Dyne's Treadwell

[Dyne]

I've had 0 trouble with scale from sketchup, it exports STLs in mm and that usually works fine.

Hm, could be some wonkiness with Fusion 360 import, though you'd think if that were going wrong, it'd at least be going wrong in a consistent way rather than two different scales.

There weren't any import option changes on my part between the two models. I don't recall ever seeing any to change aside from destination folder; the process is literally drag and drop onto the upload dialogue.

[Dyne]

As I mentioned in the edit to my next-to-last post, I was tinkering with an eye mechanism prototype while Fusion 360 was down, mostly to sort out what I need to know to design the real thing.  There are plenty of eye mechanism videos on Youtube, including a few free videos from the Stan Winston school, so if anything that follows is confusing, those may help.  

I know that Treadwell doesn't really have internal eye mechanisms, but I think it'll be fun to add one.  The entire setup will be hidden behind/distorted by the normal Paterson style lens.  It's mostly just to give an impression that something is going on in there.  Probably mostly painted black, with a few dim red LEDs somewhere in the murky depths.  It shouldn't be obvious unless you are specifically looking into his eyes.

Treadwell's head design complicates these mechanisms a little (since each eye is in a different box) but luckily, fitting servos into this space isn't a huge issue.






Here is the assembly for my initial mockup ...


Note that each eye consists of a double gimbal setup: the light blue outer ring pivots on a horizontal axis (you can glimpse part of it where the leftmost ring touches the leftmost arm of the yellow frame).  The grey inner disc is attached to that ring with a vertical pivot.  The outer ring can therefore tilt up or down on its pivot, tilting the inner disc in the process, while the inner disc can independently tilt left or right on its own pivot.

Now, imagine that one of the taller dark blue spindles is a servo, and one of the shorter spindles is another servo.  These turn their respective discs, or servo horns.  This leaves the other disc/horn at each height unpowered and spinning freely.

Now connect the horns at each height with a tie rod, like so:


These ensure that the unpowered horn at each height moves in sync with its powered sibling.  As you can see from the pair in the foreground, when either low horn rotates clockwise, the tie rod between them goes right, forcing the other horn clockwise as well.  (The angle of those two discs would be identical if the length of the tie rod was exactly the distance between each spindle, but my little wire mockups are only rough approximations.)

Now consider either of the upper horns.  If you use a similar tie rod to connect it to the the outer ring of the eye that's in front of it (in CAD, not the photo), then the upper horn can tilt that eye up (by pulling) or down (by pushing).  Now do the other upper horn the same way to the other outer ring.  Since both upper horns are moving together, as the upper servo moves, that means the upper servo will make both eyes tilt up or down in sync.

If you then do a similar arrangement with the lower horns (tying them to the attachment point on the right side of the inner disc of each eye) then the lower servo makes both eyes look left (by pushing) or right (by pulling) in sync.

I did print the eye gimbals and their frame, but these parts were so small that my print-in-place hinges weren't very effective and the frame is too flexible, so they pop loose.  It doesn't matter much, because the point ohere is just to explain how the servos drive each axis and keep the two eyes in sync.  

I am considering another method of articulating the eyes, anyway.  If you take away the gimbal setup and put a ball joint (or u-joint) at each eye's center of rotation instead, the eye piece can be a single part and -- in spite of having only one point mounting it to the frame -- it can freely tilt both up-down and left-right.  The mechanism in this video has extra servos in back for eyelids, but it demonstrates the idea with u-joints starting at about 35 seconds:





The drawback to this variant is that the prop eye must surround the pivot.  That's fine for a human-style eyeball, but Treadwell's eyes will be more like mechanical camera parts.  Having a ball joint in the center limits the design a bit.  With the gimbal I originally described, the pivots are on the outside of the eye, leaving the center free.

So that's what I've been up to on Treadwell recently.


Non-Treadwell tangent time.  (I did mention yesterday in another thread that I have the attention span of a goldfish...)

Whenever I do something like these droids, my urge to put my own stamp on them always starts getting twitchy.  That's one reason why I'm doing a Treadwell AND an R2 unit ...  I want R2 to be as close to screen accurate as I can manage, but I'm always fighting the urge to incorporate my own ideas, so I'm not trying for the same accuracy with Treadwell.  (The other reason for doing both droids is that R2 will take much longer.)

