Constraining Construction Geometry - Best Practice?

[will.astill]

TLDR: Do you fully constrain sketch CONSTRUCTION GEOMETRY?

Readers version:

When using CONSTRUCTION GEOMETRY in a sketch I always make THE CONSTRUCTION GEOMETRY fully constrained with dimensions and other constraint types so that my sketch remains fully constrained because:

 - Otherwise it's difficult to tell whether some critical geometry is fully defined.

 - I like things to be fully defined without any DoF.

It occurred to me that I don't know if this is good practice or not. Does it add unnecessary calculations for the CAD software and slow things down or does it help by fully defining THE CONSTRUCTION GEOMETRY as well as the other sketch geometry?

Note: I always fully define the rest of the geometry, this is a question about CONSTRUCTION GEOMETRY, which might for example need centre point and angle defining but not NEED a length to achieve its purpose.

I'd appreciate other's opinions on whether the additional step of CONSTRAINING CONSTRUCTION GEOMETRY is necessary.

e.g. this

vs this....

[edit]

I deleted the original sketch images and modified the question to remove context because all the answers related to how to simplify the sketch, rather than whether people fully constrain the construction geometry. 

See post later on with an example.

I'll update this in the future when I next find an example with a good screenshot.

[dave.cutting]

Have you tried making a seperate part with all of your construction geometry on it, and then deriving this into the part file for the actual component?

You can then project the construction lines that you need with worrying about accidently modifying the construction geometry.

[will.astill]

I haven't. It feels like it would be a lot more complex than just making a sketch but I'll have a play around and see what I think.

I was more interested in whether other people do or don't constrain construction geometry when they use it and their reasons though.

[swalton]

I can't speak to the sketch complexity problem, other than the basic rule of "simple is better".

Some additional ideas to make it harder to inadvertently change the datum geometry:

• You might try building a sketch block for the construction geometry.  It would help isolate the datum dimensions from the feature dimensions.
• Create datum features/surfaces/planes etc in the model tree before the geometry sketch.  Project geometry from those features into the geometry sketch.
• Create custom user parameters for all datum dimensions.  Set the sketch dimensions for the datum geometry equal to the custom user parameters.  Train yourself and other users of the model to not edit any dimension tied to a user parameter.

[will.astill]

Thanks for the reply. I usually follow the "keep it simple" rule, I just had a sketch to hand when thinking about this that was unusually complex so took a screenshot of it.

My question isn't about users changing datums or references or construction geometry. It's more of a best practice question when I use construction lines in a sketch.

I've had occasions in the past where I've used construction lines in a sketch for some valid reason (I can't think of one off hand but I remember thinking about this before - it's really annoying me that I can't come up with a valid example). In these cases, there is no actual need to define, say the length of the construction line. It's just there for whatever reason I wanted it.

For example, in this (entirely made up just for the purpose of this discussion) sketch, I've decided I need my hole to be on a line defined by two points on the sketch so I've used a construction line.

 - Does it matter that my end points can float?

 - Should I add an extra dimension to define the line fully?

(yes, I'm aware that I could just draw the line so the ends are coincident with the outline in this case - this is just an illustrative example)

[johnsonshiue]

Hi Will,

Managing sketch dimensions and constraints is an art and a science. In a perfect world, no matter what you do, it should just work. But the CAD tools are not as robust as we want them to be.

I personally prefer modeling in a strictly feature-based fashion (from coarse to detail). In this particular sketch, I would think that the most critical area is the series of the tangent arcs. I would start with a basic boxy shape with lines and add more detail along the way, instead of lumping all the detail in one sketch. Such complex sketch can diminish the criticality of the design and introduce unnecessary complexity. Also when something go wrong, it might be very difficult to make further change. Likely you will have to recreate it from scratch.

Many thanks!

[will.astill]

Hi @johnsonshiue - I have modified my original question.

I am specifically looking to understand whether constraints help or hinder when applied to construction geometry.

