unable to mirror variable fillet

[Gil Alsberg]

Every time I try to mirror a Variable Radius Fillet using the "Features to
Mirror" option, solidworks replies with an error message: "Unable to create
instances for the pattern. Try again by decreasing the number of instances
or the distance between instances.".
When I try it with the "Faces to Mirror" option, solidworks replies with an
error message: "Could not attach patterned faces to the solid to form a
closed body.".

There wasn't a single case were one of this two has worked regarding
variable fillets, and it seems like a repeating malfunction.

How is your experience regarding variable fillets feature and the
mirror/pattern feature combination?

[e***@juno.com]

Post by Gil Alsberg
How is your experience regarding variable fillets feature and the
mirror/pattern feature combination?

There are two basic categories of features (perhaps someone will chime
in with more if I am missing something): sketch based features and
applied features,

Sketch based features are obvious (they have sketches under them!).
These tend to pattern well (though I remember that at one time there
was a restriction on features based on multiple sketches, like sweep
and loft - I will explain in a minute why I don't really know if this
limitation still exists)

Applied features are items like fillet, draft, chamfer, etc, that are
applied to geometry that already exists.
As a general rule, the patterning and mirroring of applied features (in
my experience) will fail in most cases. Some things - like a certain
draft that I remember - will fail to pattern or mirror and won't even
give you an error; it just won't be applied and it is your
responsiblility to identify the failure.

So... rule of thumb: don't pattern applied features

OK, my minute is up - This response sounds a little tentative on
details because years ago I simply adopted the patterning of bodies as
my 'best practice' so I didn't have to deal with the vagaries and
quirks of whatever version of SWx I was working on. Can't pattern a
loft, a fillet, a draft, or whatever? No big deal - just
mirror/pattern the body instead. This has tended to work so well for me
that I have not felt any need to slog through what features will or
will not pattern/mirror today.

One additional benefit to patterning/mirroring bodies is (and I
encourage you to check this for yourself using 'feature statistics'),
the rebuild time is significanlty less when patterning bodies than when
patterning features.

Other than a feature used for mating (so I can use a feature based
component pattern in my assembly) I probably mirror a 'feature' once or
twice a year. I pattern or mirror boides multiple times a week.

Ed

[Gil Alsberg]

Ed,
Thanks for your insights - they are really helpful! yet one fundamental
question still hovers in the background - why the hell solidworks lets you
to try and pattern/mirror applied features, if it will fail to do so at the
end anyway?! I guess the intention of the SW guys was that this will work
too, but that seems to be impossible for them until now.

mirroring/patterning surfaces/faces seems to work for me too, although
it is a cumbersome method, which requires me to trim and replace all
involved original faces with the new patterned/mirrored surfaces/faces ( +
some additional steps and features)- so this comes to me as a surprise that
the rebuild time turns out to be quicker at the end, then with features
patterning/mirroring in which one can skip those extra steps (I will try it
myself as you suggested).

at the bottom line, I think solidworks should improve the success rate
of feature patterning in cases where applied features are involved. until
then I guess I will do the same as you.

Regards,
Gil

Post by e***@juno.com

Post by Gil Alsberg
How is your experience regarding variable fillets feature and the
mirror/pattern feature combination?

There are two basic categories of features (perhaps someone will chime
in with more if I am missing something): sketch based features and
applied features,
Sketch based features are obvious (they have sketches under them!).
These tend to pattern well (though I remember that at one time there
was a restriction on features based on multiple sketches, like sweep
and loft - I will explain in a minute why I don't really know if this
limitation still exists)
Applied features are items like fillet, draft, chamfer, etc, that are
applied to geometry that already exists.
As a general rule, the patterning and mirroring of applied features (in
my experience) will fail in most cases. Some things - like a certain
draft that I remember - will fail to pattern or mirror and won't even
give you an error; it just won't be applied and it is your
responsiblility to identify the failure.
So... rule of thumb: don't pattern applied features
OK, my minute is up - This response sounds a little tentative on
details because years ago I simply adopted the patterning of bodies as
my 'best practice' so I didn't have to deal with the vagaries and
quirks of whatever version of SWx I was working on. Can't pattern a
loft, a fillet, a draft, or whatever? No big deal - just
mirror/pattern the body instead. This has tended to work so well for me
that I have not felt any need to slog through what features will or
will not pattern/mirror today.
One additional benefit to patterning/mirroring bodies is (and I
encourage you to check this for yourself using 'feature statistics'),
the rebuild time is significanlty less when patterning bodies than when
patterning features.
Other than a feature used for mating (so I can use a feature based
component pattern in my assembly) I probably mirror a 'feature' once or
twice a year. I pattern or mirror boides multiple times a week.
Ed

