Yes Ideally this would be the case. I started to talk about this here:

http://openfarmtech.org/forum/discussion/33/cad-work-for-liberator-v2-need-criticism

At the very least, we need to adopt a standard output file, like PDF so people can get to the information easily. Honestly, there are bits and pieces everywhere. There's a Wiki, a Blog, this forum and a main website. There's even a couple "build" sites that have been used.

Organizing and sorting through all this is needed for sure. This is not my area of expertise, so I wait.....:)

The Dawg

I looked at BRL-CAD a while back and it is a really primitive tool, not really user friendly or up to current standards. This is a case where we should bite to bullet and try to acquire some floating licenses of a more modern software. The features available really do make a world of difference. My personal choice, motivated by budget limitations and a need for a serious tool compatible to the expensive software I use at work, is alibre design. A single seat of their basic software is 200 dollars, which is outrageously cheap, and their fully featured expert edition is only 2000 dollars and has tons of features.

My dealings with alibre so far have been good, and they seem very willing to negotiate on the price and set up installment plans (which is what I did). I cannot encourage this enough, and if there was enough interest I would even look into donating some personal edition seats to the OSE if it would help improve our CAD processes

heyo, i dont know much about 3d modeling, how do you guys view google sketchup? ive been using it because i know about it and its free. i know their file format isnt compatible with much of anything else. what other reasons should i seek out something else?

Are you implying that all of the tools MUST be open source? I'm fairly certain that most of the tools being used are commercial products, and while I completely support developing a free and open source CAD system, I don't believe we are close enough to an adequate system to intentionally cripple ourselves with inferior tools too early in the process.

The implication of such suggestions is that engineering so unimportant to the overall process that the quality of the tools used is irrelevant. Unfortunately this isn't a software project, its ultimately a production project and that means that the assumptions and constraints are very different. I am not simply suggesting that we use this particular software, I'm attempting to help with the costs through major donations. Now compare the features and realistically assess the benefits of having the additional modules available. If your only objection is that it isn't free and open source, I would say its a bit over-optimistic and impractical at this time to constrain ourselves in this way. That being said, I am eager to find an adequate free and open source alternative, but between sacrificing features or money I think the advanced features are more than worth it the money.

what does everyone think the "make or break" features should be?

im sorta leaning with andrew that it should be free. but i could see a world where the people that just want to try out modeling can use a free software and then the guys who do the final release drawings could use more advandced stuff. but it would be nice if we had a standard for both just so there is a consistent procedure for sharing and using the files.

what final outputs do we need? what could make or break that? a 3d model and a 2d diagram? what if it can do one but not the other.

should we use different programs for torch table/3d printing/manual welding diagrams? or should we expect one program to output for all the different modes?

I'm certainly not insisting that the _tools_ be open source but I think the format chosen for the various types of things should be something that there are open source programs to deal with, at least to some extent.

For example, if we want to distribute the CAD files with the build instructions, they are useless to the people this hardware is intended for if they cannot be viewed by some software we can distribute with our instructions, videos, ect, and that means there has to be some open source program capable of at least reading and displaying our preferred file format.

I don't know too much about the various CAD file formats but IGES does seem like it is a good one to standardize on. BRL-CAD can read and write it, and supposedly FreeCad can too. I am sure ProE and Solidworks would support it and I think I saw that alibre does as well. Assuming all these programs can actually read the format and that it stores the 3d structure well enough for us to do what we need to do (make 2d and 3d renderings, create CNC programs, create torch table routines, and create parts lists) then that should work. With this strategy you can work in whatever program you want and then just export your finished results to the standard format and upload it to the wiki. We need to make sure though that you can move amongst the various programs with no (or at least minimal) loss of information along the way.

With regard to using the same program for designing, testing, CNC programming, etc I don't think there are programs that do all this unless you spend many thousands of dollars per seat which is probably not what we want to do so we will likely have to find programs to fill each of these needs. BRL-CAD and FreeCad take care of a lot of the early prototyping work though, even if they are not the easiest to use.

Yeah, I am noticing that lossless is pretty much impossible, however if all the dimensions carry over which almost any format to any other format will preserve we we can pass around an "approximate design" which is close to the final design until we get it mostly worked out and then the final version can be stabilized into some long term storage format.

