Dyne's Treadwell

[Dyne]

Third mold's the charm?

Though I got closer to a good mold last time, I also discovered the largest source of defects on my lenses.  With the silicone caulk-in-soapy water method that I've been playing with thus far, the silicone is immersed.  While this does help curing, water may also linger on the surface (where it is difficult to see against the translucent silicone).  It wasn't air bubbles getting trapped in the mold, but water.

To address that, I decided to try one of the corn starch methods, which people here are likely familiar with as oogoo or proto-putty.  There are lots of variants with various ratios of starch to silicone, some with thinners (to make it pourable), and some with colorants (like acrylic paint).  I used the version at the second link, which uses food coloring.

The food coloring (being partly water) gets mixed into the silicone, helping it cure without needing as much humidity from the air.  Obviously, it also colors the silicone.  The corn starch fills a similar purpose as the dish soap did, keeping the silicone from being so damnably sticky (though I never quite got it totally non-sticky).  It also absorbs any unmixed food coloring, preventing moisture from getting trapped between the mold and the part, and it allegedly helps with curing.

You just have to make sure to get a nice flat surface against the lens, and of course, make sure that you don't trap any air in the process.  The food coloring helps both requirements by making the material opaque and colored, rendering the surfaces (and any possible crevices) easier to see.

The change in process paid off to a certain extent; my third mold -- the "Blue Mold" -- came out with about as good a surface as I could've hoped for without a pourable, degassed silicone.  I put the master back in the cavity, shored up the bottom of the mold with additional silicone (the center of the lens was just a thin skin due to the too-small quantity of silicone I made the mold with originally), pressed the newly added silicone onto a piece of foam board to form a level base (being originally in the form of an upside-down rounded lump, it otherwise would've been hard to keep the mold level during pouring), and added a rim around the open face of the mold to contain any excess resin.  Finally, I poured a new lens to test it out.

While waiting for that to cure, I decided to make an additional mold using this same process, primarily so I can cast two decent lenses at the same time.  This one also turned out pretty well.  I was shooting for orange, but ended up with a fleshy/meaty pink, which I will call the "Red Mold" (think Red/Blue stereo anaglyph glasses).

24 hours after the Blue Mold lens was poured, I demolded the part.  I hoped that my more precise ratio would bring the cure time back to where it was with the first lens, but once again the part was slightly sticky on the silicone mating surfaces (not as much as last time, though).  It picked up a slight texture from the uncured resin pulling away from the mold.  I washed most of that off with isopropyl alcohol and then water, but some remained.  Otherwise this would probably be the clearest lens yet.

By the time I demolded the blue lens, the Red mold had also been leveled and had cured, so I cleaned the blue mold of leftover resin (much easier than cleaning the lens, actually) and poured both molds.  I also poured mold #2 from last time with the leftover resin of this batch.  I had nothing better to use it on, and it'll bring the number of clear lenses I'll have to an even 6.

I'm not sure why my very first pour cured so much faster than the others, but I think that I'm just going to have to allow the full stated time for these latest parts to cure.  That's fine; I really need to spend more time on my Halloween projects (all this mold making also gave me a few ideas).

As for my parts in post #30, it took more than 48 hours after pouring for the clear Treadwell lens to lose all tackiness.  I still haven't sanded it.  The dyed resin was worse; it didn't fully cure until around Friday.  I could still bend the dyed Treadwell lens fairly easily until then.  It is also cloudy due to moisture fogging (which is fine for that part, since it acts as a diffuser).

[Dyne]

Getting Artie Deco rolling took up a lot of my available post-Halloween time and energy last year, so I've been in lurk mode for awhile.

However, as some of you may have read over in the Events subforum, I recently came back from Dragon Con 2018, where I had taken Artie and spent the large majority of the convention hanging around with other Droid Builders. I'm committed to (trying to) get an actual Star Wars droid running in time to take to next year's convention.

A sane person would go with an MSE droid or something, but if Artie proves anything it's that I gotta be different. Since Treadwell is already in progress, obviously I need to continue working on him. To that end, I'm back at it in this thread.


As you may recall upthread, Treadwell mostly exists as CAD. I purchased resin, primer, etc. for casting the lenses. but those are kind of general purpose materials rather than specific to this project.

So after I got home this week, I made my first official Treadwell purchases:

* A second pair of PS3 Nav controllers: $4 (gamestop has used controllers under two bucks these days)
* A pair of blower belts for Treadwell's treads. I used BDS, the supplier in the first link in this post that Kresty made (I saw other listings for over $90 each so this was good)

I also bought one of those Ozark Trail collapsible wagons from Walmart. A couple of the BB-8 builders had these at Dragon Con for transporting their droids to and from their rooms, and I was super jealous. They can carry up to 225 pounds and collapse into a small space.

