3D Printing / CAD

Black Hole

May contain traces of nut
Can anyone suggest a 3D CAD package that doesn't cost too much money?

I have a project that involves dipping a toe into 3D lithography (3D printing); I have access to a printer but need to prepare the print file. The free options I have tried seem very difficult to work with (and I am proficient at 2D CAD - they just don't seem to work the same way with regard to dimensioning) or basically involve calling up a primitive solid and morphing it a bit (yes, you can get something out of the printer pretty easily that way, but not necessarily precisely what you need for an engineering requirement).

I have looked into Solidworks, which seems to be the gold standard, but that is bloody expensive with no home user options and education options very much time limited (and not cheap either).

Basically I expect to be able to draw something in 2D with precise dimensions, and then swivel it around to work in other planes. I have an old version of TurboCAD which claims to do this, but when I actually try the facilities are poor or non-existent (maybe locked for upgrade). I don't mind investing in an updated TurboCAD, but only if it is going to do what I need (it's not just the cost; my time learning to use it is also very valuable).
 
A bow at a venture - a friend mentioned he'd tried, and approved of, a free package SweetHome3D.
He was saying this to another friend at a pub, and I haven't myself tried it, so have no idea to the capabilities.
 
Thanks, but as far as I know that lets you design and furnish houses and then take 3D rendered walk-throughs of the result.
 
I think the problem is a conceptual one. I have had a glance at TinkerCAD, which is a free modeller that works on-line through a browser. The introductory video is quite instructive.

I now realise that if I draw using a traditional pencil-and-paper drafting simulation to produce a mechanical engineering drawing, and then just extent those techniques to the third dimension, all I end up with is a wire frame (one-dimensional lines in 3D space which outline the object but provide little information about its surfaces or volume). Those of us used to reading engineering drawings fill in the details by making assumptions, but as we know computers left to make their own assumptions often get them wrong, so there has to be a way to input the extra information.

I am sure it is possible to work this way (start with a wire frame and then define the surfaces and volumes afterwards), and the likes of Solidworks probably provide for that, but TinkerCAD and several other modellers I have looked at use a different starting point which may be more relevant.

TinkerCAD (et al) start with solids in the first place, so that you create the object as a merged combination of primitive solids. For example: a table could be made up from a flattened cube as the table top merged with four stretched cubes as the legs. Solids can also be set to subtract from rather than add to the object, so a hole can be drilled into the table top by subtracting a cylinder.

I was kind-of aware of this difference in concept, but had not come to the full realisation of it until now. The problem I had accepting the build-up-from-solids approach is that I had not found a tool where the user can easily define the precise dimensions of the solid primitives (or at least not at first glance), whereas using 2D drafting the dimensions are defined precisely when drawing the outlines. TinkerCAD seems to have this capability.

The other blockage is that I now have to adapt to thinking in terms of how the desired object can be defined as a merged series of additive and subtractive primitive volumes instead of outlines. This is rather like the difference between 2D CAD and 2D illustration (eg CorelDRAW!) where the illustrator typically creates graphics by merging 2D shapes rather than drawing an outline and converting it to a shape (although that is possible too).

The fact is that I use CorelDRAW! far more often than TurboCAD, even to produce the occasional engineering drawing, because it is a more versatile tool for creating graphics in general and I have become rusty with TurboCAD (which is a one-trick pony). I should therefore be able to adopt the TinkerCAD approach, by thinking of it as a 3D version of CorelDRAW! rather than as literally a 3D CAD package.
 
Well, I have had some success (although it was a struggle due to PC problems - see another topic). Having drawn a traditional mechanical drawing of exactly what I want (using CorelDRAW as it happens; 2D drawings in plan, elevation, etc), I installed FreeCAD and sat down for a day to figure out how to use it. I chose FreeCAD because it had the appearance of being able to design parts with shapes and dimensions as an engineer would, rather than free-form as an artist/sculptor would.

Even once I understood the workflow, it took a while to make the workflow happen because you can't do the next thing until the previous step is absolutely error-free. The most difficult bit was understanding the requirements for a "fully constrained sketch" - and I'm not sure I fully understand it now.

