RoboWars Australia Forum Index -> Builders Reports

Ringspinner Antweight - Revolve

Post new topic   Reply to topic
  Author    Thread

Joined: 12 May 2020
Posts: 5
Location: Wollongong NSW

 Reply with quote  
Ringspinner Antweight - Revolve

This writeup is to document my ringspinning adventures incuding failed designs, gram-pinching tactics and the tale of being too gyroscopically stable. The creatively named Revolve project is my attempt to shoot for the stars with a slingshot.

Ringspinning designs are fairly notorious for being unreliable due to their mechanical complexity. In larger weight classes they are generally built by those with a large amount of machining skill and resouces for the purpose of showing off what they can do.

The theory behind a ringspinner is that due to all the spinning mass of the ring being far out it can store more energy than a traditional axially mounted weapon with this MOI advantage. Thus if one can also be equally damaging while spinning at lower RPMs, though the same energy has to be put in. If one can design a good ringspinner chances are they can come up with a single tooth design as well. Coupled with lots of weight being devoted to the ring due to it doubling as armour it leads to a theoretically very destructive weapon.

Main design challenges are as following:

1) Aforementioned mechanical complexity. Most designs have more than a dozen independent rollers to permit free but rigid movement. Due to tolerances the ring cannot get warped in any way lest it lead to a jam. Restricting radial motion is (comparatively) easy compared to preventing the ring from leaving to the stratosphere due to failure in axial retention from a vertical spinner's impact. Most radial bearing retainers have mounting on both sides of the axle, while axial ones have only one. The ring itself is very hard to make well, requiring sub-millimeter circularity while being strong enough to take and dish hits but being able to run on bearings without annihilating them.

2) Spinouts. Due to the large mass on the outer edges of the robot if spun up too fast it will lead to a spinout due to the wheels not having the friction to keep the inner part from spinning in reaction. This is quite hard to prevent as one is almost entirely relying on static friction and really needs the ring spun up before that nasty vert reduces the robot weight by half. Once the robot has 'spun out' it is entirely up to the reactions of the driver and mercy of opponent to prevent all that stored energy sending the robot on a gyrating bounce toward an undesirable location.

3) Gyroscopic (in)stability. Being too stable can help when punted upward but prevents controlled turning. To turn too fast will fight the spinning mass as it really wants to stay put, exacerbating any minute instability in the ring, causing a wheel to lift and spinout to occur. It is almost paramount that the centre of turning is in the middle of the ring's MOI to prevent it from being swung in an arc, effectively a very sharp and fast turn. Hal Rucker built the heavyweight ringspinner Ringmaster to have a omnidirectional drivetrain (more of a translational one) I believe to assist in preventing this by not having to turn.

4) Wedges. Not much one can do about them, wedgelets cannot be realistically mounted to a traditional ringspinner nor would it have the power to utilise them. All that is left to do is to have a sharp and hard enough weapon tip to dig into the wedge or to try and hit low.

5) Power transmission. The ring will be jostled, shifted and suddenly stopped in combat so actually powering the thing becomes an interesting problem as there is no way to axially power it like a shellspinner. The two main contenders to solve this are a gear drive or friction drive. Spur gears permit some misalignment without loss of functionality and are easy enough to carve into a ring if one is capable making it in the first place for a large robot. Friction drive is more common on insect classes due to gear teeth at this scale not lasting long without some really funky manufacturing (trust me I tried).

So with the complaining out of the way, let’s get some solutions on the ground. Did I mention this is for the 150g weight class?

1) Use less parts. Instead of a industrial scale ali-express shipment of bearings we can try to use four flanged steel rollers instead to take both the axial and radial loads. I am fairly competent with the Hafco laithe in the garage and am sure I picked up some rebar from the side of the road that was only 30% rust (this is genuinely what it ended up being made of). Fusion 360 says that they should weigh about 3.5g each and will be fitted with a pair of L-630ZZ ball bearings each (6*3*2.5). Add locktite to taste. A chassis made of two machined pieces of 3mm HDPE should be adequate to take loads too. Doing the very rough and probably wrong maths indicates that M3 bolts should do just fine. By using an HDPE top and bottom to the ring it might just have enough non-friction to roll well. That'll absorb energy in all directions to prevent the 4mm Titanium ring part from breaking hopefully, it'll need all the help it can get.