I knew from my 2016 experience with Artie that I'd want some sort of droid caddy (especially if I ever plan to also bring BBL-BBQ), it's just that once I'd reassembled Artie and was certain I was taking him, it was too late to address the matter. I was hoping I could drive him most of the time, but we were in the Westin this year; Peachtree center (crowded food court), stairs, escalators, and several bumpy outdoor sidewalks and crosswalks lay between me and the droid hangout. I also had a few drive malfunctions that kept me from driving at times. As a result, I was carrying Artie around by hand a lot.

Being a 1950s era steel vacuum cleaner, Artie weighs nearly 14 pounds just as an empty shell. All his sundry innards like batteries, speaker, motors, the wood base his motors mount to, etc. probably push him up somewhere between 20 and 30 pounds. Then add a backpack with controllers, chargers, hydration, Sphero droids, a Hardware Wars saber (chrome flashlight), etc... Yeah, I bought a wagon as soon as I found the one I wanted. I'm not sure if it counts as a Treadwell part, but it'll probably be used for him in some capacity.


Now I am pondering the drive system and what to construct Treadwell's base out of.

I lean toward using Styrene rather than wood in order to save weight. I already foresee children trying to stand on Treadwell's base, but I also know that well designed styrene droids are quite capable of supporting the weight of an adult human. (For that matter, I've seen Matt Hobbs climb on top of his Darth Porg MSE and ride it around ... I'm not sure what that thing is made of, but it is quite sturdy). In any event, my plan is to make sure he's well-supported and use the base as a "luggage rack" (load it with prop gear) to discourage people trying to ride on him.

I know from Paul's build log that Treadwells can be quite power hungry due to the friction of the treads. I'm wondering whether going the (more efficient) brushless motor route would be help with the power draw. I know they aren't cheap, but I'll have to run numbers.

[Dyne]

I know from Paul's build log that Treadwells can be quite power hungry due to the friction of the treads.  I'm wondering whether going the (more efficient) brushless motor route would be help with the power draw.  I know they aren't cheap, but I'll have to run numbers.

Following up on this, I've compiled a spreadsheet with some variants, really more as a summary of information scattered around Astromech than any new analysis.  I'm not including shipping in any of my prices here, as it depends on too many variables, but I record it in my spreadsheet if it's unusually high (motors, belts).

Firstly, as a baseline, I added up the brushed scooter motors and ESC that Paul6700 was using, even though this specific ESC is discontinued (I found a few sources that gave this price):

• Scooter Motors @ $23.16 x2
  
• Roboteq AX2550 @ $495 x1
  
• Total: 541.32
  

The RoboteQ ESCs that Oliver Steeples and others use with the Q85s (see the ESC section later) are perhaps the nearest equivalent, and cost about the same.


Oliver also posted his alternative to the Jaycar motor setup on astromech.net for a styrene R2:

• Banebots 16:1 P60 gearbox @ $54.50 x2
  
• Multistar 268kv brushless motor @ $40.24 x2
  
• HobbyKing 45 brushless motor controller @ $21.37 x2
  
• Scorpion Power Systems HK-2208 Shaft Kit @ $4 x2
  
• Hobbyking programmer @ $6.58 x1
  
• Total: $246.80
  

He mentioned considering the lower 30 amp version of this ESC, because it doesn't have a fan and thus runs quieter.

Only issue with this Jaycar alternative setup is that I've no idea if it would be sufficient for a styrene Treadwell or not.  The treads make the droid different than the R2 this was designed for, while the styrene+brushless motors make the weight and power output different than Paul's Treadwell.  

---

Moving on to other setups, I've split the options up into the motors and the ESC.  Mostly the ESCs.

Motor Considerations

The Jaycar alternative works for styrene droids, which are usually fairly light.  The current favorite brushless motor for heavier R2 units is the Q85 ebike hub motor.  These cost around $80 each, and require modification so that the droid can reverse.  Premodified motors are available for $190 each.  I'd need 2.  

The entire motor rotates on its axle and these have LOTS of torque -- even one motor is quite capable of carrying an adult human (which is, after all, the intended function) ... or bending a bicycle fork, which is why they require a fairly hefty mount.  

I have my doubts this motor would be a good fit for Treadwell, not only due to the excessive torque for a styrene droid, but also due to the physical size and mounting considerations.  

After modification (turning or bandsawing most of the spoke flanges away), I think the motor with no tire ends up being around 4.25".  Treadwell's drive wheels are 4", so the Q85 could only be used directly as the center idler wheels (which are 1/16" larger than the motor).  It wouldn't have as much grip on the treads there.  Even worse, Treadwell's setup needs two 1.25" thick wheels with a gap in between, while the tire mounting portion of the Q85 is about 1.5" wide and the motor casing extends beyond that.