[v***@verizon.net]

ed1701 overstates the case, and "throws out the baby with the bath
water". It is true that if you try to mirror things like fillets or
chamfers on their own, it won't work, and shouldn't be expected to.
Also things like shell or scale or move can't be mirrored on their own.
I also agree that patterning bodies is faster. None of this is new
information, it has existed for years from seemingly less dogmatic
sources, since disjoint bodies became available.

Still, it is incorrect to say that mirroring or patterning fillets and
chamfers "fail in most cases". This is simply wrong. When a series of
features is patterned or mirrored together, they work together. An
extrude with other attached features which has fillets around the
extrude works fine for patterning and mirroring. The fillets even work
in many cases when they go between a patterned feature an a feature
that is not patterned. In part, this sometimes has to do with knowing
when to use the "geometry pattern", and when not to, which I admit can
be a bit of voodoo. In addition to that, not every design is well
suited for mirroring/patterning bodies. Plus, the merge functions in
mirror and pattern are poorly implemented and give results or not with
baffling irregularity, often necessitating an addition combine feature,
which fails if the pattern changes.

Throwing away tools due to superstition is counterproductive.

Daisy

[e***@juno.com]

Post by v***@verizon.net
ed1701 overstates the case, and "throws out the baby with the bath
water". It is true that if you try to mirror things like fillets or
chamfers on their own, it won't work, and shouldn't be expected to.
Still, it is incorrect to say that mirroring or patterning fillets and
chamfers "fail in most cases". This is simply wrong.

patterning an individual filet is the question I was responding to. So
when you say 'it is simply wrong' that I said it fails in most cases,
thats a fair accusation - it likely fails in ALL cases when applied
like Gil attempted (though I am always hesistant to say SWx can't do
something because I haven't tried 'all cases' and SWx has a way of
pulling suprises)

Post by v***@verizon.net
When a series of
features is patterned or mirrored together, they work together.

Good observation and I think its great you added this to the
discussion.

But... (and there always seems to be a 'but' with SWx) I have seen too
many cases for my comfort where it didn't work .

For instance, there was that draft that I mentioned - the draft was
patterned with the other features it was applied to, yet the patterned
features were not drafted - the draft was simply dropped, no error
message. If i weren't disciplined in checking features after they are
created (and I'm not always as vigilant as I should be - this case was
a reminder to me to do it more) I would have missed the error and
continued with a model that did not follow my design intentl

Another case IS fillets. I have never been able to work out the magic
bullet to get them to work every time. I had a theory that it might
have to do with the order that I selected the features, or maybe things
went south if I edited a pattern or mirror feature to add the fillet
into the set. I beleive problems generally arose was when I had more
than one filet feature in the set and the order was critical, and
theorize that SWx did not follow the same order in the pattern as the
features originally appeared in the tree. Regardles, there were too
many times that I saw a fillet that caused a pattern/mirror to fail and
could not see any reason why.

So I found it more predictable, reliable, and productive for me to
adjust my thinking towards patterning bodies instead, which works
pretty consistently (I can't think of an exception right now, though
that surely doesn't suggest that there aren't exception. Daisy
mentions issues he/she had with merging bodies ). PLUS, patterning
bodies is (typically - again, I haven't tested all cases) faster since
SWx only has to move geometry, not recalculate everything - so to my
thinking it is a good best practice to shoot for.

