Face Mirror vs. Face Pattern Behavior

Makoto_Nara

Have an issue where a face mirror is failing to regenerate and I am a little confused as to why as attempting to do something similar with a face pattern works. Apologies in advance if this is a well known issue, a couple google searches and a search through the onshape forums did not yield a similar question (but I wasn't 100% sure what to search). 

Here is a test part studio where I made an extremely simple example: https://cad.onshape.com/documents/6c23fbdbfd8ffc196040db6e/w/9a6ff10f8c362454572ed2f4/e/33f6e5669233f4db62550ae8

The issue is that this mirror fails when I try to mirror these three faces across the right plane, creating a symmetrical notch: 



However, attempting the same process with a linear face pattern is successful. This is confusing as it indicates a problem with mirror rather than with the underlying feature/faces that I am trying to mirror. Does anyone have insight into why this might be a problem and what I should do differently? There are several workarounds (like the linear pattern) but the mirror is a lot cleaner, robust to changes, and thus preferred. 

S1mon

It's tricky to tell exactly what the issue is because this was shared via a link, not made public. We can't fully review what's going on with your model.

However, I made a really simple model of something similar and honestly I do not know why a Face Mirror fails. I wouldn't normally use one for something this simple - I would be more likely to do it in the sketch itself or just mirror an Extrude/Remove feature, not its faces. That said, part of the reason is that Face Mirrors are often very picky, and in this case it seems silly. What I can see is that face mirrors that don't break any edges of the existing topology - e.g. an internal round hole - will work. Mirror 1 is trying to mirror the 3 faces of the notch - it fails. Mirror 3 mirrors the cylindrical face of the hole. It works.

Here's the file in case any one wants to look at it:
https://cad.onshape.com/documents/5a99ec6b89b52bf574bfdb29/w/dabf50fe2c0eaecbcc27c974/e/33109e1727397a1a38c6a1dc

I feel like this is a really basic bug that would have been fixed a long time ago if it wasn't difficult based on what Parasolid can do.

nick_papageorge073

I often like to extrude as a new part. Add draft/rounds/etc to that part. Then mirror/pattern the part, rather than face or feature. You can add/cut during the mirror/pattern. Its more reliable IMO, and I believe it also takes the least computing resources.

That's one workaround that is good to know for other use cases as well.

jelte_steur814

I'm going to speculate, but usually face pattern/mirror features need to stick to the original face their parent is interfacing with.
They are the quickest to compute but because of this aspect not very robust. One identity change upsteam and they fail.

so in this case, because the middle slot interupts the face at the top, the mirrored faces are interacting with a new top face, rather than the same one. 

You could check this by reversing the order of the features and put the middle slot downstream of the mirrored slot in the feature tree. i'm betting it will work then.

a 'feature' mirror pattern is probably more robust. if you set it at 'reapply features' it will be slow but robust.

Nicks option is also very robust. set the original feature to "new" and mirror/pattern the new part and subtract there or later. Very robust but it does take more compute. (which may not be an issue at all for a simple part).