My list of what's missing.

traveler_hauptman

I just finished a job for a client that would have been a good test for Onshape. No existing CAD already in use. A minimal set of output information to communicate. But the capability to communicate the info I needed to just wasn't there. Bummer.

I have a list on my whiteboard of the features I'm waiting for before I am willing to use Onshape in a real project. Here's my list:

• drawings (with export to PDF and DXF)
  
• common parts (parts shared between workspaces)
• configurations (tables, equations)
• annotations in sketches (mostly to leave design intent notes to myself or other designers)
• traversal of mate relationships (which mate belongs to which part and vice versa)
• better traversal of feature relationships (Which features depend on this one, which features drive this one.)
• better sketch constraint discovery and management (trying to fix broken constraints in a recent sketch is difficult, trying to fix broken constraints in a sketch I'm seeing for the first time would be a nightmare)
• grouping: (sketch elements, features, workspaces, users)
  

The top 4 are capabilities that are essential to my ability to communicate my designs and to have confidence that I'll be able to leverage todays work for tomorrows project. The next 3 are user interface things that essential to working efficiently with complex designs.

The final issue, grouping, is another thing needed for working with complex designs. I've seen some interesting conversations here on the forum about grouping. When I say group I mean sets of both elements and (sub)sets.

Grouping workspaces allows one to reduce the total set of all workspaces into manageable chunks, and when combined with good naming can help with new team members understanding what subsets of the total set are relevant. Tags can do this a little bit and should be included for those who like them but the mechanics of tags (tags are not normally tagged themselves in the way that groups are often added to groups. tags are typically accessed via recall in contrast to recognition for groups) make them less capable than groups. In addition to aiding in the discoverability, the ability to manipulate groups and subgroups of workspaces could make some tasks such as sharing and deleting easier.

As a side note,  I think that sometimes people have the misconception that grouping implies being limited to a hierarchy (a DAG for you CS buffs). It does not.

User grouping makes granting and revoking permissions on workspace groups (and one day individual branches I hope) tractable.

Grouping features can be handy for speeding the work-flow (o-ring grooves, mechanical interfaces, hole patterns, etc) when parametrized, but mostly I just want to compress the feature tree to meaningful chunks that can be comprehended and navigated quicker than the full list.

Grouping sketch elements (blocks in SW, groups in every other graphics program) is useful for patterns and re-use... but like feature groups, often it is also a good way to reduce the complexity of the sketch, partitioning constraints so that they are easier to understand.

That's the stuff I'm waiting for before I invest in building any infrastructure around Onshape... Though if a minimally capable version of drawings + dxf export was available I probably would have tried to use Onshape in this last job.

I should point out that the list on my whiteboard used to be a bit longer and every new release I'm eagerly looking for what I can cross off.

I know a couple have used Onshape in "production". I'm curious what are the rest of you are waiting for? What are the roadblocks for you?

brucebartlett

Good List, I am also reluctant to launch into project's until things are rounded out. 

Traveler Hauptman said:

Though if a minimally capable version of drawings + dxf export was available I probably would have tried to use Onshape in this last job.

I am in the same boat.These 2 are a must have.

billy2

@Traveler Hauptman‌ 
Would a "group" be similar to an assembly? A way to gather a bunch of components that act as a whole?

I always wondered why we limit our data structures to parts, assemblies & drawings, a concept brought over from draft boards. 

Spaceclaim had an interesting grouping algorithm. Basically a saved selection list. Then when applying a function to that list, it would up-select/down-select and choose what made sense for that function. ie.. let's say I pick a vertex and add it a group (select set), then I pick a move body function, SC would up-select the vertex and grab the body and move it. This made selection sets (groups) more useful and powerful.

Maybe I should wait for your response as to why you want groups. I agree that assemblies aren't a good why to gather components.

billy2

Ok, I'm an outlier, and I'm posting this to help balance the scale dealing with drawings & dxf.

I don't want drawings and hate dxf.  

