 2019.1 | Windows 7 SP1
LinkedIn

[johnsonshiue]

Hi Walter,

Yes, it should help. The flipping can be desirable or not desirable depending on the cases. Also when there are many constraints involved, one cannot predict when the flipping can happen. Each constraint can contribute.

Many thanks!

[jletcher]

Please prove to me a part fully constrained on the axis, flipped.

In 15 years of using and animating in Inventor I have NEVER seen this.

And I did a search and found not ONE post about this, sure I may not be typing the right thing to find so I ask you to prove this to me. It was an IDEA not a issue with Inventor.

But here again I will say it please address me if you see this and address this one issue PLEASE..

I don't care you put the options in, I care you change my workflow and made more clicks for something I think is not needed and ask you to make the default option the old way so I don't have to do another click for a worthless option.

Did you see what I typed in red? Address that PLEASE.

[WHolzwarth]

Sam, I'm out now for some hours. I'll come back later with another test.

[johnsonshiue]

Hi James,

Indeed, talk is cheap. I owe you an example showing why the options are helpful. We do have customer examples but we cannot share. I need to figure out a way to show you.

Many thanks!

[WHolzwarth]

Ok. I see the difference here.

Double-click at 2019 axis mate looks like below.

Right-click on 2019 axis mate gives full access again, as shown by Sam in message 28.

[DRoam]

Wow, this is quite the long thread! ... I wish I had time to read through it fully, but just going off the first several posts, here's my understanding of the problem:

The problem: Axis-to-Axis mates default to the forced "Opposed" solution, which is not always desirable, as it can cause things to flip around and break the assembly. 

If that's an accurate summary, I think the solution is quite simple:

The solution:

Part A) We need to have the "Undirected" solution button available BEFORE making two axis selections, so that we can pre-inform Inventor to use that once the axis selections are made.

Part B) We need control over which solution the Constraint dialog defaults to when we initiate it, to avoid issues caused by an undesired solution being applied after the user has made geometry selections.

Does that sound like a reasonable approach?

[WHolzwarth]

I won't need that, Derek.

I could live with a pre-defined solution, perhaps opposite direction axis mate. There's a good 50-50 percentage chance, that it's right for me, without an additional click. If same direction or undefined setup is needed, than another click does it

[jtylerbc]

---

@jletcher wrote:

Please prove to me a part fully constrained on the axis, flipped.

---

Happens often in flexible subassemblies.  That doesn't meet your stipulation of being "fully constrained", but if it were fully constrained there would be no point in making it flexible.  The most common case for me is with hydraulic cylinder models that are set up to be used as flexible subassemblies.  Occasionally something will go slightly off-kilter somewhere else in the assembly, and the flexible cylinder will suddenly try to solve with the rod sticking out the wrong end of the cylinder body.  I'm not really a 2019 user yet (office is still on 2018, and my home computer recently went through a wipe so it has nothing as of now).  But if it helps stabilize flexible subassemblies even a little, I'm going to love these new options.

That being said, I think @jletcher is right in saying that most of the time there is no real need for a direction control in an axis mate.  As he mentioned, there is typically going to be a second constraint that will end up controlling the direction anyway.  Thus, the direction control is redundant.

For example, let's say we are bolting together two parts.  The base part has a hole and the second part has a slot, so an Insert constraint isn't practical.  When you create the axis mate to align the bolt with the hole in the base, it may go in either correctly or backwards to how you intended.  It ultimately doesn't matter much, because it will end up in the correct place when the face-to-face mate is applied between the bolt head and the second part.  In fact, when I first started using Inventor, it took some time for me to become comfortable with just letting the next constraint fix the orientation.

The fact that the second constraint will resolve the backward alignment isn't necessarily obvious to newer users.  Additionally, sometimes the undirected solve will result in the part getting shoved inside another part, obscuring it.  For these reasons, as well as the cases where it might help stabilize some constraint scenarios, I can totally understand the logic behind adding the feature.  However, it does seem situationally useful, rather than universally useful.  In that case, I would agree with @jletcher that the old method probably should have been left as the default, especially if the new options can introduce constraint conflicts that wouldn't have existed previously. 

[johnsonshiue]

Hi James,

You got good points here. No doubt about it. If the default was set to Undirected, this thread would not have existed. But, the thing is the other two options will not be used anyway and there would not be a point of changing the workflow. The existing Axial-Axial Mate constraints created in 2018 and earlier are set to Undirected to keep solve stability. I will work with the project team to see if we should change the default.

