Dyne's LD-F1 (D-O concept) build

[savagecreature]

That's lookin' pretty darn cool. You're definitely inspiring me to work harder on my builds. Very nice work

Not following the concept art exactly is one of those grey areas at work. Sometimes the concept is just to suggest a direction, sometimes the concept is *exactly* what they expect us to create. Never quite knowing how much variance is acceptable has made me a little "overly sensitive" to the subject, I think

[Dyne]

Does anyone understand levers well enough to double-check my logic here?


First, my belief is that LD-F1's lift acts as a third class lever (The Effort E is applied *between* the fulcrum F and the load L).  Because I didn't sufficiently consider this before buying the servo, I now need to find the Mechanical Advantage (M) of this lever for any given placement of the servo, to make sure that I position it where it is most likely to work successfully.  


I know the formula for M.  If A is the distance from Fulcrum to Effort, and B is the distance from Fulcrum to Load, then M = A / B.

This is also the inverse ratio for the forces acting at points A and B.  That is, M = L / E.  If we solve for E (the force needed to lift the weight at point L), then E = L / M.

Since I'm concerned with forces, a higher M is better.  The more that the load gets divided by, the better.  This also means that the closer A is to B, the better.

With a third class lever, however, M is always less than 1 (A is always less than B).  If M was greater than 1, then this wouldn't be a third class lever anymore.  If M equaled 1, it would mean that A and B are the same length, so the servo would be directly lifting the load, it could use its full power, and the lever would be irrelevant.  Obviously neither is the situation we have here.

E (the weight of L divided by M) must be below the maximum thrust of the servo.  Alternately, the maximum thrust multiplied by M must be above the weight of L.  Probably with a decent safety factor.


Figuring out length B is simple, since those are two known pivot points in my lift mechanism.  They are separated by 274.181 mm

The problem is that I found it a bit hard to understand what the value of A should be in my situation.  The "lever" is not a real physical part, and the effort isn't actually applied directly to it but rather to some point above it.  So I needed a generalized method to find A.  

I've came up with (and dismissed) a couple of low confidence ideas before I arrived at what I have below, so I hope it's correct.



My current thinking is based on the fact that the lever only cares where along its length the effort is applied.  Any given lever with the effort applied at an arbitrary distance from one end is exactly equivalent to a parallel lever of the same length with effort applied at the same relative distance.

This image may make the situation clearer:




The (non-physical) lever is the purple line FL, which has length B.  

The servo is the line between points E and P.  It has length 104 mm because the servo actually has minimum 102 and I applied a 2 mm buffer to avoid the endstop.  Neither point E nor P must necessarily be where they are shown, as that's what I'm trying to determine.  But E *must* fall somewhere along the black dashed line (which is just an offset from the real underside of the body shell -- the purple dashed line -- to account for mounting hardware).

As you can see, I've drawn a line parallel to line FL and with the same length.  That's the parallel lever.  This new line is the one that passes through point E, wherever E happens to be along the black dashed line in the setup I'm checking.  I've named its endpoints "vF" for virtual fulcrum, and "vL" for virtual load.  Now L, F, vF and vL form the vertices of a rectangle.  I don't suppose that it must be a rectangle, rather than an arbitrary parallelogram, but for drawing it out, this is the easiest way.

My thinking is that the line between E and vF is the effective value of A.  You can imagine projecting this point onto the lever (line FL) using a line drawn parallel to the small sides of the rectangle, if that helps.

Does this make any sense?  



In this diagram, this gives me A = 166.30.  With B fixed at 274.181, I get M = A/B = ~0.61.


The weight of the parts that I have currently assembled is around 1.25 kg or 2.67 lbs.  There are more parts to add, so let's estimate the final weight (L) as 3.5 lbs or 1.59 kg.  

Using E = L / M, as described above, we have E = 3.5 / 0.61 = 5.83 lbs.  The servo needs to apply 5.83 pounds of force to lift 3.5 lbs with this particular setup.  Since the most we can have is 11.5, that's well within spec, so the servo would work.

