(re)building a drum spinner

Martin · 21 December 2023 13:11

Hello! Christmas is here so we have a bit odf time to write up what went well and what went badly with robots this year. With Toucan in a fairly good spot, i decided to try to build a KE beetle for my second and being me, couldn’t make it normal, or even remotely easy for myself.

I decided early on to make a WIDE drum, pop a weapon motor on each side, slap on a drive and go for it. That simple. Hah!

Some days of CAD and a call to PCBWay later i’d got this lovely thing milled - 140mm wide, 35mm Diameter 4340 Alloy steel:

The whole bot’s an experiment: i bolted some geprc 1804 motors onto spare BBB 22mm gearboxes for the drive, they are super tiny:

Next, I got a pair of 2626 motors to power the drum - they looked cool and I had no better ideas at the time.

i wanted to get the wheels relatively far forward for balance. or soemthing, so once parts arrived we could mock up a chassis in like this :

Then fit the electronics in with a bit of a squish, and go for a spinup test - in a test box of course. Always in a test box.

At this point i’d been through aload of PLA chassis test pieces, had the internals laid out and swapped over to Polymaker CoPA nylon, i’d heard it was supposedly super strong and being an experiment, decided to give it a go. External bulkheads are 10mm thick and the internals 8mm - all printed with 4 walls / 35% guroid infil.

After some fiddling with ESC’s, it span up relatively smoothly on 3S. Even getting it this far felt like a huge achievement.

More to follow later…

hhills · 21 December 2023 15:43

Is there anything at all guarding those belts? They look very vulnerable, perhaps even from your own drum!

Martin · 21 December 2023 17:54

There are a number of significant challenges coming @hhills - i don’t want to spoil them for you… But no - for this edition the belts are exposed. Did i mention the massively overspeced pair of motors? I figured that belt redundancy could be helpful but totally forgot about power draw, which over the brief life of the robot so far has caused more damage than belt loss!

First spinups didn’t really work, the drum just sat there twitching. Eventually i lucked upon a BLHELI config that worked and allowed them to spinup smoothly on 3S. (Turning stall protection down to “low” and startup power at 75%)

Switching back to LiHV 4S, the motors twitched again for ~20 seconds then the battery let out the imporant (and toxic) magic smoke.

OOPS.

More to follow later…

hhills · 21 December 2023 20:03

Weird it’s exploding. Do you think it’s belt tension being too much or just a underlying current issue made worse by more Vlotz?

Might be worth giving it a punt on a pair of cheap ‘dumb’ escs as you can get some 40-50a no names for the price of a cup of coffee. They’re pretty immune to that sort of intelligent loop death as they’re too dumb to care and will just brute force the current.

Martin · 22 December 2023 12:00

Given the trials and tribulations revealed later in this tale the mere death of a battery, while lamentable is a price to pay for science!

Right, so where was I - oh yes, experimenting. The whole bot is way outside my comfort and safety zone, so I decided to see how far it could be pushed, hoping it wouldn’t just explode / set on fire / disassemble itself first time out.

Unfortuantely this is where things get a bit unstuck. The interior bulkheads are printed 6/6 Nylon with embedded M3 heat inserts, with a slab of 6mm UHMW providing rigidity across the back and 3mm CF top / bottom panels keeping teh guts in. Origionally i’d hoped to print it in Taulman 910 but i’m already married so don’t have the spare time to babysit a roll of that stuff.

Printing the supports allowed for beter keying for the morots and a vey nice feel. Sadly there seemd to be something slightly off and the exterior bulkheads warped a bit. not enough to kill the project, but enough to write off a few sets. I got a roughly square chassis eventually, aided by CF panels supplied by the wonderful teamSC. Bot bot drove (ish) and span up 2 times out of 3, so it got enttered into December’s RoboDojo while I finalised the setup.

Also sounds/sounded AMAZING
https://imgur.com/w5nSdnm

It became rapidly obvious the drive was going to be an issue. the 1804 conversions are a whopping 2450kV making it very hard to control even with minimal turn rates. As anyone who’s tried extremely low rates knows tehre is always a minimum value before the thing even turns - the line between “not turning” and “lol 360” was far too fine. The other issue with tehse 1804’s was the shaft swaps themselves. Somehow the pressed shafts just wouldn’t grip so I had to revert to loads of 2 part epoxy to hold them in. Finally - ground clearance soemhow i’d ended up with about 2mm.

But it’s assembled, and off we go to DoJo. It’ll be fine, right ?