2) I dunno. For weight reasons I plan to go with O-ring tyres on the two driving wheels. They'll be as far out from the centre as possible to get the best turning moment. Pre-programmed ESC ramping curves are for those who know what they are doing and have an ounce of sense. I'll just git gud at driving I guess.

3) By channeling unreasonable amounts of my spare time the ring is (according to Autodesk's good gift to the world Fusion 360) balanced within 50 microns. Let's hope the waterjet can do that good of a job. The wheels are smack on in the centre, so as long as there is little friction from the third contact point that's all I can do. Again I'll need to git gud.

4) No clue. Git gud.

5) O-ring friction drive. Compliant, cheap and can be epoxied to a motor bell. Is rubber on HDPE good enough? No clue. Chances are you reading this have dramatically more experience than I do.

Revolve Mk 1

Now that aint bad. I like the white and black colour scheme, and is well proportioned overall. The ring weighs in at 50g, should be a suitably sizeable slugger. With N20 drive (50:1?), a pair of 150mah 1S Lipos in series, a chinese D8 reciever that I was told would work with my Frsky transmitter, a turnigy multistar 7A ESC, a Turnigy D1306 brushless motor and really high spirits we were off to the races. The ring is at a 4:1 reduction from the 4000kv D1306 running 2S so it should spin at 8000rpm, a relatively respectible rotating rate.

Turning the rollers wasn't a great experince on account of the rust that had penetrated through 30mm of steel rebar. Shiny-ish parts came out at the end of the day though. My spreadsheet told me it'll make weight if I don't screw up. Bolts were ordered (M2*8, M2*18 and M3*20 with matching plain nuts)

The HDPE was machined from some beautiful creamy white stock on my FRC teams 6040 CNC rounter with a 3mm and 2mm flat end mill. I was so confident in the project that I even swapped the tools for brand new sharp ones (as far as $1 chinese bits go).

I had some ⅛" ID washer's lying around that were salvaged from converted Sunnysky X2212 motors, so used those. The ring was changed from 4mm titanium (really hard to find in small quanitites in Aus for reasonable prices) the plan was switched to 2mm Carbon steel with a 2mm 5051ish aluminium spacer.

The carbon steel was sourced from the local place known as Wollongong Sheet Metal, a location time locked in the 1940s with appropriate machinery and bakelite switchboards. Greeted by the electric wheelchair propelled receptionist (really nice guy) he deftly shifted around the office to the main building, always passing mere millimeters from obsticles. The walk through the permanently 45 degree C brick and asbestos building was hazardous enough to make any OSHA inspector faint with uncovered welding and angle grinder sparks leaking into the path. The receptionist approached one of the workers and they exchanged a quick few words, pointing to me and my more than slightly concerned mother. After fishing out what I was told is 2mm carbon steel he asked about how much I wanted. After roughly marking my 300mm strip it was carried over to an enormous hydraulic guillotine that was a mere meter from my unprotected hands (definately not the other way around). A pump whine and meaty clunk later and the steel was mine. I was charged $20 for around 300*600*2mm of (supposedly) carbon steel. As a testament to the beautiful age of the place my mother contacted her friend, a member of the historic preservation society in the city, to point it out to them. Such concludes my story of how I got a bargain on sheet steel.

The aluminium spacer was CNC machined by me, but the carbon steel was waterjet cut by the only company willing to do a walk in job the day before christmas. A trip to unanderra led to a me watch a (again really nice) bloke fight the waterjet over what it should do with Fusion 360's interperatiation of the DXF format. One $300 minimum order cost later (yowza) and I had two ring blades and a bunch of other kinds to resell and maybe make back some of that cost.