The idea is that you create a sketch, representing the intersection of the solid you wish to create with a convenient plane. The sketch is then converted into a solid by extruding it along the perpendicular (creating a prism) or by creating a solid of revolution by rotating it around one of the major axes in the plane. Having created this (relatively simple) object, one then creates further objects and uses them to combine with the existing object(s) in various ways - eg merge, or create a hole by subtraction. Therefore the designer has to be able to envisage a manufacturing process (in the memory of the computer) that builds the target object as a process of adding and removing simpler volumes.

The problem of constraints is that every point in the 2D sketch has to be precisely specified on the plane, either explicitly or as a consequence of other constraints. A single point has two degrees of freedom: where it is on the plane. This can be constrained by specifying the distance from the origin in x and y. A straight line may then start at that point, and has two further degrees of freedom: how long it is and what angle it is at. And so on and so on. What you do is sketch it all roughly, then go around placing all the features. For example: marking the end of a line as coincident with another point which is already constrained, constrains the end of the line. If the other end of the line is already coincident with a constrained point, then the whole line is constrained (or it requires two more specifications to constrain it). With a complex sketch, trying to find everything that needs constraining is quite difficult, and several times I ended up deleting all the constraints and starting again. You can't have redundant constraints either!

This all sounds unnecessarily complicated, but it is also useful. Part of my design required several identical sketches but in different places. All I had to do was duplicate the sketch and then manually adjust the constraint parameters for its location on the plane. Every property of the design is individually editable after the event, so the radius of a circular element (for example) can be tweaked and the effect will propagate throughout the design, without having to rip it all up and start again.

Here's what I have ended up with (after about 3 days head scratching and PC fighting):

Image.jpg

Now my problem is to export the data in a suitable format, and find somewhere to make it - my original contact for 3D printing is no longer there!

Incidentally, as a hobby creation, FreeCAD is amazing! They say anybody can have a crack at contributing to the code, but even if I was a competent programmer, the concept of how you go about something like this is mind-boggling.
 
Somebody tell me the answer to this one (if there is an answer): I installed FreeCAD by running the installer exe as admin. Now if I log in as admin, I can go into the Default Programs tool and assign the .fcstd file type to open in FreeCAD. But I can't do that in a user account, no matter what I try - even running the Default Programs tool as admin, or right-click on the file and try to change program - it just won't stick (and the file has become attached to Adobe Acrobat Reader as a result).
 
I have discovered that Cardiff Metropolitan University have a public Fab Lab, their web site says £8 or £10 for a four hour session (£2 per hour thereafter). Sounds good to me, especially if there is help on hand. Not limited to printers either, it looks like they have CNC, vinyl cutter, etc.

I realised I had a problem: no certainty that the FreeCAD output is a valid file. A web search indicated it needs a plugin to produce printer output, although the option appeared to be present in the export dialogue. So I installed the free STL viewer from 3d-tool.com, and it gave me this:

Image2.jpg

...so at least I now know that something else is able to take the FreeCAD output and interpret it as the object I intended. I guess the plugin requirement may have been for an older version of FreeCAD - I had to update my installation when I found no facility to create an ellipse, but that it is included in the version released in April (hopefully a future update will include support for b-splines - it was time-consuming creating the central hole with a constant wall thickness to the external ellipse profile).

Another thing gleaned from the web was that FreeCAD didn't handle the scaling very well, but I had not taken any particular precautions. Again, that may have been fixed - the 3D-Tool viewer lets you measure between selectable nodes, and the scale is perfect.

I can't imagine it will need a 4-hour session, but within that time I should be able to knock up a first-off, test it, do any mods to the design, and make a few more. Courtesy of my PC still being portable!

Update: see photos of the facilities here: http://www.fablabcardiff.com/
 
Last edited:
...alternatively I can just arrange 5 into the printable area and send the file to a print service, pick them up this afternoon for £16.47 (and hope they are right). Just one will cost about £8.

