RoboWars Australia Forum Index -> Builders Reports

Brushless Antweight Development

Post new topic   Reply to topic
  Author    Thread

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

 Reply with quote  
Brushless Antweight Development

TLDR: after various different prototypes, i've developed a combat-proven brushless antweight drive that was able to dominate all fights Storm II style. Final solution utilised 2203 drone motors with a spur reduction to a 4WD polyurethane tyred drivetrain, running on 2S. This machine could out-push and out-speed any opponent I went against, and went 3-1 in it's first event, reachine the semifinals and winning the antweight rumble. Next improvement include preventing debris from entering the open motors and a PID loop - IMU assisted control scheme.

When I started looking at making a second antweight, I thought the combat scene was kind of like early UK Robot Wars, in that 90% of machines used the same drive systems (Bosch 750s(?) and wheelchair motors). Therefore a machine like Storm II or Tornado could come in with a significantly more capable drivetrain and dominate alot of the competition (eg. Storm II with Lynch motors). Bigger brushed motors wasn't too viable as it got heavy fast and didnt get the performance gain desired, so brushless was the best option.

Experiment 1 utilised a pair of the motors I used for driving the weapon on my ringspinner, 4000kV Turnigy D1306s, that were friction driven to the inside track sprocket of a Lego track based design. Heatshrink was shrunk over the 5mm prop adapter to act as a friction surface. The robot design was based on the three-slanted side boxes with a front hinged flipper to pin opponents using a servo, which was a honest effort to put an active weapon on the machine which ended up actually very effective. Tracks meant the robot was naturally inclined to drive straight. Running on 2S with no smart drive assist aside from power curves the drivetrain was just about controllable and very fast. While I seem to have lost the footage, it could cross the car garage lengthways in about 3 seconds. While hard to get going due to sensorless low torque motors needing to cog-start, once at speed it was very driveable. That speed was half second arena crossing though, didn't have the accelerating/pushing torque I desired due to roller slip and the heatshrink rollers wore out very quickly. Improvements needed. Link to testing video: https://youtu.be/SLM7MHvrhvk

Experiment 2 attempted to fuse a 12000kV 0802 motor with a high reduction N20 gearbox. This seemed like a logical way to fix all the previous problems: high torque, decent speed, no-slip gears, no parts to regularly replace. 0802 motors were chosen as they were spares from flywheel flipper project. Pinions were pulled (destructively pried) off of donor N20 motors. Adapters were printed out of ABS to hold the back of the motor with a bushing printed an drilled to hold the 1mm shaft. Boring the gearbox holes to 2mm meant the bolts could tap into the print to hold it on. These gearmotors were part of a electric crusher ant project which was scientifically proven to be cursed, and ended up killing swathes of electronics leading to unreliable testing.

What was determined that when all works, these geamotors were very good. Plenty of torque, top speed about double that of a 1000rpm N20s with excellent range of speed control and no slip. An advantage was that they were actually lighter than stock N20s, at 7g vs 10g. Unfortunately was lost in the pair of 7A brushless ESCs vs. a malenki nano, but if it was the same weight thats definately a win. Unfortunately these turned out to be a very unreliable. Tiny whoop drone motors got quite hot, and it was hard to keep the tiny pinion engaged with the gearbox. The M2 threads in the printed motor bracket sometimes stripped, and I had issues with the 0802 motors failing on 2S after a time (which they were technically rated for). While control was good (id say more controllable than standard brushed drive due to high gear reduction for the same groundspeed) the actual performance wasn't at the level I wanted to see. A 3S N30 drivetrain could probably match in performance, and be far simpler and especially more reliable. More performance would be hard without burning the tiny drone motors, so a new revision was required.

Experiment 3 was the current revision and what ended up competing at an event. Taking the best elements of both versions, high power motors on a tracked vehicle + spur gear reduction. The D1306 motors were too fast, but the right about of power. For rev. 3 some 1400kv 2203C motors were used. These were almost double the weight of the D1306 motors (at around 19g each), but had attractively low kV and had a workable form factor. Nylon M0.5 spur gears were used to reduce the motor to the rear track sprocket. 12:40 was the reduction, the largest I could fit given space restrictions. The robot was footprint was the maximum workable size of my print bed already (114mm square), so things were quite the squeeze to add this hardware. On 8V this calculates to a top speed of 4.5m/s or 16.2km/h (10mph). Worth noting is that the wheels all ran on a pair of 3mmID ball bearings, so quite smooth.