Post by v***@verizon.net
. None of this is new
information, it has existed for years from seemingly less dogmatic
sources, since disjoint bodies became available.
Throwing away tools due to superstition is counterproductive.

Huh? Dogmatic -'Characterized by an authoritative, arrogant assertion
of unproved or unprovable principles'. Superstition? I beleive I
simply related my experiences after using SWx where I have tried out
various options.
Again, One method is predictable and reliable -for me- the other has
(in my experience) been much less so.
Just trying to share, not dictate.

Regards,
Ed

[e***@juno.com]

Post by v***@verizon.net
ed1701 overstates the case, and "throws out the baby with the bath
water". It is true that if you try to mirror things like fillets or
chamfers on their own, it won't work, and shouldn't be expected to.
Still, it is incorrect to say that mirroring or patterning fillets and
chamfers "fail in most cases". This is simply wrong.

patterning an individual filet is the question I was responding to. So
when you say 'it is simply wrong' that I said it fails in most cases,
thats a fair accusation - it likely fails in ALL cases when applied
like Gil attempted (though I am always hesistant to say SWx can't do
something because I haven't tried 'all cases' and SWx has a way of
pulling suprises)

Post by v***@verizon.net
When a series of
features is patterned or mirrored together, they work together.

Good observation and I think its great you added this to the
discussion.

But... (and there always seems to be a 'but' with SWx) I have seen too
many cases for my comfort where it didn't work .

For instance, there was that draft that I mentioned - the draft was
patterned with the other features it was applied to, yet the patterned
features were not drafted - the draft was simply dropped, no error
message. If i weren't disciplined in checking features after they are
created (and I'm not always as vigilant as I should be - this case was
a reminder to me to do it more) I would have missed the error and
continued with a model that did not follow my design intentl

Another case IS fillets. I have never been able to work out the magic
bullet to get them to work every time. I had a theory that it might
have to do with the order that I selected the features, or maybe things
went south if I edited a pattern or mirror feature to add the fillet
into the set. I beleive problems generally arose was when I had more
than one filet feature in the set and the order was critical, and
theorize that SWx did not follow the same order in the pattern as the
features originally appeared in the tree. Regardles, there were too
many times that I saw a fillet that caused a pattern/mirror to fail and
could not see any reason why.

So I found it more predictable, reliable, and productive for me to
adjust my thinking towards patterning bodies instead, which works
pretty consistently (I can't think of an exception right now, though
that surely doesn't suggest that there aren't exception. Daisy
mentions issues he/she had with merging bodies ). PLUS, patterning
bodies is (typically - again, I haven't tested all cases) faster since
SWx only has to move geometry, not recalculate everything - so to my
thinking it is a good best practice to shoot for.

Post by v***@verizon.net
. None of this is new
information, it has existed for years from seemingly less dogmatic
sources, since disjoint bodies became available.
Throwing away tools due to superstition is counterproductive.

Huh? Dogmatic -'Characterized by an authoritative, arrogant assertion
of unproved or unprovable principles'. Superstition? I beleive I
simply related my experiences after using SWx where I have tried out
various options.
Again, One method is predictable and reliable -for me- the other has
(in my experience) been much less so.
Just trying to share, not dictate.

Regards,
Ed

[e***@juno.com]

Post by Gil Alsberg
- why the hell solidworks lets you
to try and pattern/mirror applied features, if it will fail to do so at the
end anyway?!

Because sometimes it works! - as Daisy points out elsewhere in this
thread, patterning fillets and other applied features CAN work if the
thing they are applied to is also patterned in the same pattern
feature.