Not only does the Q85 need a beefy mount, but since the motor itself spins, you are really meant to support both ends of the axle.  Not sure how necessary that is, but it really wouldn't work with a standard Treadwell layout.  So I think the easiest way you could use this motor would be to turn the drive wheels indirectly using gears and a chain, like a Senna R2 drive system.

If you aren't set on screen accuracy, you could add a track carrier, kind of like the two Mountain State Treadwells in the foreground below, but at that point you can modify the treads however you like:

Mountain States Droid Builders

There are a variety of other brushless motors one could get.  Unfortunately, I don't know much about sizing brushless motors, so I'm going to have to think about this before listing any.  (Frankly, before I built Artie, I didn't know much about sizing brushed motors either, and originally bought a pair of these woefully anemic things.)

---

ESCs

On the original thread about the Q85 on Astromech, Oliver listed the Roboteq SBL1360 ESC @ $275 x2.  Some people favor the SBL2360 @ $495 instead, slightly cheaper since you only need one.  Oliver also mentioned the Turnigy TrackStar 200A 8s as a cheaper alternative.

Experiments have been done with these ESCs

• Turnigy TrackStar Gen II @ $87.03 x2 (poor low speed performance)
  
• a chinese ESC referred to as JYQD_V7.3E (pretty cheap @ $12.33 x2, but it lacks braking and has a noticable delay before reversing direction)
  
• Alien Power Systems 120A @ $89.16 x2.  
  
• James Bruton has recently been making use of the O Drive speed controller for his brushless motors in OpenDog Project, after initially mentioning it in his Exosuit build.  A dual channel 24v controller with connectors is $129.
  
• I think people were planning to test the VESC open source controllers (which James used in his exosuit video), but I don't know the results.
  

So depending on the specific options, a brushless motor setup could run anywhere from about $170 (using the $12 ESC) to over $700 (dual SBL1360).

Assuming that I can't get away with Oliver's Jaycar setup on a Styrene treadwell due to increased friction, I'm leaning most toward the O Drive, assuming it works with whatever motors I select.

[Dyne]

Treadwell got derailed a bit this year because of real life chaos, but I am still planning to build him.  Here are a couple of minor rambling thoughts, since I haven't updated in awhile.


Over two years ago (back in post 20), I started working on resin casting lenses.  I ultimately got a couple of decent parts but even those had obvious defects from various causes.  To really improve, I would need one or both of a vacuum chamber and a pressure chamber, plus significant improvements to my mold making materials, polishing to a much higher level than I've done so far, and probably cast with better resin as well (clear resin is notorious for yellowing, and this certainly happened to my el cheapo polyurethane parts).  

I could conceivably afford one or two of those things, but not several, so that's as far as I got.

Well, shortly before Dragon Con, I picked up an Elegoo Mars mSLA printer -- one of the new generation of (currently) sub $300 resin machines.  I love the machine.  It's far less fiddly than my FDM printer, and undoubtedly less prone to starting house fires as well.  Since my FDM machine started acting up right before the convention, I've been using the Mars exclusively, mostly to prototype my upgraded magnet holder for my modded spin master BB unit.

I know some of you likely already have experience with resin printers.  You'll know that even on the coarse 0.05mm layer height, prints often require little post processing to get a good surface (aside from supports).  And you are undoubtedly aware that clear printing resins are a thing ... and end up much clearer than "clear" FDM filament.

I think you might see where I'm going with this...   Yes, I just ordered some clear resin so I can experiment with printing Treadwell lenses directly.   I'll have to be careful with the post processing, as IPA apparently causes more of a frosted look, so Simple Green or Mean Green may be better.  We'll see...


Back in post #4, I mentioned wanting to add facial tracking to Treadwell to let it turn to look at people as it drove along, and linked to one of James Bruton's Ultron videos where he did something similar.  At the time, James was using a Pixy camera board with the tracking built in.  As I didn't have one of those, I started experimenting with OpenCV instead, and got pretty far with my experiments before a lack of an actual Droid and the corruption of my Raspbian install (where i was coding) put me back to square one.

In his most recent video, James has gone down the same OpenCV path for his performance robots.   See around 11:20 timestamp.  He is using DMX rather than direct servo control, but it's the same idea.


Finally, a bit of a digression here, but it sort of connects to this log, eventually...

As a giant fan of Halloween, I've been wanting to get my hands on some of the Disney Halloween themed Astromechs for awhile (R2-B00, R3-B0017, etc.), but since they are yearly park exclusives, they get harder to find after the fact... at least without paying inflated prices on ebay.  I've seen people asking above $100 for the 2017 droid for example.