There are much better geometry standards out there other than dxf. Let's please extricate dxf from our planet.

traveler_hauptman

@bill In the context of what I wrote, an assembly is a group of parts (with the important aspect that an assembly can contain (sub)assemblies. I didn't have it in the list because onshape already has it.

Working with selection sets sounds interesting. I have not had the opportunity to use spaceclaim. I'm able to do most of what I need to manage a complex design via the hierarchical organization of assemblies but there are a couple areas where selection sets would would help me. The biggest one is for estimating mass parameters for an articulated assembly.

Regarding dxf, I don't have any love for it and I use other formats when I can. It's just the format that every program can deal with and so if Onshape guys on have the resources to do one format of 2D export ASAP... make it dxf. 

brucebartlett

@bill .dxf is the format accepted by all my profile cutter subcontractors (Laser, Plasma, Flame, Water-Jet, Turret punches, etc), I could be doing a hundred or more of these in a week, so I want them to come out in as few clicks as possible and no messed up scales. 

kevin_quigley

Dxf still lives in profile cutting and sheet metal. So does inadequate user training...you would not believe the number of times I have had to explain to suppliers that they need to check the size matches the reference dimensions. So much so we always add a 10x10 mm box in each file as a scale reference.

my favourite Rhino tool...Bounding Box combined with box edit. Scaling made easy.

billy2

The fact that it's unit-less is devastating. And, it has circles, poly-lines, lacks splines and lacks rotation matrices.

I always wonder why people fear splines and I attribute it to acad and it's lack of splines.

@bruce I'm way out there, I realize this more and more every time I post on this forum. 
But I've written IGES translators for various companies and g-code for others. I enjoy looking at the 
geometry companies are using for their CAD systems and pass judgement based on how well they define their
geometry.

Remember, acad had two siblings when it went to comdex in the beginning which were desktop publishing tools.
Acad was the version that allowed dimensions and people started using it as an engineering tool.
I think it's out of place. Now, you can tell me it generates $billion dollar in sales and I'll tell you
I just don't get it.

SW had really good geometry definitions. Cadkey used a 3x3 with an offset vector (crazy).

Geometry has been defined since 1960's. We should be using a better standard.

Start with iges, then step. We should be barking at the step committee to update the standard to include hole meta data so we know which holes get taped. Some want tolerances and GD&T in the standard. We still have a long ways to go.

kevin_quigley

re the AutoCAD thing, I'm totally with you. never understood how it managed to become so popular. Triumph of marketing over sense. Back in the late 80s/early 90s when it was all to play for there were some great drafting systems around at a fraction of the cost of AutoCAD...unfortunately most were Mac only, so as Windows and DOS marched on AutoCAD won. As a Mac user I was running Claris CAD, then Ashlar Vellum and MiniCAD (now VectorWorks). Oh what might have been!!

brucebartlett

@bill Our laser will accept .iges but we are now embedded with .dxf's, would need very good reason to change. Ideally the Laser would like the Trumpf Top's .geo format, this includes material type and thickness in the meta properties, not sure about about unit's. I see there is requests for this straight out of SW have to look into that. Hopefully in the future Onshape can be customised to pull this format out for specific customers dealing with Trumpf machines. Interested to hear your thoughts on the .geo format.

billy2

Trumpf uses spaceclaim for it's geometry engine. 

In fact, Trumpf made spaceclaim's sheetmetal really good. 



My waterjet guy has one of these:


They mostly do window glazing for the architectural market and they're stuck with all the acad architects. Most of the cuts they make are perpendicular to the glass and all this 5 axis stuff is for show. It will compensate for the waterjet stream expanding and creates the cleanest edges I've seen from a waterjet. So, maybe it's not a total waste.

Guess what this mach4 machine uses for a geometry engine, spaceclaim.

I've been pushing the operator to learn how to run the machine properly and how to use step. To get the machine to cut 5 axis, you define 2 curves in space. The machine then looks at these curves by percentages and defines a vector which it uses to align the head. 