E.g. I might add a construction line to define an angle about a point. In that case the length of the construction line doesn't matter for solving the problem but if I don't define it then the sketch reports degrees of freedom for the ends of the construction line.

My question is whether people add the dimensions to the construction line (so it has no degrees of freedom), or whether they add unnecessary complexity to the sketches.

[CCarreiras · Accepted solution]

Hi!

I always fully constrain the sketches.

Instead of dimensions, i use mainly constraints horizontal and vertical with the remain geometry.

[will.astill]

So you would say always fully constrain the construction lines?

[YannickEnrico · Accepted solution]

Hi @will.astill 

I always constrain all my geometry fully. Whether it's mathematically and programmatically superior is not as much concern to me as having to use brain power distinguishing which sketches are ok not being fully constrained and which aren't.

If you always constrain fully, you'll be able to tell at a glance something is wrong. If you don't, you'll have to inspect the sketch to figure out whether it matters or not.

[CCarreiras · Accepted solution]

---

@will.astill wrote:

So you would say always fully constrain the construction lines?

---

The construction lines belong to the sketch... so yes.
First a try to create the geometry and only use construction geometry if i really need. Most of times, using the constraints wisely, i don't need to use many construction geometry.

At the end of the sketch, i check how many dop i need to manage to get the fully constraint sketch, and usually the free dof are due the construction geometry, so i use constraints vertical/horizontal (among others) to turn the sketch "fully constrained".

It's a lot more faster then use "dumb" dimensions, and this will avoid add "dumb" parameters to the parameters chart... but... in rare cases, sometimes i need to use a dimension to constraint a construction geometry, but i avoid that at all cost.

[will.astill]

Lovely. I do the same for pretty much the same reasons.

I've been training up a new apprentice recently and insisting on fully, fully constrained sketches.

The process of explaining why to someone new made me question whether what I'm insisting on is just my idiosyncrasy or whether my reasons were valid ones.

It's good to get a bit of validation from others that I'm not just a grumpy old man insisting everyone does everything my way 🙂

[CCarreiras · Accepted solution]

Not Grumpy at all... this is a basic rule for me.

Besides, if you have the policy to get the sketches F.C, you can just look to the browser to check if the sketch is stable, or need something.

[YannickEnrico · Accepted solution]

it's not being a grumpy old man.

Computers are easy to work with, in that they do nothing more and nothing less than what we tell them to.
How can you expect a sketch to behave properly in the future when edits come, if you haven't even told it how to behave?

As has been discussed and mentioned above, it's such a key thing, that you can tell if it's constrained properly by:
A: Colour of lines
B: Bottom right that tells you exactly how many is needed
C: Functionality to show degrees of freedom of geometry
d: The pin needle (what's the word?) icon on the sketch in the browser

It's so important it's one of the features with the most indicators around the program.

[jtylerbc · Accepted solution]

---

@will.astill wrote:

My question is whether people add the dimensions to the construction line (so it has no degrees of freedom), or whether they add unnecessary complexity to the sketches.

---

My thinking on this is that even if there were performance benefits to omitting those constraints, that time gain is so miniscule that it is immediately lost the first time you're troubleshooting a problem in the sketch and are mislead by the number of degrees of freedom. 

If you have to stop and think whether those "2 Dimensions Needed" are real or just under-constrained  construction geometry, even one time, the "benefit" of omitting them has already been lost.  Fully constraining the construction geometry makes future sketch issues easier to diagnose, because you don't have any such extra degrees of freedom clouding whatever the real issue is.

Similar to what @CCarreiras said, I typically use geometric constraints to do this rather than dimensions.  This is a little bit visually cleaner in the sketch, doesn't add a parameter, and will update to geometry changes without requiring you to change the dimension value.

[Frederick_Law · Accepted solution]

100% fully constrained.

Master Skeleton/Sketch user since 5.3.

Unconstrained could move on update and cause unexpected results.

[will.astill]

Thank you to all. I feel validated with such a definite response.

👍👊