How Much Convenience Would You Give Up to Get Total Security?

A few days ago there was a discussion on the SolidWorks forum about some very old videos showing some simple functionality of Synchronous Technology in Solid Edge. It looked like some people were starting to get it about how the editing is easier, the broken relationships are non-existent, and everything is a lot more flexible – for prismatic design.

And then in my way of thinking, I kept thinking about how Onshape was innovative about several things like especially file management, but really missed the boat on modeling. They copied SW’s Parasolid roots very closely. I had been hoping they would transcend history-based modeling methods. I’ve written a lot about history-based alternatives over the course of the last 8 years or so.

And then an idea popped into my head. There have been a lot of “systems” people have invented to help you avoid the tangle of references and especially broken references when modeling assemblies and individual complex parts. Resilient modeling is one I’ve spent a fair bit of time on. SSP is one I think means Skeleton Sketch Part. There’s Horizontal Modeling, Wide Tree modeling, and more I’m sure.

Anyway, most of these systems involve some sort of top-of-the-tree reference geometry from which you then build all of your references. To take this a step further, if all of the references are in a single 3D sketch, then there are no history-based entanglements even at the references level. People are bad at planning, and these systems require a lot of planning. SolidWorks was so successful in part because it allowed people to get away with modeling by the seat of their pants, without any planning. Yes, it’s a very bad idea, but people gobble it up like twinkies because it’s easy.

My idea consists of this: every time you make a reference to something like an edge or a face or a vertex, the software copies that reference and moves it to the top of the tree in a special area for references, like a 3D sketch. In essence, you’re driving everything from 2D views, like drawing views. Of course this doesn’t work so much for complex shapes, but for prismatic parts, it’s brilliant.

For example, the part below is all driven from a single 3D sketch and the reference geometry in the part. I used the Flat Tree display to show everything strictly in the order in which it was created. You can see the only sketch is a 3D sketch, and the only references are the standard planes.

This is obviously a very simple example, but the advantage of a part like this is that it will NEVER fail due to references going bad. You can download it from this link and check it out yourself. (SW2020).

There are no references to part geometry. I made this by making a 3D sketch at the bottom of the part, capturing all the information I needed, then deleting the references, putting it back at the top of the tree, and recreating the features. If the software were built to do this automatically, you wouldn’t notice much difference.

The problem is that we have been taught since we signed the PO to make references to model faces, edges, and vertices. We don’t know when to stop doing that. This example goes maybe too far the other way, but it illustrates how safe a model can be. I’m not saying features will never fail, I’m saying they will never lose their references. For those of you who make parts that aren’t simple, and assemblies, you know what a big deal it is to lose references.

And yes, you can do this in assemblies as well.

But no, you really can’t do this with surfacing. Or you could for certain types of surfacing, but not all of it. It would be very cumbersome for most of it. But you could use the concepts to help improve your surface modeling practice. You would have to give up some conveniences, however.

7 Replies to “How Much Convenience Would You Give Up to Get Total Security?”

  1. I remember trying to change people from Autocad and Pro/E in the late 90s. It was tough, but people found reasons to do it, and we sold it hard, to be honest. It eventually happened.

    SE has a different situation. Everybody hates what they have to put up with around history-based software, but few people understand that’s what they hate, or have any idea what to do about it.

    If you’re making complex parts, you understand the losing references problem. If you’re making simple parts, you’re probably making more complex assemblies, so you would also understand the losing references problem.

    SE has another problem that SW didn’t have before DS created it – the big brother syndrome. I think Siemens sends a lot of customers to SW just because they don’t buy NX. There’s a strange internal conflict that I didn’t really understand. It’s kind of unfortunate. SE is a great product that gets overlooked a lot for various reasons.

    You can follow the evolution of my thoughts about synchronous from sarcastic skeptic to true believer from my posts here on this blog over the course of the last 12 years or so. Synchronous has been communicated badly to potential customers. There have been a lot of mixed messages, and a lot of non-existent messages. I had to go visit hq to get a real idea of what it was. And even when you do communicate it properly, history-based users just cling to their sketches and their daisy-chained relations so irrationally.