Another manifestation of that urge is that I'm getting tempted to eventually do a non-canon BB-8 style droid at some point.  I came across David Scott's BaBy-8 build over on the BB-8 builder's club and started looking into the concept art it was based on.  That's when I spotted this other piece of concept art (apparently) by ILM's Christian Alzmann:



If I were to build a roller style droid, I'd be tempted to base mine on that.  It's similar to BaBy-8 (and BB-8), but clearly represents a distinct design.

The one thing I dislike about BB-8 is how that type of locomotion just seemingly sprang from nowhere (in-setting), at least insofar as Astromechs are concerned.  There are other spherical droids (Droideka, Buzz droids, etc.), but unlike BB-8, those droids aren't exactly similar in purpose or style to Industrial Automation's products.  I realize that decades have passed, but Star Wars technology has never really seemed to change all that quickly.  Unless it is WMD technology created by Sith fanboys, I suppose.

So one reason I'm tempted to do that droid is to imply that it's an earlier model in the same design lineage that BB-8 is from, back before they sorted out the whole "floating head" thing.  I'd hesitate to call it BB-5 or whatever, because if that whole thing is the model, there's nothing left to identify the individual droids.  (It's bad enough that naming one of millions of R2 units "R2-D2" is roughly equivalent to calling your car "Mustang #64".  Star Wars naming conventions can be illogical sometimes. )

Anyway, the Ralph McQuarrie "ball mounted" concept for R2-D2 could've been the earliest model in this family, stylistically tying it to the R-series.



Various other designs like BaBy-8 could also go in there somewhere.  I like the one on the lower right here; it just looks so happy.  (I'm not sure that the others really look like BB-8 relatives.  The top one looks like a cross between a Viper probe droid and a LIN droid.)



Well, that's a project for the future.

[Dyne]

For the last few days I've been preoccupied with designing and sourcing parts for my enclosure.  Made a post about that though to be honest, I found a design for the leg extensions that I like better, so I might use that rather than my own.

For Treadwell, I've been working on normalizing the eyeboxes into proper truncated pyramids, which will simplify the eye mechanism a bit (the drawings upthread were based on the plans in the site section). Also, today I received my servos and Lipo bags, and my balance cable extender should be in tomorrow (along with a new router, so hopefully no more reboots screwing with my cad software).

[Dyne]

It's been a frustrating few weeks.  I've been working on my enclosure and fighting with the printer to get it to work reliably again.  (I took the opportunity to cut the Z brace threaded rods to the length they originally should've been, and removing the braces mucked with the printer's squareness more than expected, causing a large increase in print failures.)  Stuff is getting somewhat back to normal, apart from some of my filament not seeming to like the PTFE tube I'm using to guide it into the enclosure, so I'm back to working on Treadwell.


In my last post, I mentioned I'd started to work on revising the eyeboxes as prep for fleshing out my eye mechanisms, so that's where I'm picking things back up.  Why revising?  Well, I've noticed a couple of oddities in the plans here on the site...

* Firstly, the aspect ratio of the rear face of the eyebox (1.3571) does not match the aspect ratio of the front face (0.9821).   In other words, the front face is slightly taller than it is wide, but the rear face is much wider than it is tall.  That's rather peculiar, given the simple design of the part.  It's hard to be sure from eyeballing it, due to the perspective, but I don't think that that's accurate.  From having a look at the deleted vaparator scene, it seems to me that the rear faces of Treadwell's eyeboxes are relatively square, just like the front faces.  For example:

Don't turn around, oh oh ohhhh

The change I'm making here is to normalize both faces to exact squares.  The front face is presently 88 mm on each side, and the rear face is 52 mm on each side.  These numbers represent a compromise between the width and height I originally had for each face.


* Secondly, the original prop does not have the features that I called the "eyebrows" earlier in the thread. I knew all along that those came entirely from the plans here.  It's not a bad look, but I've decided to remove it in the interests of helping others be more accurate.  As alluded to in the thread on the Paterson projector, the lens holder simply appears to screw to the front face of the eyebox.  See here:

Show me those baby blues

The lens and its retainer must protrude slightly inside the eyebox, so the front face currently has a 66 mm hole to accommodate them.  I may need to tweak this opening since I believe the lens is not perfectly vertically centered in the Paterson holder (it's a little closer to the bottom edge than the top edge; the flange on the bottom makes this less obvious than it would otherwise be).  Haven't looked into it yet.


