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

[Dyne]

Unfortunately it's still quite possible for the drive to slip, at least when turning on carpet.





I'm not sure if it's the gears or the remaining belt. It's probably the belt, simply because I don't see an easy way for the gears to slip that badly without some serious deflection (which seems unlikely) or damage (when doesn't seem to be happening).  

Unfortunately, I'm not sure what I can do about it if it's the belts, other than try to figure out an idler setup to increase tension.  I have more bearings and M4 bolts for axles, so at least I have the hardware on hand, but it'd likely require redesign and reprinting of the side walls. Again.

If it's the gears, I'd need thicker/deeper teeth to prevent it, which would probably make the removal procedure harder.  A redesign could also be tricky to fit within our constraints, though not impossible (in fact I've already done some of it). I'm not convinced I can make it different enough to matter much, though.


I'll have to figure out a way to check which issue is -- the mechanism is hard to see with the tracks on, so I might need to get a camera on it.

[Dyne]

Here's the footage of the mechanism.  it's pretty clearly the belt that's slipping.  You can see the gears and motor are turning just fine, while the belt mostly remains stationary when trying to make the droid turn.







While watching this, I did get an idea for how to cram in a tensioner without redoing the walls.  In the slot directly below the belt you can see one of the M5 bolts currently used to secure the front chassis to the rear chassis.  

Here's a side view.  The dashed lines are the approximate path of the outer edge of the belt.  The hole highlighted in blue holds the threaded insert that the M5 bolt in question goes into.


I could assign that hole double duty, using it to attach the tensioning idler (with a standoff that aligns the idler with the belt).  

I would need additional hardware after all.  Longer M5 screws for sure (exact length TBD by the design) and potentially some new bearings as well, depending on what type of screw serves as the axle for the idler (assuming it's not M4).

More problematically, space is tight. From the angle shown above, the constraints are that the idler has to avoid touching the yellow arm, which is the inboard support for the track bogies (easy to avoid since the belt needs to deflect in the other direction), and has to avoid contact with the green intermediate gear and the sprockets on the silver drive pulley (much harder to avoid)

Here it is from the front.


The standoff would be the same height as the pink mount in the image above, and the idler would extend to roughly the same height as the small pulley on the green intermediate stage.  

From the first image, it seems like only a narrow wedge of space is available, but as you can see in the bottom view below, there's a bit more room because the silver drive pulley is inset in places.



I'd need to have the mount stick out over the sprockets to use that space, though. Basically I'm imagining an L-shaped part, with the bottom of the L facing out from the center of the droid. M5 screws go through the tall part of the L into the highlighted hole. At the tip of the short end of the L is the axle mount for the idler.

My concern with this is whether the part would be strong enough to resist loosening or breaking when the belt is under tension. Also, I don't think there'd be much height left for the axle mount and bolt after clearing the sprocket.

So I suspect I'll only be able to use the wedge after all.

An alternate idea would be to print smaller belts (57 or 56 teeth). But they are already sort of a pain to install/remove as-is, and a smaller belt would be worse. Unlike the length of the belt, tensioners can be adjusted on the fly to increase slack.

An alternate slternate idea is to replace the belt with a chain. Unsure how feasible that is.

[Paul6700]

Looking great!

[Dyne]

Looking great!

Thank you!


This is the idler I came up with.  The blue lines show the belt path with and without the idler.  
I might need to tweak it (e.g. to add some washers or extend the arm) but probably not a lot.







I streamed designing this part last night.  It's nothing too complex, I just wanted to try and actually get something productive done on stream.

[Dyne]

I installed the parts to do some testing, as you can see just below the black bogie carrier arms either side of the main body.  In the process, I discovered that they are somewhat sensitive to the diameter of the hole for the M5 bolts. Too wide and it will likely easily spin around the bolt. Too narrow and the precise threading of the bolt in the hole matters -- whether it self-taps or you tap the plastic manually, if the threads don't align with the threaded insert at the right point, the bolt won't be tight when the part is at the correct angle ... so either the angle will be wrong or it will be loose and the belt tension will make it spin in the threaded insert instead.






To be honest, while it certainly had an effect, I'm not sure how significant of an effect it is. Trying to turn the droid on carpet now sometimes causes the power draw to exceed the current limit I'd set on my power supply (a relatively low 2 amps), which was not previously happening.  That tells me the belts are staying engaged a little bit longer instead of immediately skipping teeth and letting the motor freewheel when more (mechanical) power is needed.  Consequently, the current draw to overcome the motor's stall can go higher than it was going before, hitting the max I set.**

On the other hand, it still mostly ends up skipping teeth instead of turning (or accelerating too quickly), regardless whether I'm using the actual battery or a raised max current on the bench, so it's not a huge improvement.

On the gripping hand, the carpets I'm testing on do have deeper piles than the carpet at e.g. dragon con, so it may still improve things in the real world.  Due to weather, I have yet to test it outdoors on concrete, which is another situation where it sometimes skips when turning, albeit not as often.

