Parametric Model Breaking Project and Offset Functions

[dallinN5N4D]

Hello,

I've been have a bunch of issues with a parametric model breaking the project and offset functions whenever I change the parameters. In this model I created a vertical section by projecting off the horizontal section then offsetting that projection. When I update the parameters the projection or offset seem to get "lost". It looks like it breaks the offset dimension into different dimensions as well. I don't understand what's going on. Any help or suggestions would be greatly appreciated.

The first is the original part, the second is the same part with new parameters loaded.

 

 

[TrippyLighting]

Can you explain the thought behind that dimensioning scheme?

If I had to design this shape I would not dimension it this way, but there might be a specific purpose behind that hasn't quite revealed itself to me.

Either way, I've attached a file with a different workflow not only avoiding the buggy sketch offset but avoid the second sketch altogether. 

@Phil.E Here's a puzzle for the sketch team.

[Phil.E]

Is there a step I can take with the first file to make what is shown in the second file?

[dallinN5N4D]

Thanks for the help! I had no idea the split faces function existed, that is awesome. That should work for what I'm doing.

The idea for the dimensioning method is based on the how the measurements are collected for the part being made. The points 1-6 are triangulated from two known, fixed points. (a and b) The goal is to fit the part being made onto an existing object. Because I'm taking measurements from a real world object and putting them into a model I had to find a way to accommodate the variance in the measurements. Arcs with "floating" endpoints, coincident with the "midpoints" (2 and 5) of the curve is the best solution I've found so far. Accurately measuring the actual radii of the object isn't a realistic option.

This brings up the next issue I haven't been able to figure out. Updating the parameters appears to break the concentric constraint on the right (b side) radius. Any thoughts?

Thanks again for the help! 

[dallinN5N4D]

I'm using the parameter.io extension to update the parameters in the model. I'll attach the parameter files. Is that what you're asking?

[Phil.E]

I'm just trying to understand the model and what is expected to update before I pass it to the sketch team. I have enough for a report. Thanks for posting this.

[TrippyLighting]

---

@Phil.E wrote:

Is there a step I can take with the first file to make what is shown in the second file?

---

It was easy to reproduce at the time, but I cannot reproduce it now.

[Phil.E]

I changed some of the "raw" values that aren't formulas by just a little bit and it gave a warning for the second sketch. That doesn't seem right, but it was repeatable for me.

[TrippyLighting]

---

@dallinN5N4D wrote:

...

This brings up the next issue I haven't been able to figure out. Updating the parameters appears to break the concentric constraint on the right (b side) radius. Any thoughts?

---

I think the parameter update creates a situation where the dimensions are conflicting with constraints. While I understand the thinking, that is exactly why I would not dimension it this way.

I would definitely not use a plugin to update all parameters at the same time, because that makes debugging the sketch impossible. 

I would use a simple dimensioning scheme and then use driven dimension to check the dimensions you are interested in.

[laughingcreek]

I completely agree with @TrippyLighting , A simplified approach is needed here to prevent dimensions from conflicting with each other when making changes.

---

@dallinN5N4D wrote:...The idea for the dimensioning method is based on the how the measurements are collected for the part being made.

---

would be interested in what the physical means of collecting measurements are.

---

Because I'm taking measurements from a real world object and putting them into a model I had to find a way to accommodate the variance in the measurements. Arcs with "floating" endpoints, coincident with the "midpoints" (2 and 5) of the curve is the best solution I've found so far.

---

how are you "accurately" determining the mid point of the curve on the real-world part inorder to take your measurement?

---

Accurately measuring the actual radii of the object isn't a realistic option.

---

They do make gauges for measuring radii, albeit pricey.  but if accuracy of the curve is all important, it is your best bet.  if this is a one off, I've had surprisingly accurate success making my own gages cut  from card stock with a compass w/ blade attached.  

2 of the most important dimensions in this shape are the depth and the angle of the side walls.  in your setup these are inferred by the various triangular measurements.  measuring these directly would induce less error.

[dallinN5N4D]

Thanks for the feedback,

I guess I should better explain the purpose of this model and the dimensioning scheme. The model will be used to make custom fitted parts. The goal is to use measurements that are simple to take and easily repeated, then update the model with these measurements. I have a little input on what measurements are taken and 'll see if I can come up with a way to use a simpler dimensioning scheme, but based on initial tests the current method works very well in every step but modeling. I can go back and manually fix the problems with the model for each iteration, but this will be a high enough volume product that doing this isn't feasible. I'd really like to get the model to work every time without having to significantly alter the dimensioning scheme. (Though I'm realizing this may not be possible.)

The measurements are taken using a laser measuring tool. A marker is placed a points a and b, then the laser measures the distances from points 1-6 to the markers. Points 1,3,4, and 6 are supposed to be at the endpoints of the radius, but it's difficult to get that exact. I used floating endpoints in the model to account for that. Points 2 and 5 are supposed to be at the midpoint, but again it's not exact so I make the points just intersect the arc in the model rather than be actual midpoints to try to account for this.

 

I'll look into radii gauges, but there is a lot of variance in the sizes (anywhere from around 7" to 20"+) which might make them a difficult option to use.

[dallinN5N4D]

I guess it makes sense that as the parameters update it will create conflicting geometry and dimensions. For this reason I can't really update parameters one at a time and check for errors, almost every time I update an individual parameter it creates impossible geometry.

With all the new parameters updated I can go back and delete and re-create the tangential and concentric constraints to get the right geometry. I guess I was expecting the model to update correctly because technically everything is correct, but it looks like things break along the way to being updated.

But if creating impossible geometry as parameters update is the issue, won't that still happen even with a simpler dimensioning scheme? I need to play around with that more and see...

Thanks for all the help!

[laughingcreek]

since you can't determine the exact endpoints of the curves or midpoints, it seems to me your inducing error in your method from the start.  errors will just get compounded from there.  which is on top of a dimensioning scheme that will cause conflicts when making adjustments.  the three measurements I indicated before are measured directly, least room for error. add a decent value for the radii and your set. 

these work-

https://www.amazon.com/CGOLDENWALL-indicator-Measurement-changeable-measuring/dp/B0127I4KHQ/ref=sr_1...

and these are fun to have i the shop-

https://www.amazon.com/Baileigh-IRG-3600-Inside-Variable-Universal/dp/B00F6GVOC6/ref=sr_1_21?crid=1P...

and there are several other types.