    Technically, I really belong on the NX side of the fence in this debate, but I just haven’t done anything about it. NX almost requires a big corporate presence behind you to justify it, and I really don’t want to go there.

    How long is it going to take to get people to give up their history-based view of CAD? it may just happen slowly by products offering alternatives like ST, and some of the other products that use direct edit type ideas. Of course a lot of the direct edit products didn’t do that well back in the late 90s, so there is probably a lot of prejudice due to that.

    1. I’m not as talented as you, but I use SOLIDWORKS Direct Editing every day. It really saves time. I teach this method to my students at Palomar College too.

  2. Some people just “rave” about SolidEdge, and I understand why. However, in 22 years, to 45 companies, as a contract Designer, I’ve never heard 1 person mention SolidEdge. Just sayin’.

    And many years ago, around 2004(?), SolidEdge opened a huge reseller office here in my home town Carlsabd, CA. They closed up less than 2 years later. A friend of mine was a salesman there, told me “they couldn’t give it away”.

    Cheers,
    Devon Sowell

  3. Matt-
    I’m glad this is being explored and shown to the world of linear history-based modelers.The Auto industry has been using this type of design for a decade plus now.
    For Solid Edge users people have to understand that it is not an either or with synchronous technology and history-based modeling. The blessing of SE is that you can work in both. Actually, with the ability of convergent modeling you can now work with 3 tools! The rub is knowing when to use which technology and to apply it correctly.

    Ryan

  4. Matt,

    This idea is very similar to what we demonstrated in a prototype called To3DNow yrs ago. This worked for parts and assemblies. There were no sketched references that relied on pre-existing geometry, however sketches did not become unconstrained or hanging out in space when things were deleted. Neither models or assemblies could fail in the spectacular ways that happen with current modelers (usually at the worst possible time) when a feature is deleted or changed. To3DNow was also a feature-based, parametric and associative modeler.

    We were not successful in garnering funding, but we did demonstrate an evolutionary methodology that would allow people to design and change designs faster and ideally, advance the current thirty-something year old paradigm.

    Good ideas like this do not exist in a vacuum and a lot of designers and engineers focused on organizing designs in SOLIDWORKS, Creo, etc… have found methodologies using skeleton parts and master model techniques to facilitate the notion of deign vs. modeling and also make the risk of changing models late in the design cycle much more understood. These techniques also facilitate team involvement in projects because the skeleton or master model pre-determines how parts and sub-assemblies work together.

    Skeleton parts in feature-based modeling go back to at least the 1990 time frame at PTC when this idea was used by application engineers in the creation of a new demo and the consulting group in trying to provide robust design methods while working with customers on complex projects.

  5. hello,
    I’m french (sorry for my poor english language) and I use SE for more than 10 years. Due we mainly work in my society on complex parts (around 80% of them are linked to complex design surface), we use history-based construction and nobody use synchrone technology.
    I know RMS method and I try as far as possible to use it or to keep it in mind.
    Similar to your description, I try to link my sketchs related to plan reference or to another sketchs, never (as far as possible) to edge or vertices of faces. I feel this construction more robust and I added buttons on SE “quick launch bar” to easily and fastly show or mask solid body. For example, if I have a extrusion define with a separated sketch and I have to create another sketch, I mask the solid body and I link my new sketch on element of first sketch and not on edge or vertices of the extrusion.
    I also got into the habit of taking my sketches and reference plans to the top of the tree (I also rename them).
    In this way, I find the CAD construction more robust and above all easier to repair after heavy modifications because the skeleton created by the sketches and reference drawings is stronger.
    Have a nice day!

  6. You can get burned by synchronous too. Especially by not paying attention to design intent settings. Especially, I’ve personally found, with parts with long thin “walls” where you find way down the line, usually in a draft, that some line has the jaggies when it definitely shouldn’t. By that point it’s often to late to correct. It’s due to the fact that visual feedback in the model didn’t come through because the angle was minute. Synchronous depends heavily on visual feedback. That’s why, in my opinion, it is best for prismatic parts. That’s probably why all the marketing gimmickry uses prismatic parts almost exclusively. Of course, I wouldn’t use anything but sync for imported models. There are no dumb solids with sync.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.