ATTENTION Inventor Developers! This part has repeatable crash

[Anonymous]

Open this part, unsuppress fillet12 (just above EOP). This action causes features 25 steps previous to fail. How does a fillet reach back in time to corrupt other features?

I couldn't find the email address for anyone to just send this to. If you are a developer who deals with this kind of problem, please post your email or an email I can just send parts to rather than posting. This time I will write the email down on my monitor with sharpie.

Okay- I cannot post the part here!

"The content type of the file 'Corrupt Part.ipt' is not allowed"

So if someone wants to reply with their email contact, I will be happy to send the corrupt part.
Thanks
Phil

[Anonymous]

JD- Thanks for taking the time to look at this.

Perhaps some explaining of my process will help.

This part is designed to mate with several off the shelf components. I get those from vendors, standard parts in 3D format. The parts are mathematically related to each other. I place them into an assembly in the proper spatial relationship accordingly.

There is an envelope size I must match. This part has to be no larger or smaller than a pre-determined size. Other than these pre-set definitions, there are standard shapes which must be used per vendor/industry standard- ingredients if you will. That's it, the rest is me being creative, not necessarily disciplined as a proper engineer or machinist need be.

The venturi type shape contains certain mathematical relationships. You can see that in the sketch with a floating cloud of points used as defining points for a spline. This flare is essential to the function of the finished assembly and therefore happens early in.

After that I hack my way right thru to the finished part. There are extra sketches and workplanes and all kinds of stuff that get in as a result of my hacking. Yes, hacking, that's how I do it. Most of those extra workplanes happened automatically and took no time to create.

Look- If I knew the finished size and shape of everything, and therefore didn't have to relate the separate parts, or project geometry back and forth, I too could create the thing with fewer features. But- the sheer number of features has no bearing in my hacking. There is no clock that says "Phil, you've got extra extrusions here, quit wasting your time." In the end, I get a part with the correct envelope size, the correct ingredients, and the correct special geometry that makes this thing so wonderful once it's a product. Along the way, I broke Inventor and all this mess started.

Consumer goods are a mix of standards, and creativity. Art. My artistic parts tend to be wilder, because I am chiefly concerned with a perfect appearance or rather, the result. The destination, not the path. All the constraints, or discipline in the world will not unlock that extra bit of creativity I get from flowing thru a part this way. IV gets a standing round of applause for letting me make these messes into things people look at and wish to purchase- bad process, extra sketches, too many workplanes later.

[tahdesign1]

OK so I have read this thread for the humor in it for a while now and final decided to take a look at the part in question. Where I do agree with Troy and Josh about some of the half way thought out and uncompleted features and/or upgrades ADSK has done (this website is a shining example), I also must lean toward Dennis and JD when it pertains to this actual part. Call me anal but I truly believe there is a right way and a wrong way to most anything and this modeling method leans toward the ladder. Parametric models are co-dependant on the parameters of the features in the model that were created before them. There will always come a point when the math fails in a model such as this. I am just surprised it did not happen earlier. I am just glad mechanic working on my car (extra bolts left in the engine) or framer building my house (un-nailed studs in the walls) does not have this approach to their work. I have always worked with the idea that someone coming behind me in 5 years can open one of my models and it be rock solid and ready to edit. If this model made it to completion in this state and was ever accessed again, I am sure it would fail pretty quickly. I have had to rebuild way to many models by previously engineers to get my job done in the past to let a model go through my hands that it not nailed down.

I know the original poster was not looking for a model exam here. But if you are going to post to this site for answer on a sick part, you should expect to have at least part of the answer come from your methods.

[JDMather]

> Perhaps some explaining of my process will help. After that I hack my way right thru... But- the sheer number of features has no bearing in my hacking.
I take more interest in the complex stuff like this at the boundaries of what Inventor can do. But I think you are actually working too hard and probably taking more time than is necessary. I do a lot of hacking too - but only as far as it gets me in understanding the geometry. The interrelationships between projections, constraints, parent/child features can get so complex that it really pays to do your hacking and then start over. Often several times, lest the quicksand get too shakey. With the complexity of your work I'm sure you will run into more interesting problems that Autodesk should take a look at.



Welcome to the forum. I hope you will share more.

[Anonymous]

Hey Phil, what curve are you using for the
flare?...it kind of looks like a Traktrix...

I play around with horn loading every so often and
just want to keep up with the times 🙂

[dan_inv09]

I think that most of the time it is economical to redraw the part, unfortunately you will not know when until it is to late. You see it after you spend an hour or two and still have not fixed what would have taken you 20 or 30 minutes to reconstruct better.
Someone else mentioned all the trouble reconstraining and finding what other assemblies it's used in, but the whole point is not to wait until it's too late. If you stop and redo it immediately after your first try you will be placing your better, more robust model and you shouldn't need to fix it after it's in a couple of assemblies.
I still have the mindset of just a little adjustment here and there and it will be all better. I'm working on being able to start over from scratch, I know for a fact that it's really hard, but I can also see the benefits. And the time it saves when you do need it will far outweigh the time you spend when you don't.

[dan_inv09]

I think it's inappropriate for him to be jumping on you, but poor practice is poor practice, and he's trying to say that sloppiness in any aspect of one's work might not reflect well upon the rest. (But then again, if a cluttered desk equals a cluttered mind what does an empty desk mean?)