Let's say I have 2 curves in space and I want to cut this shape:

Curves always have a start/end (closed splines have special start & stop conditions, but they do start/end). This highlighted vector represents the start on both curves and represents were the waterjet will cut 1st:

This vector represents 25% for both the top & bottom curves and will dictate were the cut head is positioned when 25% complete:


This vector represents 50% and you can to see how 5 axis movements are controlled:


What if the top/bottom curves didn't have a common start point and twisted? Spaceclaim has tools to fix this but it's way out the reach for the kid who runs this machine. 

When we waterjet, we have'm cut c'sinks. Why not? just not with DXF. 




I looked up .geo which seems to be a trumpf format and is probably gcode. Can you view it? Do you see a bunch of G1, G2 & G3 commands?

How hard would it be to write a .geo output for OS? Not hard, just need access to the API, access to an ordered selection list (possibly chain select), and a little code to remember where you are so you can make proper left/right arc decisions, then write to a file (oops) I'll just email it to you or create a tab with a text file. 

One thing I'd check is to send your laser a spline  and watch how badly it segments the curve. With iges you'll preserve the b-spline and trumpf may produce a G06. This would produce a nice curve. 

There's a lot of golf companies around here and many guys don't like linear interpolated splines. That's what DXF does when converting splines into polylines. I've been fighting for circle interpolation for a long time or some better way to get curves to a controller properly. I'd love to send a controller spline parameters and have the controller produce motor steps based on a spline definition. A lot of guys just grind the facets out after cutting. It's a lot of work.

California Precision has many trumpf machine that do run spaceclaim. I'm not sure what the advantage is other than trumpf now supports step. We just send them SW files and they handled the rest. 



Bruce, if all you're doing is lines & arcs. Then use dxf. 

brucebartlett

@bill getting out my league now, I just have to pump out flat profiles made from lines and arcs for 95 percent of work I do. I  know the machine struggles with detail when splines are converted back to steppie polylines, I will have to look into this for the more detailed spline based profiles, the later version .dxf seem to handle splines better, may have to experiment with iges next time I do a part with text or spline detail. Back in another late 90's I was working for toolmaker using Cadkey and machining strategist and we would run our programs back through post processor to convert the G01's to G02's and G03's where applicable, we could knock upto 90% of some programs and make the machines run a lot smoother.

billy2

I haven't keep up on DXF specs.

But I know units is still an issue and wonder why autodesk doesn't fix it.

If I send out a DXF I have to change SW units to inches, export dxf, then change SW units to mm. Doesn't sound bad but when you have 50 parts to export, you get tired of this routine. 

Do you design in inches? that would make exporting easier.

Most the g-code generators I've created were in Cadkey using CADL. Robert White was a good friend of mine who wrote Fastsurf and got me going with surfaces and curves. That was a long time ago and you don't look that old.

brucebartlett

Cadkey 97 was my first intro to 3d as toolmaking apprentice, have seen it for the last 10 or so year's but have some good memories of the direct modelling capabilities. 

billy2

Cadkey a direct modeller?, I guess it was. 

DeaconDave242

With the advent (although not really...it's been around for decades) of "light-weighting" in almost every manufacturing discipline, the thing that I find missing that would be MOST useful is the ability to assign materials and/or densities to the various parts of an assembly to determine mass properties. Every industry is keenly interested in savings due to lighter weight objects in the assemblies and units they offer for sale and/or for complying with customer specifications. I've been told that this functionality is being worked on, so I'm anxious to see just how this will be implemented in Onshape.

brucebartlett

@DeaconDave242 yes I'am waiting for this too. Functionality is there, if you click on a part in the Part Studio parts list you will see a set of scales come up down the bottom right. Looks good but I still don't think you can apply materials so you won't get a mass. There is a whole thread on this somewhere.

DeaconDave242

Thanks Bruce...I see what you mean...needs a lot more before its very useful tho...

brucebartlett

@DeaconDave242 see this thread https://forum.onshape.com/discussion/comment/519/#Comment_519