Specifically, it would work IF this servo has a long enough stroke in this arrangement to lift as far as I need it to, and it doesn't collide with any other parts or occupy any impossible positions.  I know already that it's problematic to put point P concentric with the center of the lower stalk axles, as shown here, because the hole size in the servo is about half the diameter of the axle.  Also, to reach the full 45 degrees of lift, the servo would need a longer than 50 mm stroke.  Not by much, but still.

That's why I'm checking different solutions.  Gotta find the optimal combination of stroke length, lift range, and mechanical advantage that doesn't cause other problems.


Edit: I did think of another possible way to find A, which is to draw an arc centered at F with the radius FE, down from point E until it intersects the line FL at a point I'll call R (it'd be near the letter B in the diagram). Then A is the length FR. This would give me a slightly longer A than above, but not by much since the rectangle isn't very wide.

Or I could project the SERVO through an arc centered at P to figure out where it intersects FL (where a servo of this length would attach, if that lever was a real physical part). No idea if either of these make any mathematical sense as far as calculating the lever is concerned, though.

[Dyne]

T-Minus 4 weeks (and 1 day)

Since post #18, my prints started to be plagued by rapidly decreasing quality and increasing failures (including skipping and gouging at the extruder).



Eventually, I concluded that my print head's PTFE liner was likely to be damaged or deformed.  Swapped it for the spare print head that came with the printer and re-leveled, just in case the new print head was at a slightly different height.  Aside from probably needing to dial in the level a bit more, things appear to be working normally again.


For the tracks, production is in full swing.  As of right now, 50 of the final links have been printed and assembled -- just short of the 53 needed for the first full track -- as well as 16 spare links using the slightly older v5 model.  Turns out that that one wasn't final after all. The only real change, though, was to make the centering mechanism centered from the front of the link to the back; previously it extended out further on one side than the other, over the center interlocking "tooth".  The old links are still more or less compatible with the new ones in an pinch, especially if used one-by-one, but notice how the teeth don't line up at the same height in the closeup below?  I wanted to fix that before full production started in case it wound up mattering, so we're now on V6.


Enough links to complete a full track and then some should be done by tomorrow.  They aren't being printed in Inland Silver PETG as the ones shown here were, but instead with their "Silver" PLA, which is really just a light grey.  They will be mostly hidden beneath the traction pads, so I can always paint the pads if I want a different color.


I've also got the parts for one of the two front driven sprockets printed and the main body for the other, so I just have six toothed wheels and two bodies for the rear sprockets to go.  I have some concerns about how well the teeth are meshing with the links -- it initially seemed right on the money, but I think it's very slightly off.  The design allows me to attach new toothed wheels if I need to tweak them, without having to redo the entire sprocket.

Most importantly, I got several parts of the chassis designed and printed them this weekend.  As such, I'm feeling a bit more confident that I can at least get this thing mobile, even if I have to forego the lift system for now and it doesn't get aesthetically complete or fully programmed until after Dragon Con.


Here's how the chassis is organized:


The blue assembly -- which is not fully designed yet -- is essentially the rear of the droid's chassis, though some of it protrudes into the forward portion.  The main part will carry the lift system, rear axle and sprockets, and the mounts for the track carriers outboard of each track, as well as the mounts for all 6 idlers in each track.  The bit that extends into the forward part of the droid will house many of the electronics and contains a curved recess for the main joint in the neck stalks to rest on when the head is lowered.  The blue parts are all essentially bolted together.

The red assembly carries the front axle, drive sprockets, the motor and pulleys.  This assembly attaches to the blue parts that extend into the forward section via M5 screws through the three horizontal slots that sit just below the motor and the second stage pulley.  This allows the red assembly it to slide back and forth relative to the blue assembly to tension the track.  The forward blue parts also have a slot for the front axle to slide with the red parts.