Post by Gil Alsberg
mirroring/patterning surfaces/faces seems to work for me too, although
it is a cumbersome method, which requires me to trim and replace all
involved original faces with the new patterned/mirrored surfaces/faces ( +
some additional steps and features)

Maybe I didn't make myself very clear - that happens sometimes. I tend
to pattern or mirror bodies, not faces. I don't know that i have ever
patterned faces (other than playing with it when it first came out).
I'm not throwing it out, I just have such success with bodies that I
haven't needed to go elsewhere (but certainly don't discourage anyone
who uses it if it works for them - it just seems extra complicated to
my mind for the reasons you state)

Post by Gil Alsberg
- so this comes to me as a surprise that
the rebuild time turns out to be quicker at the end, then with features
patterning/mirroring in which one can skip those extra steps (I will try it
myself as you suggested).

If you do test it, please think to share the results - I am little
curious how the rebuild time of patterning faces works in comparison to
features and bodies,

Good luck,
Ed

[Life in Mono]

Ed again - great insights !!

but why-oh-why does this have to come from you and is not included in
the manuals or even help.?
I think I know the answer - to keep it all looking nice and simple -
but then why not have 2 versions of manuals etc. basic & advanced
!!.......Another one for the top 10 Solidworks improvements (unless
that was posted by say Acad .... ohhhh cynic).

[Bo]

I think there is some actual construction logic behind which trying to
mirror a fillet, say on the corner of a 'box' won't work.

When you want to mirror the feature, the feature is internally defined
in terms of all the surfaces it connects to on the solid. The fillet
does not "stand on its own". It is defined by its dependencies.

Thus when you try to mirror only that feature, the definition of what
that feature connects to limits it to ONLY what it is connected to.

If you create only one quadrant of the box, put all the drafts and
radii on it and then mirror twice, you can get the full 'box', as I
have done many many times.

I just did it again with both constant radii and variable radii on a
quadrant of a symmetrical box and then mirrored them all successfully
twice to form the finished box. It is my preferred way to construct
symetrical objects like this.

Bo

Post by Life in Mono
Ed again - great insights !!
but why-oh-why does this have to come from you and is not included in
the manuals or even help.?
I think I know the answer - to keep it all looking nice and simple -
but then why not have 2 versions of manuals etc. basic & advanced
!!.......Another one for the top 10 Solidworks improvements (unless
that was posted by say Acad .... ohhhh cynic).

[TOP]

Ed,

I do believe patterning of applied features is covered in the
Essentials training in one of the exercises. In that case a rib is
patterned around an axis and prior to patterning there is a
fillet/round applied to the rib. Are you limiting this to mirrors or
mirror/patterns in general?

Post by e***@juno.com

Post by Gil Alsberg
How is your experience regarding variable fillets feature and the
mirror/pattern feature combination?

There are two basic categories of features (perhaps someone will chime
in with more if I am missing something): sketch based features and
applied features,
Sketch based features are obvious (they have sketches under them!).
These tend to pattern well (though I remember that at one time there
was a restriction on features based on multiple sketches, like sweep
and loft - I will explain in a minute why I don't really know if this
limitation still exists)
Applied features are items like fillet, draft, chamfer, etc, that are
applied to geometry that already exists.
As a general rule, the patterning and mirroring of applied features (in
my experience) will fail in most cases. Some things - like a certain
draft that I remember - will fail to pattern or mirror and won't even
give you an error; it just won't be applied and it is your
responsiblility to identify the failure.
So... rule of thumb: don't pattern applied features
OK, my minute is up - This response sounds a little tentative on
details because years ago I simply adopted the patterning of bodies as
my 'best practice' so I didn't have to deal with the vagaries and
quirks of whatever version of SWx I was working on. Can't pattern a
loft, a fillet, a draft, or whatever? No big deal - just
mirror/pattern the body instead. This has tended to work so well for me
that I have not felt any need to slog through what features will or
will not pattern/mirror today.
One additional benefit to patterning/mirroring bodies is (and I
encourage you to check this for yourself using 'feature statistics'),
the rebuild time is significanlty less when patterning bodies than when
patterning features.
Other than a feature used for mating (so I can use a feature based
component pattern in my assembly) I probably mirror a 'feature' once or
twice a year. I pattern or mirror boides multiple times a week.
Ed

[e***@juno.com]

Post by TOP
In that case a rib is
patterned around an axis and prior to patterning there is a
fillet/round applied to the rib. Are you limiting this to mirrors or
mirror/patterns in general?