TBExtent · 22 December 2023 12:19

I think you might’ve put the wrong photo in for your brushless conversion, cos that’s an xing 1806 and doesn’t match the ones in your bot

Looking sweet though!

Martin · 23 December 2023 17:39

Ah yes thanks - bougth some lower speed 1806’s hoping they would work but missed the 3mm shaft in the spec, so they were useless.

So - How did RoboDojo go ?

Well, it went a little bit like this:
https://i.imgur.com/npKt6xa.gif

Arrived, got straight through safety fine and discovered that with the TPU side bumpers attached it doesn’t drive in the arena as the floor is interesting after a year of FC Feathers smashing it up. It would sort of grip, then lose grip. And still won’t spin on 4S. Besides this it was fine.

Round 1 - Icebreaker. @&*$
NGL - it was the best 30 seconds of the bot’s career. A couple of massive hits and it didn’t explode. Victory? no, but i’ll take it. https://imgur.com/knA0rYG

Rounds 2,3,4. Mean control bots who pushed me about and OOTA’d

How did the bot survive #1 though - well on first glance not very well. At the end it was beached on a floor seam due to my shoddy 2-3mm ground clearance. The weapon had stopped, one belt missing and the second had slipped sideways and cut into the Nylon frame and pulling the weapon off to refit spare belts I discovered the 8mm Mounting bolt was compeltely banana’d.

A few days later stripping the whole thing while wondering what i’m doing with my life it turned out the bulkheads were not exactly happy - I wonder what could have caused that

So what did we learn from DoJo - here’s a bit of a summary:

CoPA Nylon’s not strong enough.
Existing deadshaft supports are not good enough.
belts are too long, so chasssis flex eats them
3mm Bolts with heat inserts worked fine.
2450kV brushless drive is way too fast.
Weapon motors are ridiculously overspecced - a 1 minute fight drained a 850mah 3S lipo.
The drum hits stuff really hard. especially verts - which was kind of the point.
Ground clearance is trash.
Doesn’t drive upside down.
The drum’s too far from the floor giving weird bite angles.
In general the bot concept works. But needs many major revisions as above.

Back to the drawing board these things are totally fixable, Swapping to smaller 2208 motors lets them sit forward behind the drum, shortinging belts and neatening thigns up. A slight geometry tweak lets the drum expand to 42mm for the same weight (higher MOI maybe IDK) though slightly less tip speed. Dropping back to trused brushed drive, mounted further back to drive inverted and with a dual ESC, tweak the body a bit, upgrade to a 10mm Ti deadshaft. Lets go.

And it now spins up on 4S. Just like that @hhills with the same ESC’s but new, ‘smaller’ motors.

Merry Christmas VoV

Martin · 4 February 2024 19:10

Happy new year everyone. This wonderful little drum has just completed it’s winter refit and is fit and raring to get into an arena near you

Since we last spoke there have been a few changes. The outer chassis has changed from Printed PA6 nylon to Milled 10mm HDPE. The inner supports are still nylon, but Taulman 910 rather than the brittel PA6. Fixings have been replaced for more accuracy - using 4mm Plastite type screws straight into the Nylon/HDPE supports instead of the Heat Insert combinations. (Heat inserts drift a mm or so to either side as you insert them, which over time warps the chassis.)

I ended up getting a little benchtop pillar drill for my workshop which suddenly makes making holes super easy. Kept the CF top / Bottom plates but not really sure how much stability they add now the main frame is fixed, but we have weight and they look cool so

Drive has been fully downgraded from the 1804 conversions to trad BBB 22mm brushed, along with a nice dual ESC from the bits box. These are mounted using Pololu hubs with the grubs drilled and tapped out to M4 for more grip, keying into grooves filed into the Motor shafts. Wheels are up to 50mm from the anemic 44mm ones last time so not only do they touch the floor both ways up, but ther eis significantly more ground clearance. Cast PU 40A with a bit of red pigment.

Along with that the weapon has had a refit too. it’s a totally new drum from PCBWay - which looks to have been wire FDM’d rather than milled. It’s about the same weight as the last one but up to 42mm OD with more outer mass for a higher MOI. Weapon motors are down from 2x 2626’s to 2x 2208’s, again belt driven straight onto the drum via printed GT2 pulleys bolted onto the motors. It took a bit of fiddling to find the correct length belts but it is now fine.

Got it fully assembled. And ready for the new year <3
It still makes an amazing noise.
https://imgur.com/6gSNm2z