I got a bit fed up a few weeks ago, when I was unable to order R5-B0019 -- this year's Droid -- from the local Disney store or their web store, only five weeks after it was released and well before Halloween.  (I have since gotten one, but at the time I hadn't.)

But I did end up getting a Black series Chopper.  From there, I soon learned that while the black series is nominally 1/12 scale, their R2-D2 is undersized compared with the rest of the line.  However, there are 1/12 Bandai droid models that were in the correct scale.  That gave me the idea -- I could build a few of these Halloween droids in the larger scale using the model kits.  They'd fit with Chopper and be more detailed than the figures. So I did exactly that with R2-B00, using the R2-D2 and R5-D4 kit.



I plan to also do 1/12 versions of R3-B0017 and R5-B0019 (R4-B0018 is decorated like candy corn; I'm not a big fan of that one).  Just one problem:  R3-B0017 is cast in translucent green glow-in-the-dark plastic.

So how does any of that relate to this log?

Well, you may recall that one of my lens casting experiments was adding highlighter dye to get UV reactive lenses for Treadwell (and BB-8).  This sort of worked, but the dye is water based, and urethane doesn't like water.  At all.  

I did some research.  As it happens, Glow-in-the-Dark printing resin IS a thing.  It is called Liqcreate Hazard Glow.

It is unfortunately very expensive... Matterhackers carries it for $85 per 250 ml. (For comparison, a liter of my normal Siraya tech grey fast resin is $40.  The same amount of the Liqcreate would cost almost $100 more than my printer did.)

I doubt I could ever afford to print a full sized Michael Baddeley R-series body with that stuff, even on that Peopoly phenom large format resin printer. But a 1/12 version with a R3 dome printed in mixed clear/GitD resin, or some Treadwell or BB unit lenses ... maybe.


The other relationship is that I'm now tempted to model and print a 1/12 treadwell.  I know there was a resin kit released awhile back, but I don't think it's very easy to get.

[savagecreature]

fun post. Thanks for letting us know what you've been up to. These kinds of posts are always an inspiration to me, and a bit of a relief to know that life ruins other builders plans, too.

As far as lenses for the treadwell go, that's been something Iv'e been considering as well. 3D printed resin seems like it might be a good solution, but I don't happen to have a resin printer. However, I do have 3 CnCs, which just might be the way to do it. Check out this video

https://www.youtube.com/watch?v=ZW2lj0KjyT8

Since we have the correct lens frames, sure would be cool to have high quality lenses. Now I just need to find the time to spend really testing it out.

[Dyne]

Check out this video

https://www.youtube.com/watch?v=ZW2lj0KjyT8

Since we have the correct lens frames, sure  would be cool to have high quality lenses.  Now I just need to find the time to spend really testing it out.

I don't have a CNC, but I like that video. It does make me wonder if you could put polishing wheels on the toolhead instead of an end mill. Might save a bit of manual work.

I have the clear resin now, but without thinking about it, I'd discarded my first bottle for the opaque resin when I put the last of it into the tank, and the second bottle is unopened and full. So I can't test until I either use enough resin to empty my tank without worrying about overflow, find another bottle to use, or Elegoo gets around to releasing their 4-in-1 tank pack on Amazon so I can have multiples.

The drawback with printing the lens is that the lenses would have to be printed solid. That means either printing on supports and dealing with the defects, or printing flat and hoping that the peel force of such a large area doesn't cause a print failure. And because it'd be a solid lens, it'll use a fair amount of resin

[Dyne]

Oops, it appears that I completely forgot to post the results of my printed lens test.



The best of the lenses that I've cast (the one shown at the end of post #25) is on the right.  The printed lens is on the left.  It is partially sanded in this pic; I think the circular ridges around the outside are artifacts due to the quality of the mesh when I exported it and not so much the print.  I pulled it out the other day and cleaned those up, but I'll need to re-polish it before I take another picture.


In other news, I have a large format printer on the way.  Hopefully I can resume actual progress on my builds soon.

[kresty]

That's pretty cool. How "clear" do you expect it to get?

[Dyne]

That's pretty cool.  How "clear" do you expect it to get?

Pretty clear...

Resin printed lens after polish

That's after going up to 2000 grit sandpaper for a few minutes, and a few more minutes of Novus 3 and 2.

Compare to the picture from back in post 25 that I took with the old lens, which took a lot more work to get to this stage...

Old polyurethane cast lens

The new lens is not without defects, but they are mostly faint layer lines on the back or weird edges. The worst is visible in my previous post, in the left lens near the top of the image, kind of a grey horizontal smudge. It's actually one of those artifacts caused by the a part of the print partially detaching and reattaching. But I can live with it.

[kresty]

That's pretty clear - and it's cool that it even acts like a lens! Very fun.