Tilt housing is printing with holes

[maxB]

Can I ask what kind of modelling technique did you apply with your last try?

Did you use Boolean operations? Lofting?

Edit: If you hide your front face you see some internal problems. Maybe from accidentally copy-scaling or using JointPushPull at some point?
Edit2: Personally, I would try to make the model using a lofting technique (see pic2).

• you could manually control the amount of edges in each curve
• turn on the endpoints so you can see where you add vertices/details
• model half if its 100% symmetrical. Mirror when all is done. More easy to handle the model, more easy for SketchUp.
• model each surface as a separate part. make sure you re-use the same contours for surfaces that are connected.

[Zxen]

@maxB

Boolean operations? Lofting?

I don’t think I used these. I did what Cotty showed me above to create a perfect dome, then I sliced into the two sides using four rectangles on the grouped dome. Then I ungrouped the dome, intersected selected faces, then deleted the cuts. Then I simply pushed rectangles through things, changing my mind as my design decisions changed. Each time I pushed rectangles through the cylinder, I neatened things up each time by using curvizard weld and deleting surface lines, as well as using alt+eraser tool to remove external lines. Sometimes I also selected all and then clicked AMS Soften Edges.

I didn’t use jointpushpull this time.

I don’t understand how I would draw half the model at a time. It would be too hard to imagine the parts as they work together. Remember, there are many other parts to the overall robot. I cannot attach a lot of parts for context because the upload limit is 3Gb here on this forum, and most skp files are at least 7Gb.

[maxB]

There are many ways to make this thing. It’s a nice challenge so I tried making it using a loft method. Still work in progress … will post the skp when its done in an hour or so.

[maxB]

This should be it. Use at your own risk.

I always try simplify a complex object into smaller more easy to make parts. I named and color coded the individual groups so you can get an idea of the workflow (start is 0 - 14 = done).

By using Curviloft you can control the smoothness (amount of segments) of the curves. Also, the shape tends to be very clean because you control the contours and vertices yourself. For 3dprinting in my experience this workflow is more fast than intersecting-cleaning etc etc.

broken tilt segment.skp (2.5 MB)

[Zxen]

Thats extremely kind of you, maxB. You have gone above and beyond. As you can see, this is one of my prototype designs as I figure out how to build beautiful elbows with internal wiring cavities for the motors I’ll be using. At the moment, I cannot open your sketchup file because I only have Sketchup 2015:

But I have continued to work using a clean method which allows me to make constant changes without breaking anything - to group every part of a shape and not combine them at all until the end.

My elbow consists of two motors - tilt and pan.
Now the shaft of each motor is only 5mm wide, so it is flimsy and weak in the extreme without a support lip/rail.
The design you made without errors is part of an idea I had that I may still use, to screw lips around edges to lock the tilt segment to the pan segment. But since that idea, I have been redrawing a cleaner design. I have a lot of prototype prints in front of me with the motors and electric ruler in front of me when I’m drawing.

At the moment I am trying to create a tilt/pan combo dome cylinder without any visible seam at all - the rotation will be concealed underneath at the base of it all. This will lessen the lines on the robot arm, so I am refining the design to make all connections including the lips internal.

I need a minimum thickness of 3mm of PLA or ABS plastic for strength for all parts of the structure. Unfortunately, due to the fact that artificial shafting by way of using cogs required more strength than anything I could print could bear, I had to consider not using cogs for prototyping cheaply with my printer. Cogs need to be made out of metal to tolerate the torque, otherwise the teeth snap off. This design in this thread eliminates the need for any artificial shafts, because I have pushed the servos all the way to the edges of cylinders so the shafts are now centred. However, this is supposed to be a woman’s arm with an elbow diameter of 62mm. This is too narrow for the required motors.

As it stands, the cylinder is already 70mm diameter. Nevertheless, I need the cylinder to be wider so that it can encompass the pan segment inside it. While remaining uniformly 3mm thick, I want to resize the cylinder part of this drawing to be 76mm diameter, and more importantly, I need to know how to do it. I can see there’s a scaling tool, but I’m not sure how to get the result I want.