There are five red parts in total, and these are the ones I printed this weekend: The bottom plate, plus the inboard and outboard plates for each side, all bolted together.  The inboard plate forms the walls of the opening in the middle front of the droid, and the outboard plates faces the tracks.  There is a gap between the two, as well as a captive nut for the forward axle.  The side plates also have bolts that hold on the front shell of the droid.


The outboard plate has the mounts for the motor and the second stage pulley, as well as holes to attach an as-yet-unmodeled brace for the other end of the threaded rod that the second stage pulley turns on.  Between the belts, pulley, idler, and the teeth in the track links, space in this area is tight, as you can see below, so I may end up securing the other end of the pulley's threaded rod in the track carrier instead.


It also has a slot between motor and pulley for the motor wires to pass into the gap between the side plates, where they run forward and into the blue parts that house the electronics.


This is what that red assembly looks like in reality.  Note that the sprocket axle will be a threaded rod all the way through the droid, not a bolt like I currently have installed.  No reason you couldn't use a bolt, but I want a through axle for transverse rigidity.  I have temporarily placed threaded rods in the holes at the front where the front body shell attaches.



I like that the droid's left inner side plate (the one furthest from camera) has a scar defect near the top ... this will be visible once the droid is complete and suggests some "damage" weathering.  I, too, have a recent scar defect, as today this droid received his traditional blood sacrifice (I jabbed myself with the printer's scraper while trying to remove the support from that bottom plate).


I haven't even tried sourcing sounds for the droid yet, or really addressing audio beyond digging out a spare transducer and amplifier.  Since this droid hasn't appeared on screen, I can do pretty much whatever I want.  I think I mentioned previously that I've considered using Michael Perl's Human-Cyborg Relations D-O voice modulator (since this droid is technically what D-O could have looked like).  Unfortunately, that app isn't out yet.  Another idea is to use the sounds that D-O had on stage at Celebration 2019 (before he could speak).





Speaking of sounds ... does anybody have any clean Babu Frik dialogue samples?  Asking for a friend.  I've looked around, but haven't found any yet.  (I do have the stuffed toy that plays them, but I want to trigger them remotely, and accessing the wiring to trigger it is not trivial, and recording the sounds from the toy wouldn't give great quality.  Playing clean samples through the droid's amplifier would work much better than either of those methods.  (Maybe I'll ask Matt Hobbs about the sounds he has in his Mouse when Babu is driving it rather than the Porg.)


At any rate, I hope to have the rest of the main chassis design done and at least partly printed by the end of next weekend.

[kresty]

Ack, you're mixing lbs and mm!   Which I suppose could be more confusing because 5lb mass is only 5lbf if the interesting direction happens to include gravity.  

I'm confused.  It's not rotating around P?  In the earlier animation, I thought the bar rotated around it's endpoint?

I guess you're saying that the "real" lever is BC in the diagram you'd posted earlier?  Since that's the hinged point of the cover and the place it rotates about?


So, the lever is kinda like BC, except that it's even longer because the weight is all over at A?  

Two thoughts:

First is that the diagram of forces is going to change as the model moves from it's closed to fully open position.  You'll want to position your servo in a place that is highly effective in the worst positions.

Second is that you can simplify a rigid structure to a line for the calculations if it's rigid.  If the loads are at the ends, and if the effort is along the line, then it's easy enough to so the math.  if the effort isn't along the line, then it gets more complicated.  You'd have to have a second line where the servo attached to the fulcrum.  But the math's the same, presuming the servo's effort is perpendicular to it's line.

The math is basically 'just' saying "how much does this mass (or effort) want this lever to rotate around the endpoint?"  Which is really easy, it's just the force times the distance.  If you have 12 points with mass and 15 efforts in all sorts of weird places on a strange rigid structure all rotating around the same point, it's still the same math.  + or - forces multiplied by the length of their lever arms.