The reason you don't get lossless transfer is because not everyone stores materials the same way, some formats don't store it at all, some use BREP whereas some use CSG, etc. However, most of the common formats are convertible if all you want is the basic shape of the parts and the dimensions. And it is only the stabilized design that we are interested in producing CNC and torch table routines for so all that can be ignored until the design is finalized.

so what did we decide on?

software doesnt matter, file format doesnt matter. how do we dictate uniform drawings?

i downloaded brl-cad. i guess "archer" is the drawing program? i probly need to read up on the website instructions before i try to start drawing.

k ill try freecad too and see how i feel

I was talking about the layout of things like title blocks and so forth.

I also thing some training and discussion of GD&T, http://en.wikipedia.org/wiki/Geometric_dimensioning_and_tolerancing would be very helpful... you can never review GD&T too much, and its a somewhat difficult subject for those who have never really encountered it much before. My previous messages were all from my phone and for some reason Android (or at least the Dolphin browser android) doesn't really like the OSE wiki or this forum very much. It would be nice to have a standard file type (digital) but since it seems unlikely that efforts to implement that would be particularly fruitful I'm willing to settle on simply standardizing the drawings.

As for the other standards you mentioned, yes they are absolutely vital. In order to develop those we should develop a process similar to the one used to choose the 50 GVCS technologies, only specific to each of the particular domains (manufacturing techniques, materials, standard fasteners, etc.)

My vision here would be a system in place that would allow interested parties to contribute to a project but not necessarily be expected to output finished drawings or even standard solid models. As long as they can find a way to interact with their design team they're good to go.

A core group of CAD members would then be responsible for taking all the input from various members (programs) and converting or even totally remaking them in order to produce presentable finished drawings and instructions. This way, the people that already have access to the best software and training to do a particular job can worry about polishing everything up.

All I need as an experienced draftsman is a way to communicate with other team members effectively. Once I understand them, I can do the rest.

The Dawg

The requirements for the finished drawings that I see are
(1) capable of being read with free software on any operating system.
(2) capable of being printed out for use in the shop. A laptop is not terribly compatible with torches for cutting and welding, for example.
(3) separating text notes and instructions from the geometric layout. Since this is intended as a world project, that will make it easier to translate into various languages.

Solidworks is a nice program, and the program I use can read and write in the solidworks native filetype. I also agree with the assessment that we only really need a core group of dedicated cad people. At work I frequently get photographs, sketches, verbal descriptions, and occasionally physical prototypes. A design engineer will be given a cobbled together prototype and have to convert it from spot-welded and duct taped to a finished good that can be easily manufactured, eliminates the inefficiencies of the design prototype, and irons out all the wrinkles.

In this engineering model the other engineers mainly focus on the effect they are trying to produce, and the physics behind that effect. When I look at the efforts thus far, I see that the projects have an abundance of electrical or mechanical engineers (the scientific effect guys) but not a single design engineer. Now this process of innovation is only the beginning of a long chain of processes that improve the product and prepare it for release. Especially if these tools are intended to be built and used in all sorts of different environments. Our future users may not be literate and most likely won't have many of the tools we have access to. How do you teach them how to read drawings, follow maintenance schedules, adhere to safety procedures, and so on? The only reliable solution is to attempt to design the machine so it can only be put together correctly, is super robust and can survive infrequent maintenance, and has safety interlocks that mandated safe operation. To implement even those requirements may force a design engineer to radically rework the whole product. Wanting it to be built anywhere means we have to dramatically limit the fabrication methods we design the product to be built using. This means vast swaths of otherwise valid designs will have to be scrapped if they contain even one part dependent upon advanced manufacturing capacities. This is very much like an insanely complicated Sudoku puzzle, with hundreds of columns and rows.

I personally love doing design work of this type but for it to work the design engineer needs the support of the rest of the team to "meddle" with the other guys "perfect" design. Patience is also needed by everyone involved since design engineers go through hundreds of iterations in which they gradually "evolve" the design towards the specification. Most of these iterations are virtual, but there are still numerous physical tests that must be built and run to ensure particular solutions work as anticipated.

The process may seem difficult and long winded, but its what separates dabbling from engineering.