Please remember that I have been working on this project for over a year (mostly braking systems and electical components) so I already have around 100 elbow designs. That equals around 10,000 if you count the ones I scribbled out along the way. So when I ask for 76mm, I might mean 72mm later, or even discard the idea of 3D printing anything at all because in the end this will be made of much thinner aluminium. My priority at the moment is to make an elbow from the printer that is as streamlined, sturdy and user friendly as possible. My own elbow is 82mm wide, so a 76mm wide elbow is not unusable. It seems a woman’s arm demands that I return to the far narrower artificial shaft designs.

But for now, can someone please tell me how to resize a shape based on diametre in millimetres?

I am attaching the work done so far in my redesign in case anyone has any ideas that could streamline this prototype better using 100% 3D printable components. If you have any ideas, please ensure you leave hollows for the wiring. The motors are 38x19.5x40.5mm.

70mm.skp (1.3 MB)

[maxB]

@Zxen

But for now, can someone please tell me how to resize a shape based on diametre in millimetres?

Hi Zxen,

Scaling the object would also scale the thickness. There’s no easy way around it.

I think a loft based workflow might be what gives you the flexibility you need.
If you save the individual base curves, you can always just change the diameter of any curve, even change the segment count, and rebuild the object just in a few minutes.

I attached the file as V15 format. Have a look at it. If you have any questions, feel free to ask.

Max

broken tilt segment.skp (2.5 MB)

[Zxen]

Thanks for the 2015 file. Flawless. Weirdly, your file has changed sketchup behaviour - The spacebar toggles between the push tool and the eraser tool. This behaviour changes in my other files back to normal.

Do you think ‘lofting’ is easier than any other method overall? Curviloft has been very troublesome in the past…

[maxB]

@Zxen

Do you think ‘lofting’ is easier than any other method overall?

For making ‘blobby’ models for 3d printing I would definitely use the loft method:

• for 3d printing you need clean shapes. Modelling a 3d solid object, intersecting that shape with other shapes etc etc can result in a messy end result because the software algorithms are really shaping your model. By using lofts your contours are defining the object.
• by using a loft method and saving the original contour files you can always, quite easily, slightly alter the model.

Curviloft can be troublesome if:

• you try to make a complex shape in 1 go using all contour lines at the same time → that’s the reason I split your shape up in separate elements
• if your contours don’t line up nicely. You need equal amount of segments in the individual curves.

Edit: That spacebar toggle seems like a bug to me. I don’t see how a file could change your keyboard shortcuts. Maybe make a new topic reporting this bug?

[Zxen]

broken tilt segment(2).skp (2.5 MB)

I have gone back to working on the model you sent me but all dimensions are warped and do not fit with any of the required specs. As you can see, I have drawn a 70mm circle under the dome and it doesn’t fit.

[maxB]

It’s no surprise they don’t match up 100%. SketchUp’s circles aren’t real circles but ngons. The higher the segment count the better they look but still, they are not perfect circles.
Also there’s a difference in the segment count you use for the 70mm circle and the one I used for the loft (see pic). If there should be a 100% match you should look at the segment count for both parts and line them up perfectly.

Finally, the loft I made was only using the side-contours. You could also pick the bottom contour and feed that to curviloft at the same time so it will use that contour as well.

If you have the time to learn and analyze the way curviloft works this should work.

.

[Zxen]

I can see my circle has 96 sides. How do I detect how many sides the base of the dome WOULD have were it a full circle so they match?

Yes I will look at curviloft more closely after New Year

As you can imagine, many people I speak to about sketchup blame me for my impatience when it comes to learning each plugin that counteracts the polygon limitations of this program. But there are other programs such as Rhino that do not ever have such limitations, so I would not need to work around problems constantly. If I could start again, I would learn Rhino first, because it looks better suited to drawing. So if this trend continues, that I need to spend more time learning how to get around curve issues on Sketchup using plugins than it would take me to learn Rhino from scratch, I would chose the latter. Sketchup has allowed me to draw in 3d for the first time in my life, so I am grateful for it, but it seems to like straight lines too much. Curves constitute 99% of the world around us, so creating a drawing program that cannot draw curves properly seems extremely flawed. I hope nobody here is patriotic to sketchup in any way, and everyone understands what I’m saying. Drawing a circle should not be this hard.

[maxB]

Maybe look at it in another way: look at you biggest part (70mm?) and decide what segment count you want / need to use for that one. Your 3d printer will have some limitation in accuracy anyway.

Then start drawing those parts using that count OR for the base of the dome you just create a ‘circle’ using that segment count and break the ‘circle’ at the endpoints of the base. At that moment all the points of the base will match up to the other part. You get my drift?