In other words, if, in your diagram, the lever rotates around F, and your mass load is on L (It's not, there's all that light green stuff to the left of that, so the mass has even more leverage).  But, assuming that it was on L, then the load would be the mass on L times the length LF (B) - if the B was parallel to the ground.  It's not, so you'd have to figure out how much the load was perpendicular to the LF (B) line.  Or just go with it because it's close to parallel.


Anyhoo. If you're applying the effort from your servo to E, then it's lever arm around F is going to be the line EF, not A.  AND that's presuming the effort being applied perpendicularly to the lever.  (The obvious other extreme would be the servo being mounted parallel to the EF lever.  In which case it doesn't matter how hard it pushes, all it can do is break the lever, it can't make it pivot.)

Here's a video I found from a quick search.  I didn't check for accuracy or whatever, but it shows what happens if a force isn't perpendicular to the lever:

 Ch 8 - Torque - Calculating Lever Arm and Torque - YouTube

[Dyne]

Ack, you're mixing lbs and mm!   Which I suppose could be more confusing because 5lb mass is only 5lbf if the interesting direction happens to include gravity.

If you're pointing out that I'm mixing metric and US units ... that's mostly because the servo rating was given in lbs of thrust rather than kg or newtons or whatever, and I typically weigh things in that as well, while I prefer using metric as the default length unit in CAD.

I'm confused.  It's not rotating around P?  In the earlier animation, I thought the bar rotated around it's endpoint?

In the earlier animation, I was showing the parallelogram.  I skipping calling out most of that in the more recent diagram, for simplicity's sake.  It just seemed to confuse the issue when I tried to describe the problem to friends.

To answer your question, the green arm does indeed rotate around P (and in the arrangement shown, the servo does as well, but this arrangement isn't final).  

The body shell (or rather the imaginary line between the pivots on each end of it) forms the arm parallel to that green bar, and this arm rotates around F.  Those are the two "ground" pivots of the four bar mechanism.  The other two pivots are L and the unlabeled green circle a bit forward of L.

I guess you're saying that the "real" lever is BC in the diagram you'd posted earlier?  Since that's the hinged point of the cover and the place it rotates about?

So, the lever is kinda like BC, except that it's even longer because the weight is all over at A?

Yeah, mostly I was using the old diagram's BC (FL in the new one) as the "real" lever because that's the closest part of the parallelogram to where the servo force is actually being applied, so it seemed best to use that bit, omitting the rest of the parallelogram (on the grounds that if I can't calculate the values when there's just one arm in the lever, adding an entire four-bar linkage into the mix is just going to make it worse).

It wasn't as clear in the old diagram, but in the real build, A and B are just two pivots on the same single part that forms the upper stalks of the neck (the "light green stuff"). So all of the weight is bearing on the entire length AB. I guess you could say SOME of the weight is all the way over at A, and presumably the rest at B, but your guess is as good as mine as to what percentage goes where. Or maybe all of it is *effectively* at A.

Two thoughts:

First is that the diagram of forces is going to change as the model moves from it's closed to fully open position.  You'll want to position your servo in a place that is highly effective in the worst positions.

Yes.  In one of my earlier attempts I was trying to calculate the effective distance of the effort from F by another method, and I realized that it always gave a range of results between the two positions because the equivalent of point E was shifting, sometimes quite substantially depending on the servo placement.  

I think it's sufficient to consider two cases: when the head is lifted, and when it is lowered but trying to rise.  The latter is, I assume, typically going to be the worst position, since it's fighting gravity and the angle is generally the furthest from perpendicular.  

When the head is already raised, the servo is just handling the static weight (which it is pretty good at, even unpowered), and when it's lowering, gravity is helping it, so the servo is mostly acting to keep it from just dropping.

Second is that you can simplify a rigid structure to a line for the calculations if it's rigid.  If the loads are at the ends, and if the effort is along the line, then it's easy enough to so the math.  if the effort isn't along the line, then it gets more complicated.  You'd have to have a second line where the servo attached to the fulcrum.  But the math's the same, presuming the servo's effort is perpendicular to it's line.