You can't really do a thorough analysis of failure points on a conceptual design, but on machines that are fairly mature like the CEB press we definately should do this. And we should look not only at catastrophic failure points, but also on points that will receive excessive wear and thus fail quietly later on and render the machine basically useless until they are repaired or replaced.

As much as I welcome the structure I think we need to also realize that open source is a process.  Even more than the destination, I think it's about the journey.  So once something functions well enough to get the job done and we've met a basic set of conditions we need to do a release and get it out there.  Then we let the process work it's magic.  We let first adopters give us feedback.  We let the design evolve over time.

What I'm trying to say is lets not let the structure take over either.  We are all volunteers, lets make the structure as easy to work with as possible.  As welcoming as possible.  Words like inclusive come to mind rather than exclusive.

The Dawg

Aneiren, what format are you planning to make the title block for the drawings in.  I plan on uploading a new 2d sketch of the rototiller drawings with the actual tiller teeth on it so the measurements of the existing machine can be added.  If I had your title block in SVG format I could use that for the page background in FreeCad when I do the export.  It would be nice to get that in use right away.  If you are doing it in something else I can convert it to SVG with inkscape easily enough but it would take 1/2 hour or so to do.  As a starting point, you may want to look at the one FreeCad ships with as it is already a nice starting point and it is already in SVG, which is the best format for something like that (at least a vector drawing anyway, if not SVG).  Have a look at the rototiller page on the wiki to see what it looks like.

Cool.  I am just about to upload another drawing for the rototiller now but I will do the next version with the new title block then.

EDIT:  Just in case you find this useful, here is a category with several versions of the OSE logo and also the CC license emblem, which is what I am assuming we would use for the 2D sketches that will have the title block you are designing.

http://openfarmtech.org/wiki/Category:OSE_Promotional_Images

It would be good to get a vector version, although the regular logo can be embedded for now as that is better than nothing.

I added several tables to the CAD team page on the wiki for standardaized building materials, fasterners, and the like.  There should probably be a few more tables, and the only items I filled in were just random picks from the first bolt sales website on google, but the tables are in place now so people can see how they are made and hopefully can fill them in.  If no one does it I will eventually fill them in but it would be good if someone with fabrication experience would pick out good choices for each of the various options.

We probably don't want to "hunt" for the best price available for each component, but rather try to find the typical price you would pay without too much searching.  Also, we want to try to get the best deal we can for the building materials if we have to buy them so if there are common sizes that are cheaper those are the ones we want to use.  Not only will startup villages have an easier time getting going, but as they acquire tools and begin to make some of these things themselves, being able to sell the extras they make is a key part of the overall strategy so we want them manufacturing the common sizes that are easy to sell.

Draftsight seems like a decent 2D tool, but for most of our applications I think we really do need to use 3D Parametric modeling. Of course, practice and experience with a 2D tool is still valuable and will help you throughout this process.

Well I haven't found any good 3D parametric CAD solutions that are complete. There is FreeCAD which shows progress but is still largely incomplete, and there is BRL-CAD which is a somewhat outdated system. If you are really good with computers and programming I would suggest you help out with FreeCAD. Otherwise there is BRL... but I am personally not a fan and I think it will teach you bad methods that won't transfer to a modern CAD system.

Will It Print?

One of the things that I find most wonderful about sketch-driven parametric CAD programs is their ability to automate the actual design process within the software as well.  Once the general layout has been done these programs allow the dimensions of the sketches to be driven by spreadsheets (such as Excel) and so you can make a design that can be modified and controlled without actually having to go in and individually manipulate the elements.

And to extend the thought... what do people do who can't get to a home depot or etc.?

I had hoped that this topic, as technical and practical concern would have garnered more feedback, but we're slowly making progress.

What would be helpful (to the whole domain of open source hardware) is an open source equivalent of the ISO or the ANSI organizations. Perhaps the OSSO (open source standards organization)?

I was trying to get more information to be able to populate the title block with meaningful info (like project codes or manager approvals). The title block isn't really much more than a container for that sort of data, and we don't know what that data will be.

I guess I can just bite the bullet and make it up, and if people don't like it then tough. Everything the OSE will have to be done the hard way because Generally Accepted Practices are dismissed by the leadership as too much work... just wing it.

Anyways, I'm done complaining for now.