Beware: because its not a real perfect circle, once you rotate the 3d printed part slightly it will not fit perfectly but instead you will see tiny gaps again.

If having real circles is absolutely necessary for your design, maybe you could have a look at Rhino (or talk to people who use Rhino). I have no experience using it so don’t know if it uses real circles or ngons. The ‘I need real circles question’ has been asked here before and until now there’s no sign they will ever be/can be implemented in SketchUp.

[Zxen]

Basically you’re implying that I need to restart the entire robot from scratch each time I change my mind about the diameter of a curve in a single component. This elbow needs to flawlessly connect to all other parts, but if I decide something needs to be about 0.5 mm thicker or wider or steeper, it will throw out all other sizes in the entire body of the robot. Therefore I’d better not ever want to change my mind about something after I have begun. And rotating parts around will nearly always break things, cause curviloft skins to twist and bulge erratically, etc. I don’t know how you guys put up with all of this. Rhino is free for 90 days, and I just drew a perfect circle without learning anything at all. I zoomed into the circle, and even at maximum zoom, the circle consisted of one pixel per line. What projects are you using sketchup for? As far as I can see, it has not kept up with it’s competitors at all. Without plugins created by people who do not even work for or get paid by Trimble, Sketchup would not have any users, I’d wager. The community is probably the only thing keeping this software afloat.

[mac7595]

Zxen: On you assy. drawing and in the edit mode, select the face of what I assume is some kink of support and then select align view. That gives you a normal view of that face; on the very bottom do a right to left select : move / copy down. I thought that would give me info to set circle center and help but, first go seems like that may not be real circle and leads me to ask how are you doing the scaling.
Have breakfast appt. with daughter so cannot check my work so take with grain of slat.
Back later:smiling_imp:

[john_drivenupthewall]

@Zxen

What projects are you using sketchup for?

here’s an old one of mine, all made with standard tools…
the masts are 10mm carbon fibre tubes glued into 3d printed housings…
it uses SMA wire driven by capacitors housed in a separate base unit…

I have also automated articulated mannequins from film stunt work in the past, but I tend to use off the shelf car ‘universal joints’ and heavy duty cables for that type of work…

john

[maxB]

@Zxen - SketchUp is a unique kind of modelling software. Its has its pros and cons. A big con for your modelling (while still refining ideas) for 3d printing: there is no modifier stack in SketchUp (yet). Whatever you do is applied the object and you can’t change the original parameters while keeping all your modifications (which can be done in 3dStudio, don’t know about Rhino).

Maybe someday someone / some company will make a plugin that gives us some sort of modifier stack in SketchUp… you can only hope…

[mac7595]

MAXB;
Confirmed you contention and in fact the wall of the unit is not circular. Even if one uses different # of vertices the measured radius should still be the same through out. Just using the perpendicular bisector of the two adjacent chords give diff center point ref then your test ref. and two sides give diff CPs also
REF your stack issue is because you do not how to use Su yet, you have a number of undo steps, you can make the one under modification unique , you can replace the one you modified by one in the component browser; you can use proxies; you should assign the items you have to layers to control visibility; the guide lines you used to find center point seem a " little" off; use wire frame rendering mode helps some time to get better inference engine ref, TURN OFF LENGTH SNAPPING.
My experience all software has its own set of problems , one needs to learn to work around them
If there are design constraints because of housing just lock it so you can not accidentally change it

[mac7595]

I even worked on one with el over az. & we had to run wind tunnel tests on it because one design spec was 70 M/sec ( 156 mph) wind with blowing sand and it supported a top having ballistic coef spec also. It eventually failed because the labyrinth seal allowed gears to jam with sand.

[Shep]

Hey Everyone,
I was thinking about this model today while getting the car serviced and thought I might show one of what I’m sure are many ways to approach this item. It occurred to me that it may be possible to use considerably less edges on the inner shell, making intersection simpler and the model a bit neater.
The files below are the progression (multiple copies) and the end component.

Shep

Tilt1.skp (2.3 MB)

Tilt comp.skp (398.7 KB)

[bill_ruggirello]

there is a lot of good info here but I haven’t read it all.
However, I tried to print a part with 1mm wall thickness and my Slic3r put holes in the wall.
I had to increase to 1.5mm to stop it.

just a thought