* Thirdly, as I hinted in my previous post, the truncated pyramid that forms the eyebox is skewed in the plans.  If you look back at the very first eyebox schematic that I posted, you'll see four angles that give the slope of the long faces of the eyebox relative to the rear face.  All four angles are different from one another.  This essentially moves the rear face of each eyebox closer toward the horizontal centerline of Treadwell's head, and upward from the vertical center.  My change here is to center the front and rear faces with each other.  Since both faces are also square, that means all of the slopes will be identical.


So those are a few things I'm fixing.  Here's a cross section of what I have rebuilt so far.

New Eyebox

This is arranged in the orientation that it would be printed in.  The two holes at top are for the lens holders.  As you can see, I've now added the vertical split (so it'll fit on my printer).  Each eyebox would be held together with three screws shown, and/or glue.  I might also work some captive nuts into those screw holes.

What you can't see in that pic is how the parts join together away from the centerline.  There is a little ridge that goes around the inner edge, except where the screws are.  Here's a closeup of just the rear half.

Eyebox Joint

This ridge was originally just a straightforward rectangular extrusion and helped hold the halves together, but then I realized that the other half of the eyebox would need support for the matching surface.  So it is now chamfered instead and just helps center the parts.  I'm not sure that I like this setup (among other things, it removes a good chunk of the plastic that'd be attaching the front section to the print bed).  Since I plan to use the screws regardless, and those are perfectly capable of aligning the parts, I might abandon it.

Printing in this orientation does mean that the layer lines won't be optimal if the printer doesn't produce a good bond, but I don't think the eyeboxes are going to weigh enough to cause delamination issues, even with my eye mechanisms.  Maybe they would if you used actual glass lenses, but I don't have those.

The alternative is printing with the eyebox lying flat -- which would require even more support -- or printing the faces flat, individually.  Or a compromise ... lie the part down, but leave off the topmost face to be printed separately.  The latter idea might be the most practical from the standpoint of putting in an eye mechanism, but my print bed is too small to print the eyebox length as one part, no matter which way it's turned, so I think it'd make the split I have now even worse.  I'm sticking with the design above for now.

[kresty]

Cool!

[Dyne]

Been awhile since I've posted or done significant work.  I've been distracted by other things for awhile.

I took some time over the last few days to merge Treadwell's revised eyeboxes into the main branch of my project, incorporate the neck piece, and fiddle with the project drawings to make use of the multi-page drawing feature that recently went into Fusion.  This is primarily for PDF output.  It makes drawings much more manageable.  Because of the way exploded views work in Fusion, I haven't yet found a way to integrate my lens subassembly drawing into the main PDF, so that one remains a separate file, which is fine.  Two separate drawings is still better than seven.

Most of the drawings are 1:1 scale, or larger.  I made the sheets A2 sized to make them a bit less busy.

This is the resulting PDF.  If the link doesn't work, the file is in my user gallery.


There is a bit of a glitch with Fusion 360 drawings where a line drawing will show components in the correct position, but a shaded drawing sometimes will show one or more components slightly shifted, or even in entirely the wrong place.  This can happen simultaneously, meaning the edges in shaded drawings are in a different position than the colored surfaces.  The result looks like this:

F360 Glitch
Just an illustration of wonky drawings in Fusion

Thus, I could only get away with shading a few drawings.


In mostly unrelated news, I was recently given one of these things, pretty much because it resembles an R2 unit.  Mine lacks the handle.  It's about 4 inches in diameter, and maybe 6 inches tall.  I think I will turn it into a miniature droid of my own design, though probably not an R2.

[kresty]

Wow, if that flashlight is the criteria for people giving you "artoo" stuff, you must be drowning!

[Dyne]

Wow, if that flashlight is the criteria for people giving you "artoo" stuff, you must be drowning!

Heh.  Without the handle, the resemblance to Artoo's silhouette is more obvious -- cylindrical body, domed head, and two short "shoulder mounts" that some clearly missing part(s) once attached to.

I guess that's why it was originally described to me as some sort of "robot" that the source had gotten from some long ago promotion.  That description blinded me from seeing what it actually was at first.  I was imagining it was meant to be something along the lines of a toy Cobot, like this one

Though it clearly has a flashlight bulb and a reflector in the "dome", I thought these were intended as some sort of cheesy flashing light for the robot.  There's also a big sliding actuator in back.  It raises the bulb up out of the body a short distance.  In reality, that's a selector for a wide or narrow angle beam, but when you are lead to believe the thing is a toy robot from the early 80s, the phrase "Now with long distance scanner action!" runs through your head.