[Anonymous]

The curve is a tractrix hybrid. We will never claim that on the box, just using the good math (ironic considering.... ) and also postitioning the tweeter to time align with the woofer. Pulse testing the machined nylon frames based on this model showed a need to move the tweeter a skosh. That's when I broke IV.

Nice to see another audio guy out there!

[Anonymous]

I am anal, (about some stuff) and it's "latter".

If you consider the high school dropout working on your car to be equal to the guy who designed it, you aren't anal enough. If you consider the skilled labor that stick built your house to be equal to the architect, well, you get it. I wonder if you'd take that as an excuse: "the wheel fell off because the designer had poor workflow and now I can't remember to bolt all 5 lugs..."

Seriously, do you think the people who listen to music out of my products give a flying hoot about how many unconstrained (un nailed) sketches went into it? Nobody is questioning the ability of the part to be an injection molded resin speaker frame, now are they? Apples and oranges.

I didn't post here looking for an answer to a sick part. The parts (8 different sizes all based on this very one) are all complete and on their way to production. No sick parts here. If you note the title of the post you might get a better clue. I don't mind people's input, heck, I even plan to tighten up my anal powers to work better in the future based on this advice, however, to say that these would "fail pretty quickly" is a huge bit of conjecture on your part.

My methods, the very same methods that produced this glitch, are also the same methods I use to edit parts like this, make minor changes, and move on. Nobody is going to open these and dig around later. They are all dumb solids (.step/.sat/.sldprt) by the time the mold guys get them- BTW I wouldn't want our vendors changing/ripping off my parts anyway!

Thanks for the input everyone.

[Anonymous]

So are you horn loading a high freq driver that was
made originally for horn loading or did you find a good "dome tweeter" candidate
for this type of loading?

I have been into Audio since equipment was still
advertised with "solid state" as a selling feature and they still had
tube testers in stores.

We have a local audio club active here in Southern
Oregon if you ever want to come down to a meeting 🙂

[Anonymous]

>he's trying to say that sloppiness in any aspect of one's work might not reflect well upon the rest.

To Gary, and Dan, yes that is what I was trying to say. Gary you've been a fantastic wealth of information on this forum, and I do thank you for that. I just hated to hear the justification for sloppiness coming from you.

>But then again, if a cluttered desk equals a cluttered mind what does an empty desk mean?

A clear mind 😉

[Anonymous]

> Call me anal but I truly believe there is a right way and a wrong way to most
> anything and this modeling method leans toward the ladder.

Hope the guy on the ladder is wearing his safety belt.

[tahdesign1]

"They are all dumb solids (.step/.sat/.sldprt) by the time the mold guys get them"

This part of the work flow is probably the only reason there has not been any manufacturing issues.

[Anonymous]

It's a dome tweeter. We are using the flare as a wave guide, not so much for loading but for good pattern control to minimize room interactions and reflections.

[Anonymous]

Maybe. Is that a bad thing? Getting stuff made with no problems.... hmmmm... sounds like success.

[dan_inv09]

tahdesign, you think the "mold guys" are just using the dumb solids as a reference to create more usable models?

[tahdesign1]

No, with a dumb solid they most likely are just importing to their cam software and using the dumb solid a base for their programming. There probably is no need to remodel anything so that is why the delivery of the dumb solid negates the original model integrity as being an issue in this case. I believe it was indicated earlier that these models are translated to SW for china. So my point was if these modeling methods were used to produce this in SW (which I doubt SW would allow) and it was then passed onto a cam software that uses the actual SW file, the models integrity would then produce issues in the design of the mold itself.

[Anonymous]

> dumb solid

Otherwise known, to professional CAD modelers, as a b-rep solid object.
I doubt anyone using Direct or Explicit Modeling techniques and
software refers to the geometry as 'dumb'. Problems associated with
b-rep integrity often manifest themselves as a failure to export or
import or prove to be unsuitable foundations for subsequent features.
If the integrity of the b-rep object (one of the reasons for "solid
modelers") is in question, and perhaps it is relevant to the particular
failure mode being discussed, developers should be asked about the IV
equivalent of Mechanical Desktop's AMFeatCheck command / variable.

> So my point was if these modeling methods were used to
> produce this in SW (which I doubt SW would allow) and
> it was then passed onto a cam software that uses the
> actual SW file, the models integrity would then produce
> issues in the design of the mold itself.

So many assumptions based on so little knowledge. Not only of what
other programs will or won't do but the precise nature and cause of the
failure is, as yet, unknown. Was a previously undetected defect in the
b-rep geometry detected or one introduced due to lack of constraint or
was a pointer to a referenced object lost or corrupted or ...? [I would
love to see an export of the model in its original, pre-cascaded
failure, state. Trying to export then import may provide a clue to
what's going on, lacking the developers' (wonder if they'll ever
provide any feedback re the specifics of the defects and reasons for
the behavior?) debug builds and analysis tools, as well as the reason
for what was apparently an existing failed fillet.]

[JDMather]

>I would love to see an export of the model in its original, pre-cascaded failure, state...
http://labs.autodesk.com/technologies/inventor_lt/

[Anonymous]

Thanks for the suggestion. I occasionally recommend LT to people wanting to
get their feet wet at no cost but I'd have to buy a new computer to meet the
install specs. Not very cost effective for me. ;^)