As I promised earlier, there are legitimate benefits of the new options. Please see attached file. I got it from the following thread. The assembly demonstrate the issue we are trying to solve clearly. If you open the original assembly, HydraulicCylinderAssembly.iam and you drag the arm a bit, you will easily get inconsistent constraint error. If you open HydraulicCylinderAssembly_2019.iam and do the same, you will see the solve is more stable. You won't get constraint error easily. This is just one of the examples.

https://forums.autodesk.com/t5/inventor-forum/hydraulic-cylinder-flipping/td-p/5273607

You and John are right. The benefit is more obvious in under-constrained assemblies. In a fully-constrained assembly, you may not see much difference. However, it depends on the change in assembly. For example, if you make a drastic change in angular value or offset distance, Undirected Axial-Axial Mate can flip.

Many thanks!

[Xun.Zhang]

Hello John, James;

Personally, still think that either option to be default is not perfect solution.

This is not something about default option change, but the mind changes instead. If axis-axis constrain start with directional way from day 1 as Face-Face constrain did, and then add Undirected in later, what’s your opinion? You may have thought Undirected is useless.

In theory, Axis-Axis do have two possible direction solutions, which is similar with “Face-Face” constraint like Mate and Flush since faces own its normal direction.

So, let’s start with “Face-Face” constraints - Mate and Flush. As you know, there is no Undirected for Face-Face constraint, is it right? Besides, Mate was selected as default option. Why nobody complains?

The same situation for Insert constraint as well, it does not contain Undirected either and Opposed is the default option. Why nobody complains?

Meanwhile, why nobody complains the exactly same behaviors introduced (include “more clicks” or “abnormal flip” or “downstream constraint confliction”) around “Face-Face” constraints - Mate & Flush or Insert – Opposed & Aligned?

“Axis-Axis” constraints - Opposed (=Mate) and Aligned (=Flush) which is the same behaves compared with “Face-Face”. Additionally, Axis direction control was missing in legacy versions. Now we add this important direction factor back to persist the unique result of constraint chain solving.

As for why Opposed is selected by default, please take a look at “Face-Face” or Insert constrain which is selected as default even 50-50 percentage chance and Opposed is the one compared with Face-Face mate or Insert.

Please correct me know if above logic is incorrect for you.

Below is a case you are asking for which was created from scratch in Inventor 2018 and every part was full constrained, however, if the design changes a little bit, the result is not persistent any more. Please check out the data from enclosed file and the demo video here.

Original (correct)

After update (incorrect)

Hope it make a little sense to you.

[jletcher]

You got good points here. No doubt about it. If the default was set to Undirected, this thread would not have existed.

True

But, the thing is the other two options will not be used anyway and there would not be a point of changing the workflow.

I am not following this, are you saying because I will not use them my workflow has not changed just because they are there? This would be correct if you left undirected the default now I have to move my mouse click another option so it does change my workflow and add time. If this is done 50 times a day you just cost me 20 minutes extra. My clients did not charge their customer this extra 20 minutes because he did not know you were going to change his workflow. Is Autodesk going to pay my client the money he lost?

This is why you don't change workflows, it is costly to the end user. 

The existing Axial-Axial Mate constraints created in 2018 and earlier are set to Undirected to keep solve stability. I will work with the project team to see if we should change the default.

 What do you mean see "if" we should change it? If this is not changed you have just proven 100% you don't care about the user and the money you cost them.

This is something you should jump on and fix asap to prove you care and understand this change cost them money. Time for you all to step up. This should be nothing more then a registry setting.

As I promised earlier, there are legitimate benefits of the new options. Please see attached file. I got it from the following thread. The assembly demonstrate the issue we are trying to solve clearly. If you open the original assembly, HydraulicCylinderAssembly.iam and you drag the arm a bit, you will easily get inconsistent constraint error.

OK thanks for the sample but I think you should have looked at it.

It does not flip anything the axle constraint did not flip, what you have here is the is no stop for the arms to over rotate but there is no flip of the constraint.

If you open HydraulicCylinderAssembly_2019.iam and do the same, you will see the solve is more stable. You won't get constraint error easily. This is just one of the examples.

Please watch the video I have attach, I had no problems making it do the same in 2019 with the so called fix.

Now the reason it did not fix it is because it is not a flip constraint issue but a under constrain assembly with no stop in the cylinder.

 I would never teach this method to anyone it is doomed to fail from the start. I find drag animation unstable in Inventor but I don't blame Inventor for that.

Again these options was NOT put in to fix anything, it was a idea from someone that had no clue how to use Inventor. If you can go back and read the tread and find my post saying this idea was stupid you would see I tried to stop it from the start, and it did not have that many votes at that.

My "Don't add anything and just fix it" had more votes then this idea.

[WHolzwarth]

I've been playing with Xun's fileset. Interesting motions can be set with only a few constraints (Use Driveme for animation).

But I need to look, why Part 4 does no continuous rotation around the Y-axis. Perhaps some task for development ..