[kresty]

Though it clearly has a flashlight bulb and a reflector in the "dome", I thought these were intended as some sort of cheesy flashing light for the robot.  There's also a big sliding actuator in back.  It raises the bulb up out of the body a short distance.  In reality, that's a selector for a wide or narrow angle beam, but when you are lead to believe the thing is a toy robot from the early 80s, the phrase "Now with long distance scanner action!" runs through your head.

Oh, I'm sorry I was mistaken, I didn't realize it was a toy robot ;-) (Don't wanna spoil your fun)

[Dyne]

Almost a year has gone by since I started on my Treadwell, and I thought it was high time I actually made something.  Other than CAD models, that is.  Something tangible.

Where have I been, you ask?  

I've mentioned here and there that my printer has been sick.  Specifically, it has been printing parts with a consistent lean toward the negative Y axis, rather than the random layer shifts you'd get with a slipping belt.  This was a problem for any sort of functional prints (i.e. Treadwell and Artie Deco's drivetrain).  Wheels ended up with a pronounced wobble, screw holes were distorted enough to cause the screws to bind up, etc.  I've been puzzling over the problem off and on for months.  It completely blocked Treadwell, because I can't print my eyeboxes if they are going to be bent all out of whack.  I spent a long time where I just didn't print anything at all.

A few weeks before Dragon Con, I finally started printing again; a mostly aesthetic project -- my teapot racer for the alternate history track -- so I didn't have to worry about the distortion.  But as a result, it was fresh in mind during the convention.  After Dragon Con ended, I took advantage of the fact that we were in Atlanta to drop by the Micro Center there (since we don't have one locally) to replace filament I used for my racer and get another Pi Zero.  I mentioned the problem to the guy staffing the 3d printer area.  He was as puzzled by it as I was, and didn't really have any suggestions.  

As it happens this was the beginning of the end, because it started really annoying me and forced me to think about it rather than randomly trying things, especially once I started discussing the issue with my friend during the trip back home.  I realized it couldn't be an issue with any one axis, and the two that are affected (Y and Z) are mechanically unrelated.  Once I got home, I was absently staring at the printer when the solution suddenly smacked me.  Or possibly my palm connected with my face.  

Remember back in march when I was building the enclosure, I said that I took advantage of the downtime to remove and trim down my Z braces?  The Z braces are there to hold the Z gantry rigid, and do a very good job of it, so I basically never have to think about that part of the printer as a source of problems.  And so I didn't.  When I put the braces back on, I squared the gantry as usual and then promptly forgot it was even a thing.

It was blindingly obvious... the gantry was not actually square to the rest of the printer.  It leaned forward.  And therefore, so did every single part printed since then.  But since I had (seemingly) squared it, that possibility literally never occurred to me.

Massive fail.  I'll be turning in my 3d printer operator's license soon.

In the meantime, I said at the start of this post that I would actually be making something tangible.  I commented elsewhere that I've been considering resin-casting my Treadwell lenses, and one of the first prints I did after the printer was working was a test of that part.  I just did another print today with a lens retainer to test how well they fit together (perfectly, as it turns out).  

So here it is, the first official part for my Treadwell (the blue retainer)...

Treadwell Lens and Retainer

Someday I might move up to printing something I've actually designed from scratch...

As I mentioned in the other thread, the lens was sliced with Prusa's Smooth Variable Layer Height feature.  Basically the entire upper surface of the lens uses 0.06 mm layers for smoothness.  I could potentially go thinner than this, but I'm not sure what the lower bound is on this printer, and I know that this value works.  If I sand the concave surface a bit, and possibly wax coat it, I should get something that might actually be a reasonable master for resin casting.  I just have to be careful about dimensional accuracy.  Much of rest of the lens was printed with 0.3 mm layers for speed, aside from the area where it shifts gradually from thick layers to thin.

Regarding the rest of the droid, it has occurred to me that I might want to employ the trick used by the red carpet BB-8: control Treadwell's head tilt with bowden-style bicycle cables.  That would let me move some stuff to the base, which might help avoid some of the wobble-producing weight at the end of Treadwell's stick body.  It would require moving the slip ring and probably the head rotation mechanisms to the base as well, as the tilt servo has to feed through the slip ring in order to allow the servo (and the cables attached to it) to rotate with the head.  If not, the bicycle cables would get twisted up and quickly prevent full head rotation.