Once all the parts had arrived I assembled the chassis and ring with and ESC dangling out the side to test. Measurements were pretty spot on for my taste and it felt fairly free spinning in my hand so I gave it a shot.

No dice, barely even turned at 100rpm with a hand start.

My assessment of the situation was that my already slightly rusty rollers were not machined precisely enough (they were within approx 0.5mm of spec). Things I should have tried but didn't for some reason are:
a) Lubrication. Seems obvious but nonetheless here we are.
b) Testing the D1306 without the ring. That goddam 7A esc was my bane until the end of the project. I didn't know about programming ESCs at that point and the X2212 motors I had played with it before worked quite well with the 30A Turnigy ESCs I had on hand. Why wouldn't this work? Even with starting torque set to 150% and no load the motor would stutter for two seconds and the start to spin about 50% of the time. With hindsight I ended up switching to a 20A esc.

Armed with this knowledge and better bar stock I may return to this design for Revolve Mk 3. The HDPE chassis is a really good concept and the rollers do have promise.

At this point panic began to set in. I (a NSWlander) wanted to attend the ARC janurary meet, and only had about 2 weeks to get the robot ready. One late night on my education-liscenced-because-I-cant-afford-a-normal-liscenced copy of Fusion 360 later and a new design was born.

Revolve Mk 2 Electric Boogaloo

Oh no, I have done the forbidden deed. This uses an industrial scale ali-express shipment of bearings (16 of them, same ones as before). The chassis was printed out of ABS (all I had) which is not what I wanted but time was critical. At least a theme was decided: black chassis with white painted ring that looks like a halo when spinning, thus the robot is imaginatively named halo.

The bearings were mounted on the original pylons and on M3 bolts threaded into the print horizontally. I hope the layers are strong enough. Same overall dimensions and parts, it was made with parts I had on hand. With time running low, I set it up on the bench with a 20A esc. The ESC choice was purely because I had watched my first ever robot part (7A esc) evaporate into smoke because the multimeter had the polarity of it's probes reversed. The ring was definately a bit less friction-y but I think that unintentional ESC swap saved the project. So, yet again, I mounted the hellish contraption in the vice betweel two rolls of electrical tape and started it up.


Hand started, spinning in reverse and still no lubricant but it worked enough to give me hope.

After this it was trial and error to get the following improvements:

a) Turns out the 6040 isn't the best machine to make circles out of. There was a 1mm difference in ID where the machine's x and y axis were when cutting the HDPE rings, consistent to both pieces. Fixed with some sandpaped and time.

b) There was slack in the ring's mounting to the chassis. Turns out it needs to be accurate to about 0.1mm, so after four chassis prints later with the radial bearings shifted by 0.1mm outward each time I got a free but rigid fit.

c) O-ring pressure was important. Just some tightening and loosening of bolts.

d) Washers. The washers on the radial bearings were rubbing on the casing and so didnt permit free movement when tight. The only solution? Making my own 4*3*0.25mm washers on a lathe more designed for turning parts two orders of magnitude larger. My entire understanding of how to make parts on a lathe was inverted, leading me to question the meaning of life itself and why I chose a ringspinner as my first completed robot. The parts were reduced in radius from brass bar stock with passes, then half-parted with a parting tool, then drilled with a centre drill (happened to be 3mm) to pop off a washer that didnt need much cleaning up. One pass on a light file was all that was required to remove burrs (two passes would make it dissapear into shavings). My sanity limited me to only making the 8 required for the robot, no spares.

e) I eventually added some silicon spray lubricant to the ring to permit it to spin up with the D1306's paltry starting torque. It disappears very quickly and I have been recommended some silicon grease.

Gram Grabbing

Before we get to the results of the Mk 2, I want to reward those who have made it this far with James K's surefure way to make robots lighter by weight but heavier by wallet.

As I was very unsure of how to make a 50g ring ringspinner make weight I took all precautions necissary. I kept a spreadsheet of the weight of every nut, bolt, washer, component etc. Thus an 'if this then that' method could be taken to playing with the design.