(Putting this next quote here, out of order, because it's related...)

Anyhoo. If you're applying the effort from your servo to E, then it's lever arm around F is going to be the line EF, not A.  AND that's presuming the effort being applied perpendicularly to the lever.  (The obvious other extreme would be the servo being mounted parallel to the EF lever.  In which case it doesn't matter how hard it pushes, all it can do is break the lever, it can't make it pivot.)

I think I reached a similar conclusion (as suggested in the edit that added the last paragraph or two) by extrapolating from thinking of the Effort from the fulcrum's point of view as a torque.  If I draw a circle centered at F that touches E, that's where the servo's force is applied anywhere that it can possibly move (in other words, E is a constant distance from the Fulcrum, being connected to it, albeit in a roundabout fashion).  

The math is basically 'just' saying "how much does this mass (or effort) want this lever to rotate around the endpoint?"  Which is really easy, it's just the force times the distance.  If you have 12 points with mass and 15 efforts in all sorts of weird places on a strange rigid structure all rotating around the same point, it's still the same math.  + or - forces multiplied by the length of their lever arms.

In other words, if, in your diagram, the lever rotates around F, and your mass load is on L (It's not, there's all that light green stuff to the left of that, so the mass has even more leverage).

That's true.  I was treating it all as being at L, mostly because, like I said, if I couldn't get a handle on calculating the leverage for ONE arm in a simplified arrangement, I certainly wasn't going to be able to do it for four arms and a weight distributed in some unknown ratio along the entire length of the line AB (in my old diagram).

And not just that, the arms of the mechanism also add some of the weight that has to be lifted by the servo.  I did include this weight in my numbers, at least...

To be clear, using the old diagram, some unknown portion of the weight is bearing down at A, some at B, some is the part connecting AB, some is the part AD (the green arm), and some is the body shell and accompanying stuff.  Maybe I should've treated all the weight as being at A in the old diagram.

But, assuming that it was on L, then the load would be the mass on L times the length LF (B) - if the B was parallel to the ground.  It's not, so you'd have to figure out how much the load was perpendicular to the LF (B) line.  Or just go with it because it's close to parallel.



Here's a video I found from a quick search.  I didn't check for accuracy or whatever, but it shows what happens if a force isn't perpendicular to the lever:

 [/url][url=https://www.youtube.com/watch?v=_HuxmF_1Z90]Ch 8 - Torque - Calculating Lever Arm and Torque - YouTube

I will have a look. Thanks for the thoughts.


For now, while I'll still be able to raise the head manually, I'm making the executive decision to punt -- I'll set aside the lift servo, at least until after Dragon Con.

It's not so much because of the lever stuff, but mostly because I started playing around with how to fit the various external components in, and it looks like the only space big enough for the battery is directly between the two green arms.



That will put severe constraints on where point P can be and may end up requiring me to move to a dual servo setup, which I am unlikely to be able to do until after the con.  I will probably take some time at the con to run the issue past other builders like Matt Hobbs and David Ferreira and see what they think (assuming they are coming this year).

[kresty]

Punting sounds like a good plan

Though it is probably a bit annoying that the battery is probably where the servo'd need to go.

Browsing I found this picture...

Looking at the upper hydraulic ram that lifts the top arm of the crane, there's a short lever arm. When the arm's about 1/2 way extended, that lever arm would be perpendicular to the ram. It also doesn't get very far from that.

Of course a hydraulic ram can apply a lot of effort. A similar position for your model might be if the middle section were your lid, and the top hinge was your pivot point, then a lever arm forward of that pivot point might work. But unless it's really far from the pivot point, it's going to have poor leverage.

Another picture I stumbled upon.


Note how the piston is kind of far along the arm. That helps the leverage. Since it's hand pumped unlike the crane, that's important (though it'll require more pumping). Also notice that when the engine hoist is in the working position (parallel to the ground), the ram is nearly 90 degrees to the arm. That's actually kind of what I was thinking when first considering your problem - you really need it mounted vertically (but there's not enough room I'm pretty sure).