I could maybe continue tweaking this part to add more tension, but beyond that the logical next step I see is to switch to a chain drive for the second stage. I know little about chain drive hardware beyond that it would likely mean integrating off the shelf sprockets into my intermediate and drive roller parts. (I suspect printed sprockets wouldn't cut it, though I've been wrong before.)



** I'm not sure what the stall current on these things is supposed to be.  Specs have always been hard to find in general because iflyrc's website is terrible, and I believe the product has since been discontinued so the semi-broken page I knew about has also been removed.  I had to resort to the internet archive copy.  I'm just going to screenshot that here for posterity in case I haven't already done so somewhere else:

[Dyne]

This update is about a minor part, but one that is a major Quality of Life improvement.

One of the more troublesome aspects of working on LD-F1 is reinstalling the tracks after I've had to take them off for some reason.

The problem is not that it's intrinsically difficult, it's that you really need three hands or a second person to do it, neither of which I have handy (badum tish) when I typically have a need to take the tracks off.

The process looks like this: grab the links on the two disconnected ends of the track, pull those links together, hold them in the proper alignment so the internal holes match (which normally requires one hand on the inboard side and one on the outboard side), and while holding them there, insert the nylon filament connecting "pin" through said holes (which is where the third hand is needed).

It can be done solo (e.g. by putting the filament in the insertion hole for one section of the joint, aligning the links with two hands, then using your body to nudge the filament further in.  If you can get it to go in through another section of the joint with that method, then that is probably enough to free up the hand nearest the filament insertion point, as the filament would keep the links in alignment on that end afterward.  Then you can finish the process without need for the extra appendage.  Occasionally this workaround goes smoothly, but mostly it's just frustrating.


No more. This is now my full set of tools for attaching and detaching the tracks:


The new Track Clips are the grey pieces in the photo above.  They're very simple parts that fit around the inboard and outboard drive sprocket pins on the pair of detached links, holding them in roughly the correct position for inserting the filament.  The only trick to them is not to let the clip fall off and cause the tracks to flop around freely (especially the rear portion, since the rear sprocket can freewheel).

The nylon filament in the brown clip is just a spare of the segments used as pins. I use it to push the inserted filament pin further in to seat it after the entire length is inside the insertion hole.  The brown clip is just a part from thingiverse for attaching loose filament ends to spools -- its purpose here is merely so I can see this clear filament segment better, to avoid misplacing it. But the segment is not special - really any random piece of rigid 1.75mm filament would work just as well.

The safety pin is used when detaching the track sections. You push it into the (tiny) holes on the inboard side of the joint to shove the filament outward.  I've also used a needle for this, but the safety pin is better. The tool doesn't actually need to be sharp (ideally it wouldn't be), but it does need to have a smaller diameter than your average paperclip and still be stiff.

Speaking of paperclips, the one here is also used when detaching. At the same time as you are pushing the filament out with the safety pin, you push the paperclip down into the vertical hole on the outboard side of the link, shoving the outboard end of the filament down toward the insertion/exit hole. I could use another safety pin for this, but this operation definitely works better with a blunt-ended poking tool since the filament is round.

(In theory I could use my spare filament pin "tool" to do that, but in practice the vertical holes came out slightly too small. On the other hand, I could just seat the inserted filament pins with this paperclip instead of the spare piece of filament - it doesn't really matter as I'll tend to have both the paperclip and some spare filament pins around either way.)


The photo below shows a track clip installed on the outboard drive sprocket pins of two track links.  There was also one on the inboard end of the same pair of links.  When I took this photo, the track clips were the only thing holding these two links together; I'd already taken the filament pin out of the connection on the left side of this pair.

[Dyne]

Today is the final day of Droid Con, and I've had a great time. I met a lot of people I only knew as forum names as well as getting to hang out with friends from Dragon Con with more focus on our favorite hobby (and less distractions from that).

LD-F1 has been in attendance.  Having a soldering room at the con is very helpful; yesterday I did a minor repair on the adafruit sequin LED on the side of his head to resolder the 5v wire that broke at Dragon Con which I forgot to fix.  

He's also had the first field test of the new gears.  They didn't really improve things as much as I'd hoped ... the belts still skip (sometimes even on tile).  Also, his port-side motor gear came loose twice.  Luckily the locknut that is supposed to prevent that merely fell down into the track one of those times (and I found it on the floor afterward the other time).  I put threadlock on it afterward, so hopefully that solves the problem, but I haven't test driven him since then.

I do have some rewiring to do, and I'm starting to think that I want to redesign his head for easier access, because reconnecting the power cable to the adafruit servo board is a major pain, and I have some loose connections to the receiver, and the ESP32 boots up the head lights about one time in three for unknown reasons.

The head mechanism for Professor Huyang (as seen in the panel) has given me some notions of redesigning that as well.  Right now it's too flaky and twitchy.

I've been meaning to get back to my promise of releasing the files, but things have been busy and chaotic recently.

Such is droid life.