The lessons never end. The shaft needs to be 26mm according to the design dimension, but the toothpick piece I used to test the assembly is 27.3mm. Which to use?
Conversation
I go with the toothpick. Figure the compliance of strut will create a nice compressive fit. But I forgot steel is much stiffer than wood, so it’s a tighter fit! Still it fits. Btw I’m very impressed with my hacksaw precision. I’ve matched the toothpick exactly 😎
1
1
Second axle I decided to go a little shorter. Note the scratches on the shafts and leftover stock btw… hard to wield a hacksaw precisely without it jumping around a bit. Next time I’ll make a cutting guide or something.
2
1
Passive wheels in!
1
2
Aaaannnd drive motors in as well 😎
The assembly is done! Note holes on drive motor strut fork caps… can add screws there if necessary but interference fit holds motor in place well enough. That was a bitch to get right btw.
1
3
Here’s some posing with steering angles… stylish lil guy eg?
2
4
Let’s add the computer, battery and soft assemble the partial wire harness (battery and servos connected; drive motors not yet… will add 2x6= 12 more wires/4 more cables to this picture).
1
1
2
Let’s power it up 😎
This is just led blinky running but it’s still neat. I got much further today than I expected. Now for toughest part… rest of wire harness. Unfortunately crimping and/or soldering may be needed and I suck at both.
Then the hellish challenge: programming
0:04
788 views
2
2
13
Back to the mechanical design of the main rover... just spent 2 hours figuring out how to get a mcmaster-carr ball joint and some standard nuts and bolts into my assembly, and detected some geometry issues that would have been annoying to discover after printing...
1
5
Finally actually earning the mechanical engineering degree they gave me in 1997. I think the mechanical design is now about 60% done. I'm now aiming to build a model accurate enough to simulate.
1
5
Turnbuckles and bogies need to be added, then the 4 drive units and 2 passive wheels. Another couple of days to rough it out, then 2 more to get all the finicky details doubled checked before I start printing.
Replying to
Another good day. Design is finally roughed out. Lots of the dimensions and stuff are tolerances, but now I have something I can refine.
1
2
Especially proud of learning to integrate components from other sources properly. SG-90 servo motors are now in the design (random open-source), as well as a ball joint from McMaster-Carr
1
9
I'm still astounded by how I can do all this insanely powerful engineering design work for free, and my only cost is materials and time.
2
17
The turnbuckle was the hardest thing I've had to model. 2 ball joints at either end of a variable length rod... I left out the screw constraint and the lock nuts. Made it cylindrical joints instead.
1
6
And FINALLY, the full assembly is sorta done with all major linkages and joints represented. It was a bitch to get this right... there's like 6 ball joints, and 5 pin joints here.
1
1
8
The thing is, attaching the 2 turnbuckles would immediately mangle the assembly into a weird train wreck. Since the geometry is coarse, all lengths are rough approximations, and OnShape doesn't allow ball joints to be constrained, the constraint solver creates pretzels.
1
1
And initially I had the turnbuckles set too long, so the thing would really contort. So I suppressed one end of each turnbuckle, rearranged the whole thing to look right, then manually adjusted the turnbuckle lengths and positions to be approximately right...
1
Then I unsuppressed end connections and the solver "found" the nearby solution and didn't pretzelize the design. Then I used the measuring tool to measure the turnbuckle length, and went back to the part subassembly and constrained cylindrical joint to be a few mm around that.
1
1
Effectively, instead of solving a hard 3d geometry problem to compute the right nominal length, I found it "experimentally" in the 3d CAD design. Now I have everything close enough that I can do projections and back out final actual geometry. 😎 🥳
2
1
9
Back to rover after 2 months. Finally got beaglebone blue to drive a motor. Result of a half day of work that should have been 20 minutes. 🤬
0:03
316 views
1
3
First attempt didn’t work because it turned out one of the 2 drive motors in ABTF build had terminals shorted 😭. Took a bunch of continuity testing to figure it out.
Luckily I have 2 more plus half a dozen coming. They’re not cheap. $13 each, with integrated quadrature sensor.
1
1
Or at least I think that’s what’s wrong. Will need to do some more testing to be sure.
1
Feeling the urge to add a pair of RGB leds on my rover and have the leds go red for evil. In general I want to make my rover evil. You guys are lucky I’m pretty bad at this stuff.
1
8
Servo control accomplished, kinda
1
3
Video
0:13
642 views
3
10
Next up: testing the quadrature sensor. 4 encoder wires hooked up to 4 encoder inputs with alligator clips ugh. Motor will run off dc supply for this test
1
1
Holy crap worked on first try. Im always amazed when this happens.
0:05
596 views
2
9
Though I have it counting down for forward motion. Need to reverse wires.
0:02
720 views
1
3
Back to rover after 6 weeks. I have 3 weeks to build a face-saving demo for demo day 😬 Dec 12. Printing parts for NIAM (Nature is Attempted Murder, the draft build for final build, Nature is Murder)
1
4
Urgh. This part was sized big enough to hold the servo, but not big enough to allow it to be maneuvered in. 🤬
1
1
Design for assembly is a bitch. I thought I had enough room to wiggle it in but no. Will try to saw in a compliance notch like I have on this mating connector. PLA is pretty stiff, but these notches work well.
1
1
Alt solution is to mount with flange on the outside of the box, which wastes space and increases vehicle height by about 3mm but might be better than hacking away.
1
Ah crap, this problem only affects the outer 4 wheels, and there may not be room to lower the middle wheels 3mm
Quote Tweet
And FINALLY, the full assembly is sorta done with all major linkages and joints represented. It was a bitch to get this right... there's like 6 ball joints, and 5 pin joints here.
Show this thread
2
1
But the assembly was already off anyway, since I am using 65mm wheels instead of 60, and adjusted the motor forks by 5mm, which means the outer wheels are now about 7mm higher than middle. Which is fine since the bogie is hinged, but it will be weirdly sunken middle look.
2
1
Edit, 5mm, since only half of the extra diameter affects height... so the bogie crossbar will now have about a 5-7 degree off-horizontal gradient, making the thing look a bit like a grasshopper
1
This has been an exercise in demonstrating to myself why premature optimization is the root of all evil. I was procrastinating on printing because I hadn’t worked out the exact target geometry of the design for it to be even keeled at nominal. Finally said fuckit just print.
1
11
Good thing because this draft assembly is going to so wildly off in so many ways, I’m going to have to do a significant redesign anyway. But this will be good enough to test a bunch of things
2
5
6 parts printed, 2 more printing now, 4 left. Already aware of a dozen design errors (all fixable in post with a hacksaw and sandpaper so not showstoppers for the prototype). But damn mech design is harder than they let on in school. Dunno how my professors let me pass courses
2
1
The basic mechanical frame for NIM has 26 3d printed parts, 8 bits of aluminum tubing 4 off-the-shelf complex fasteners (ball joints and turnbuckles) and probably a couple dozen ordinary fasteners. It’s about 10x more complex than anything I designed or built in mechE school
1
3
This is going to be a bitch to put together. But milestone coming up… metal-work. Need to cut that aluminum tubing into pieces with either a hacksaw or a dremel.
2
3
Show replies