Anyway, you're build's looking good! Impressive pace you're setting.

[Dyne]

Punting sounds like a good plan

Though it is probably a bit annoying that the battery is probably where the servo'd need to go.

Yeah. If I changed the pivot of the green stalks to two bolts pointing outward rather than one long through-bolt, I could concievably move the battery a little further back, but the rear axle prevents it from going far enough to get out of the way of the servo.

I could also make the rear axle a pair of bolts like I had in my photo of the real chassis, but that's a bit trickier because it's also the rear body hinge, and like I said for the front one, I favor a full axle for strength.

Also notice that when the engine hoist is in the working position (parallel to the ground), the ram is nearly 90 degrees to the arm. That's actually kind of what I was thinking when first considering your problem - you really need it mounted vertically (but there's not enough room I'm pretty sure).

Yeah, space is an issue. The servo I have is 102 millimeters minimum ... very slightly more than 4 inches long. It has to have some structure both above and below the ends for mounting to allow for rotation and wire clearance.

Between ground clearance and the thickness of the shell, there is probably around 3 - 3.5 inches maximum vertical space in there, and that assumes the battery isn't in the way.

Anyway, you're build's looking good! Impressive pace you're setting.

Thank you.


It seems I spoke too soon on my printer being back to normal, however. I got those black PETG chassis prints and my Four-nines prints done in orange PLA without issue.

As soon as I put the silver petg on to run some toothed wheels for LD-F1, I once again had quality issues near the top of the print. I'm not sure if the spool is bad or something else is going on. We'll see how it goes with other spools.



Sent from my SM-A326U1 using Tapatalk

[kresty]

I think I'd aim for the servo mounted in front going towards the rear a little. So that when it raised a little it'd be perpendicular to the shell. Eg: a little less than perpendicular as it started, then perpendicular, then a little more than perpendicular, and then by the time it didn't have any leverage most of the weight would be on the frame not the servo.

That's backwards of the initial diagram though. Of course, an artist doesn't have to consider nasty things like real physics

[Dyne]

I think I'd aim for the servo mounted in front going towards the rear a little. So that when it raised a little it'd be perpendicular to the shell. Eg: a little less than perpendicular as it started, then perpendicular, then a little more than perpendicular, and then by the time it didn't have any leverage most of the weight would be on the frame not the servo.

That's backwards of the initial diagram though. Of course, an artist doesn't have to consider nasty things like real physics

As it happens, I was placing it oriented that way for awhile, though not so far forward.

I'd have to check, but I suspect that the limited stroke length becomes a big constraint on getting enough lift. It's 50 mm on my servo, and while they have servos with more, that also increases the overall length. I havent seen much in this small size that improves on a 2:1 length to stroke ratio.

I'm almost inclined to build an axial gearbox at one of the lower pivots with some ridiculous gearing. Michael Rechtin (same channel as that DIY actuator video) recently posted a stackable brushless gearbox. I'm not sure on the exact size of his, but it'd likely have to be more compact to work

Sent from my SM-A326U1 using Tapatalk

[Dyne]

T-Minus 3 weeks

Only a little bit of time left to get this thing done.  I do have a spare week's worth of wiggle room, but that's not much, and I'd prefer to have the droid assembled by then so I can do some actual testing and relax a bit.

(I did say that I'm terrible at estimating how long things will take.  It's been a full month since my deadline to have the design largely wrapped up, and I'm still working on it.)

Indeed, we're now entering a race to make sure that there's actually time to get everything printed, since many of the parts that I have left will require quite a bit of time, even assuming no failures.  I may have to train someone to keep an eye on things so I can run some prints while I'm not actually here.  I'm also considering changing to a larger nozzle purely to speed things up.  If I haven't got most of the big parts printed in the next two weekends, I'll definitely do that.