It's a dilemma. If I knew the design is right and the material/construction will not cause a problem (eg too brittle), I would just print the 5-off and be done with it. If the design needs tweaking, the FabLab is the way to go. Bugger.

Update: dammit, time is precious. I've sent the 5-off print file to a fabricator on www.3dhubs.com and I should be able to collect the result next time I am in the area (maybe even this afternoon) - well before I would be able to schedule a session at Cardiff Met (however interesting that might be). If the first-off doesn't work out, I'll have to think again.
 
Last edited:
Harrumph!

The bloke I chose to place the order with accepted it, and the 3DHubs system took payment and marked it as being printed, so I thought "great, I can pick up while I'm that way this afternoon". No such luck. I got a message "I won't be able to run this until tomorrow night", and I won't then be free to pick up until Friday. Should have gone to Specsavers FabLab!
 
I thought about cancelling and spending Friday afternoon at FabLab, but it would be inconvenient. Instead I have revised the order - refined the design slightly, and added a separate simple cup shape as a lens cap for the SJCAM. This is still a better option (using somebody else's time rather than my own) unless I need that print-test-revise-print loop.

The selected printer has a print volume of 200mm cubed; as long as everything fits within that (and are all standing on the base, not floating in free air!), there is only one set-up charge and the rest if the cost is driven by total volume, ie the amount of plastic used. Not that the set-up charge is huge (£4+VAT), but by adding a small part such as the lens cap to an existing print run, the incremental cost is 64p!

Incidentally, the printer didn't want multiple objects in the STL file. He just wanted the files for the single unique objects and a repeat count (quantity is a parameter in the order system), and will arrange the print run himself.
 
There are problems. The plastic shrinks as it cools resulted in some warping (particularly at the ends of the arms) and dimensional errors. However, there is less shrinkage in some places than others, maybe where the dimension was constrained by being in contact with the build platform, or maybe where there is more structure to resist shrinkage, so it wouldn't be a simple case of applying an initial scaling factor to compensate.

For a critical part it would be as well to expect some mechanical finishing, for example print a hole slightly undersized and bore it out afterwards, or print slightly larger and file down to the required dimension. The lens cap internal diameter has gone down from the design 20.75mm to 20.1±0.1, so it won't fit without reaming it out.

However, that said, I can't remember exactly when I started work on it (the file says created last Wednesday, but I don't think I got properly stuck in until Saturday), but I reckon that was 6 days from being a 3D CAD novice to a finished product (I had a message that the print was finished about 10 last night, so I asked if I could collect straight away - I didn't realise I was dealing with a hobbyist running his home-brew printing outfit!).

I have also been tossing up whether to buy into this myself. Prices start at around £500. On the other hand, £150 buys a year's membership at Cardiff Met's FabLab, with access to several printers, CNC, 40W & 60W laser cutters... (although members still have to pay for time on the machines, membership gets you access at extended hours). I could just pay for the induction training and then pay-as-I-go after that.
 
Not too bad actually. I went for ABS (the other commonly-printed material is PLA, which is sugar based and doesn't like getting warm), and I have cracked the ellipse on one of them (it is intended as a deformable structure) through excessive handling, but it superglued back together fine and I have had no other accidents putting them into service.

One of my projects will be to run off some test structures to measure the standard mechanical properties, but I find it hard to believe nobody has already done that and the results must be on the web somewhere if you know where to look.

How about this for an idea: replacement cooling fan blades with feathered edges to reduce fan noise.
 
Interesting stuff :thumbsup:

There are problems. The plastic shrinks as it cools resulted in some warping (particularly at the ends of the arms) and dimensional errors. However, there is less shrinkage in some places than others, maybe where the dimension was constrained by being in contact with the build platform, or maybe where there is more structure to resist shrinkage, so it wouldn't be a simple case of applying an initial scaling factor to compensate.

This is where a professional designer will earn his keep. He(she)'ll know how and where to tweak individual dimensions.

As to buying in, it's still a rapidly developing market. I'd think that year's sub would be the best way to go - that way you can get experience of the machines and materials and in a year maybe decide on a system to suit you.
Or decide you've had enough and go play with something new :)
 
Back
Top