Paul,
Beats me. I am no authority on this .

What I do is mostly informed by Jim Wilkisnons FAST presentation (or at
least the spririt of it) at SW World in new Orleans, and a memorable
'whats' new' demo a few years back where they showed patterning of
bodies to overcome limitations in patterning of features. Both of these
touchpoints could rightly be called-out as pretty out of date.

If I have to make a symetrical part, my general practice is to build
half then mirror the body. If the part is 'mostly' symetrical, I build
the portions that are symetrical, mirror that body, then add the unique
features on each side as needed. So I almost never need to mirror a
feature (I can't recall the last one I tried... it just hasn't come up
with the kinds of stuff I get asked to do), and when I have I manuever
it so I can mirror a body instead and carry along a lot of features
without worrying about limitations or suprises.

When I am tempted to pattern more than a single sketch-based feature,
its usually a relatively complex series of features to make mating
details on stacking items &stuff like that, that look
pretty-clearly-not candidates for feature patterning. I don't run into
much stuff that is similar to the models in the essentials.

Good candidates for patternng are mating features like holes used for
assembly mates. Or craploads of holes (perfs where I need ot see the
perf or my vendor needs the geometry) Rib features can be, but most of
the time my ribs are not on even floors or cover more than one face,
and I need to use parting line draft at the root of the rib to contol
my wall thickness so the applied feature (the draft) wouldn't be a
candidate for including in the pattern.

Ed

[e***@juno.com]

Paul,
Right after I posted this last message it occured to me that you have
in your posession a good sample of the vagaries of patterning features,
but had to wait to revise my post because I had to leave to watch some
fire-works and drive some Russians around (long-ish story)

When I responded to your modeling contest in March (or thereabouts) I
sent you a sample that failed inexplicably (to me) - a feature pattern
using 'countour select' would not pattern, but when the feature was
made from just a sketch (that contained only the contour needed to make
the feature) the pattern worked just fine.

When playing with a contest on my time I can roll the dice and see what
happens. I can target fewest features and fastest rebuild time, not
fastest modeling time. But when working for a customer, I can't go for
anything other than fastest total MODELING time for the job -
otherwise, I would be sitting at the craps table pissing away someone
elses money, not mine (and that just aint right). If its a long job,
fastest rebuild time becomes part of the equation. No one makes more
money - and final products sitting on a shelf simply don't care - if
there are a fewer features (unless that impacts editability or rebuild
time - damn, I hate having to add all these disclaimers).

If I had to develop a theory for the discrepancy between the
reliability of feature patterns vs body patterns, it is that some
programmer or group of programmers have to work to make feature
patterns work for and deal with every variation of everything that SWx
has to offer. This is what - maybe 50-100 features now (or more - boy
I don't want to have to count), plus all of the variations of geometry
plus variations of selection offered by contour select, smart select,
and things I can't think of right now.

When patterning a body, to my mind, those same programmers only have to
PATTERN THE BODY - they have one, completely self-contained thing to
deal with.

As Daisy RIGHTLY pointed out, patterning bodies comes with its own
issues - but I have a bit of experience with the BREP and always can
get bodies to fit in (another really long post or lesson - darn, these
disclaimers suck)

So, for me, the question is to deal with the devil I know and can
control reliably, or deal with the devil I can't know (and am unsure if
I could ever know). So I default away from techniques that I have seen
produce spotty results, even once (like contour select), and to prepare
for patterning bodies any time there is any doubt that the pattern will
work (like when parting line draft or fillet are involved). And when
its a mirror, I go for rebuild time - build half of the model and
mirror that body instead of mirroring the features which need to be
recalculated.

Does this make sense? I'm curious about your experience and the
experience of others - I bet this could be a fascinating and
educational thread.

-Ed