We are currently up to 96 track links out of 106 in-hand plus 16 spares and one "iffy" spare (it's not really that bad, but it's slightly deformed on the top of one peg, so I'm keeping it aside), and I believe that the remaining ten are already done.  I've got all four sprocket bodies and 6 of the 8 toothed wheels printed, though I had to switch to black PETG for the rear parts because the spool of silver I had was just causing too many problems, even after swapping hotends.

I've got a new spool of silver and the leftovers from my original spool to compare with, so we'll see how that goes when I get around to using them.


Over the last week, I have worked on finishing up the chassis design as planned.  It's not quite there, but it's close enough to get many of the parts printed.  There are a lot of parts in here, so here's a breakdown of the structure:




This black model is the Rear Bottom Plate of the droid.  Like the Front Bottom Plate, it has holes along the sides for attaching the side plates.  It also has a couple of holes in the bottom and a rectangular recess that the battery sits in.  

You will also notice the rear curved section has a large central structure and two vertical openings.  These are to support the rear body shell hinge and to allow it and the two L-shaped brackets that attach underneath the rear body shell to rotate.

I probably should've added short walls along the edges to make the part a little less flexible, but it'll be stiff enough once the side plates are installed.




The orange part is the Rear Outer Side Plate.  There's another on the opposite side, of course.  This bolts to the Rear Bottom Plate and to a part later on.  One of the things I need to wrap up on this part is to add a couple of mounting points for the track carriers and idlers, which I will be working on in the next few days.

Also shown are the Middle Support for the lower stalk axle (white) and the Battery Cover (blue) that sits over that rectangular recess in the bottom plate.  I'm using my typical 5Ah 3S LIPO in this build.  The cover is not very thick and exists more to keep the battery in place and hidden from sight than anything, though I suppose it could give the battery a tiny bit of protection from random stuff bonking it.  

I didn't get fancy with decorating the Battery Cover; I figure I can always add some greebles to it later.  It screws to the bottom plate (from below) on either side and has a latch to hold it down in the rear.  In the front is the opening for the wires and balance leads.  Fitting it and its mounting points in this space (without changing the design of parts I've already printed to allow more clearance) required a bit of fiddling around, and the tolerance is pretty tight, but I think it'll work.



The purple models are the Rear Inner Side Plates.  There is a recess for the locknut on the axle, just as there is on the Forward version of this part.  I did not put a hex recess for the lower stalk axle because I want to be able to insert and remove the axle without disassembling the chassis.  The bolts will be mostly covered by the tracks anyway, and I can always make a magnetic cover if I want to later.

The pink part is the Rear Upper Cross Brace.  It bolts to the Rear Inner Side Plates to help prevent the chassis from splaying or folding, removing some of the stress from the bottom plates (the next part will also help, as will the locknuts on the main axles).



Forward of (and on either side of) the battery area is an important part, the "Electronics Box" (yellow), named such mostly because many of the electronics go here and I didn't have a better name for it.  But it also serves another purpose: It's the main structural connection between the front chassis and rear chassis.  

Remember in a previous post when I mentioned three M5 screws that go in the horizontal slots in each Front Outer Side Plate and allow the front part to slide for possible tensioning?  Well, there are two more M5 screws just aft of those (not in slots) in the rear section.  The insets for all of these M5 screws are installed in the sides of this part.

So ... electronics.

The fuse panel sits underneath this part, just forward of the battery on the Front Bottom Plate.  There will also be some terminals for the various ground connections in here, though I might have to DIY that.  I'm planning to put the audio transducer in the area forward of the fuse panel, underneath the front axle, also attached to the front bottom plate.

The raised platform above the front axle is where the ESP32 and voltage meter are mounted.  I've test-printed this part already, making sure I can attach any jumpers that I need to the ESP32 pins, as well as connecting to USB if I need to update Penumbra.  



And above shows the cover for the ESP32 and voltage meter (Olive green), held on with magnets in each corner.  I call it the Brain Box.  It's the sloped bit that sticks up out of the opening in the body shell in the completed droid.