This version, with some ESC and control curve tuning, ended up working great. Definately required practice to drive as startup cogging ends up making it not go in the intended direction half the time initially, so the method was to act like a great white and not stop moving. Low speed control was pretty good, minimum speed about 1/3rd the top speed of a 1000rpm N20 drive (on 24mm wheels). Top speed was blistering but not too insane, and spur gear drive meant it could actually reach it within the length of an arena. It liked to drift with the lego tracks though which limited its acceleration and handling, so I invested in Barnes Flexicast 45 Polyurethane resin to cast my own tyres. Now 4WD, the 10mm wide 24mm diameter wheels made an instant difference in performance. It stuck to the ground as if it had magnets even on a polished wood floor, gone was any drifting (maybe a loss actually) and acceleration was almost instant. O-ring belts were used to transmit torque to the front wheels, tightly strung and very effective. The drivetrain could easily carry a Beetleweight to speed. If you've seen Storm II's match against the Steel Avenger in OG robot wars, it's got the power to do that type of thing to other Antweights. Very pleased, I made the following changes to the design before heading interstate to my first ever event in Melbourne (australia):
- Hooks on the servo arms. This meant the arms now functioned as a variable height/retractable catchment system to hold robots for wall or pit slamming
- Ram air duct for cooling motors, definately made a difference given the power requirements to move a 150g mass (it did nothing but looked cool)
- 0.5mm cheapo Aluminium armour plate to take the brunt of abuse off the ABS printed chassis

Competition was the first event back for MRC (Melbourne robotic combat) since early 2020, and again my first ever. Alongside my ringspinner, flywheel flipper and Beetleweight holonomic drumspinner I brought the brushless and (named Offset).
It's first fight was against a front hinged servo flipper. In this fight I was just re-getting used to driving it so had some difficulty with the startup cogging. Driving got much better as the day went on. The sectioned arena isn't quite enough to use full speed as one rarely gets an opportunity for a long run, but the potential for dominating speed is evident. Surprisingly to me, the righting arms very effectively flipped the opponent on their sides. Won by KO, video here: https://www.youtube.com/watch?v=H7A5qLI0640&ab_channel=JamesKirby

Second and third matches weren't videod, but i've got some memory of them. The second match I effectively controlled the match by out-speeding and out-pushing the opponent, eventually into the pit. It was by this point I realised I was in the non-destructive league, where the machine was designed for the destructive league but too late to fix it now. The third match was the semifinals (event was 6 hours long including setup to packup). Again overpowering the opponent, a 2WD scoop, as even when scooped the two wheels on the ground could power both machines into the wall. Lost this match due to pushing the opponent into the pit, but with a little too much power leading me to be just underneath them and technically entering first.
Rumble link: https://www.youtube.com/watch?v=CqA5ud0MnP4&ab_channel=JamesKirby

Finally the rumble, where the full beetle arena got to be used against a multitude of destructive opponents. One drive motor (left?) was locking up from debris between the rotor and stator which led to intermittent control. I spent the first half assisting general chaos while taking corner hits from the undercutter and gaining one big hole in the front from the meltybrain. The speed again was on display, causing high-speed spinner collisions by pushing other machines directly into eachother eg. at 1:20. Across the match I dispatched of a white vertical spinner using the righting arms, managed to pit both a 2wd pusher named Pushy and have a good tussle with a fellow newcomer's midcutter, to end in a pitting but not after he cut off one of the self righing arms. Offest won by being the last machine standing.
To summarise, this design ended up being quite effective as it only lost a match due to driver error. The brushless drive was a great asset, and achieved the dominating performance I desired. While calpable of handling it, it was never flipped all competition likely due to both having wide, low stance and rarely being caught. There are improvements to be had though:

- Lower/hinged front wedge or forks: in recent sparring with my grandfather's servo flipper with a hinged wedge I found that it was very susceptible to being stuck up on a wedge when caught due to low ground clearance preventing wheels from being on the ground. The chassis protects the wheels well, so I'll try to get under them first.
- Prevent shrapnel from entering motors: by the end of the rumble the large, exposed electromagnets I call motors had collected enough steel shrapnel to be one almost and another completely siezed. Need to resolve this, though it is easy to clean the motors.
- Add IMU assisted controls: this will permit the user to input desired angular velocity and translational velocity with closed loop controls. Better utilises the power available, as the human is definately the limiting factor as to what it can do. By losing the aluminium plate and lightening the unnecissary bits of the chassis I can gain the weight required.

CAD link to Offset as it competed: https://grabcad.com/library/offset-ant-mk-1-2-1

Offset Mk 1 vs Mk 1.2 drivetrain comparison:

Post Tue Jun 07, 2022 10:29 pm 
 View user's profile Send private message Send e-mail
  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.