One step I took was to use Titanium bolts for the radial supports. At about $3 a pop these are not normally considered a sensible investment but it saved 1.5g with no functional losses. Now that's priceless. I tried to find titanium M2 bolts but to no avail.

The robot was made to be as small as reasonable (even if just barely). Electronic components were laid out in CAD and dimensions formed over the most efficient way to pack them, alongside the most difficult way. DF robotics N20s were used for the small motor controller and packed with 2mm of space between them. One cannot realisticall make the wheelbase any thinner with N20s, and as the ring is a circle it is therefore the smallest realistic ringspinner one can make in the 150g weight class.

Wheels were held on with slightly undersized 3D printed collars that tightly fit onto the N20 shaft. A very simple (perhaps common) method of holding the wheel on but is very light and easy to re-assemble.

M2 bolts are an absolute pain to deal with but are plenty strong for most antweight applications. The weight difference between a M2 and M3 bolt is fairly substantial in number.

Lastly, no connectors were used in the entire robot beside the batteries, all soldered. Maintenance nightmare? Maybe, but it sure is light. Space was actually the main driving force her.

Alright back to the epic and turmulous adventure that is a high school student trying to make a tiny robot in his bedroom.

Somewhat Functional

By now my grandfather (quadcopter wiz) had helped me reprogram the ESC and finally depict that the 20A option was the best. I hadnt bothered to do weight analysis on Mk 2 so had no idea it it would make weight, but was more concerned in not wasting my school holidays without some form of success.

With the ring showing promise the rest of the robot was assembled. Three days of soldering and re-soldering with the 25W bunnings soldering iron and it was ready. V-tail mixing was set up on the transmitter (open-tx seems neat) and I got some driving done. Those two 150mah Lipos can drive for about 10 minutes straight without showing significant voltage drop, impressive. Handling isn't bad too despite the wheelbase width of 65mm. I'm used to a full sized joystick for robot control so the small stick will take some getting used too.

With the radial bolts tensioned just right and the night already fallen I sic'd the robot on a piece of neighbours decking (hardwood block). Due to the incredibly uneven sandstone floor I couldn't get the weapon up past 50% speed, but it survived quite well. Not too visible on camera but the sandstone maked really nices sparks when grated on right. Attached below is the test footage.



A few good hits and one Lipo ejection later the robot was now non-functional but had proved it's potential. It left some nice gashes in the wood and only stopped working after the Lipo left.

Before each fight I have to pre-tension the radial bolts just right to permit freespinning. Im sure I could use some form of spring washer, but that would be an extra gram or two.

And that's about it at the moment. I missed the Adelaide comp mostly due to not wanting to be stuck out of NSW and miss the start of the new semester. I weighed the robot and it turns out (sans 2 M2 nuts) it weighs in at 135g. Those gram grabbing tactics worked well, perhaps ill add 10g to the ring to make it to 40% weapon to robot ratio. I aim to get the chassis printed out of Carbon Fibre infilled nylon to further reduce weight while retaining strength. Also, ill see if anyone can show me how to harden carbon steel to make the tip a bit better at cutting.

Thanks for reading this recount, best of luck with all your future and current projects!

Maybe ill get around to that last one, we'll see.

Post Fri Jan 29, 2021 11:25 pm 
 View user's profile Send private message Send e-mail
Site Admin

Joined: 15 Jun 2004
Posts: 8511
Location: NSW

 Reply with quote  

Welcome VaSIMR200!
The bot looks really nice, looking forward to seeing it evolve and progress Cool

Post Sun Feb 07, 2021 11:28 am 
 View user's profile Send private message
  Display posts from previous:      

Forum Jump:
Jump to:  

Post new topic   Reply to topic
Page 1 of 1

Forum Rules:
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum


Last Thread | Next Thread  >
Powered by phpBB: © 2001 phpBB Group
millenniumFalcon Template By Vereor.