I would like to put the droid's main power switch next to the ESP32 and voltage meter, but it's going to be tough to fit one with a high enough current rating into the remaining space (especially making it fit under this cover vertically, though I suppose going with a rocker switch could address that).

The brain box is the sole part of the droid's lower body that is currently printed in PLA, just because I already had some loaded at the time, and because in no way is this a major part.  If I want to add some more blinkies in the rear-facing surface of this, I can either cut an opening or just reprint it



The newly added Cyan part above is the "Electronics Box Lid".  It includes the curved dip that keeps the elbow joint in the stalks from lowering any further than it should, as well as a rear stop for the upper stalk to give it some support and remove strain from the mass of the head trying to wrench backward when accelerating.

It's a separate part mostly to allow me to get at the fuse panel without disassembling the droid.  Making it a separate part also helped reduce the height of support material needed during printing.  I separated it from the main electronics box just below that curved dip, so the latter can rest directly on the print bed at its lowest point.  If I want to, I can also attach some electronics (MP3 board, for example) to the roof above the fuse box or the sidewalls or whatever.


And that's it for the chassis structure.  Of those parts, I've already printed the Rear Bottom Plate, Electronics Box Lid, and Battery Cover this weekend.  I am printing the Lower Stalk Middle Support and the Rear Upper Cross Brace right now.  That leaves me needing the Electronic Box and the four Rear Side Plates (inner/outer, left/right) before I can do an assembly of the droid's lower body.

I also need to wrap up the body shell parts.  Those parts will have to be printed standing on edge because I don't want to cope with supports on the curved outward facing surfaces.  I already added some Baddeley-style breakaway supports to the opening in the flat part of the Front Shell, and once I add something similar in the curved portion, I can get that printed.



The rear shell also needs one of those supports, to have the rear body shell hinge added (basically just two arms which hold bearings to connect it to the rear axle), and to have the mount for the upper stalk's rear pivot added.


That leaves the track carrier and idler structure as the major design elements I need to finish.  As mentioned earlier, I have to get those at least partially modeled before I can start printing the side panels of the rear chassis, because that's where they'll attach.


It doesn't stop there, though...

I still plan to use the new traction pads, and to date I have zero out of 106.  These aren't strictly required, but they'll make the tracks look much better.  They take 45 minutes each so I can fit some in spare moments here and there when I wouldn't have time for a larger print, as well as printing bulk once printer time clears up.  Most will be silver PETG (if the new spool works) while maybe 10 - 15 will likely be in grey TPU.  I need to tweak the design slightly before I start them, because the clips currently have a tendency to break off due to the layer orientation.



Beyond CAD and printing, I did take the opportunity Saturday to do some quick priming of many of the previously printed neck and head parts, though it's looking very much like this droid will not really be fully prepped, much less painted, before the convention.  

I took the upper stalks apart for this.  They are meant to be glued together permanently so the seam toward the middle doesn't bow out and open a gap, and so the gaps that are there can be filled, but I haven't done any of that yet so it's still possible.

During disassembly, I realized that this creates a design flaw:  Once the stalk halves are glued together, there's no way to remove the rotation platform that sits beneath the head servos.  That becomes a problem if it breaks and I need to swap it for a new print.

You can't remove it because there is a hex nut on the top of each threaded rod that passes through the rotation platform.  Both are recessed into it, so there's no way to tighten or loosen them except by turning the threaded rods.  Which you can't get to (especially the forward one) because they are inside the glued-together stalks.

If not for the potential bowing and gap opening, technically the upper stalks wouldn't HAVE to be glued together.  They're securely held together at the bottom by the nuts that retain the bearings in the two lower pivots, and at the top by the M3 screws inserted through the top of the rotation platform itself.  I'll have to come up with some sort of solution for the middle, be it drilling screw holes, some sort of clamp, or only limited gluing toward the middle with something I can loosen later.