Revron (Ant) Build and Iteration Diary

[Jordan (Jordan Gell)]

Revron (Ant) Build and Iteration Diary

Preamble

In February 2023 I attended my first Antweight event, SCAR 2 in Sheffield. I entered with a fairly simple 4WD pusher called Shoveron.

(shoveron before)

I had a great time, met some great people, learned a lot and even managed to win a fight. One thing that surprised me as a newcomer is just how effective some of the spinners were and by the end of the trip home I’d convinced myself for the next event it’d be fun to build one.

(shoveron after fighting Icebreaker jr and learning what a well designed spinner can do)

A bit of research later and I set a few targets for my build

1. 4WD
2. Vertical spinner, ideally a drisk
3. Use more durable materials
4. Go positive W:L for the next event
5. Separate armour from chassis

Revron V1.0

(Revron pre SCAR 3)

For the first iteration I took a lot of design inspiration from the US beetleweight scene and robots like Lynx and Ablation (the peterbar kits).

For the electronics I used 2 x N20 BBB 1k RPM motors for direct drive at the rear, Rx and ESC’s were handled by a malenki nano and for the weapon i opted for the BE1806 and BBB brushless esc. The battery was a 350MaH 2s pack which in hindsight was overkill and a bit of a waste of weight. I wanted to start out with pretty well know reliable components so I could spend a bit more time on the mechanical design without having to worry about respecifying parts that it would depend upon.

On the mechanical side I decided against tyres and tried out a full cleat system using laser cut stainless spikes attached to printed pulleys and with 1mm thick o-rings acting as belts between the rear wheels and the fronts for 4wd. The chassis construction consisted of a polycarbonate printed pocket for the motors and electronics to be housed. 1mm CNC’s polycarbonate sheet was used for the top and bottom panels as well as the front armour. The side armour was machined from 4mm HDPE and was attached to the chassis using 6 Aluminium standoffs and aluminium bolts. One of the standoffs also doubled up as the axle for the front wheels. The drisk was made from two laser cut pieces of 304 stainless one directly mounted to the motor and the other attached through a polycarbonate spacer. Finally the motor was supported both sides by 4mm HDPE uprights and I turned down the prop mount to allow for a flanged bearing to be used on the rotating side.

SCAR 3

I managed to finish the build just before SCAR 3 and in testing a few issues had become apparent that were fixed the night before. Firstly the front wheels would sometimes bind, this was remedied by a small amount of grease on the standoffs. Secondly the cleated drive did not provide the traction expected. It turns out Ants don’t carry enough weight to make this as effective as it is for US beetles so I swapped out the rear cleats for few O-rings and hoped for the best.

SCAR 3 went well with the weapon working better than expected. Revron won all three qualifiers with a few nice roofshots and made it into the round of 16. It was knocked out in the round of 16 by a very well driven GNC. Revron took very little damage throughout the event and overall I was happy with a lot of the design choices I’d made regarding the chassis construction, armour package and weapon.

The loss to GNC showed some of the areas that needed improvement. Mainly traction, drive speed and my driving ability.

Images of early build

(First chassis test assembly)

(First drive motor tests using borrowed shoveron parts)

(Final printed prototype with all internal components before commiting to machining and laser cutting)

Thanks for reading and I’ll update this post with the series of updates I made over 2023 as I get chance,
Jordan

[Joe (Joe Brown)]

Big fan of your design style and how your ants look like the larger bots! Looking forward to seeing more of the builds. Was awesome to see you do beetles.

[Jordan (Jordan Gell)]

Cheers Joe,

I’ve got a few new Ants in the works for this year and I’m hopeful that when I work through the snags on the beetle I’ll get the reliability improved!

[Jack (Jack Franklin)]

Very cool. Agree with joe - impressively beetle-like in construction for an ant - really nice to see that construction style scaled right down

[Jordan (Jordan Gell)]

Revron update package 1

Components upgrades

Problem number one to fix was drive speed. Revron is, even for an Antweight, pretty tiny and as such runs pretty tiny wheels. 19mm OD cleats up front and for the first version 20mm o-rings on the back. This coupled with 1000rpm motors meant many other bots were much quicker and it was easily outmanoeuvred by most control bots. To fix this I had two options;

1- Fit bigger wheels and as a result scale up the bot to deal with this
2- Find motors quicker than 1000rpm

I really didn’t want to alter the size of the bot as I was pretty pleased with the form factor so I set about shopping around. Eventually I came across Turnabots spicy N10’s and bought some of the 1200rpm and the 2200rpm motors to try out. I also picked up a couple of their batteries which were slightly smaller than the turnigy nano’s i’d been using up to that point. I decided to run the 2200rpms after some brief testing.

Problem two was traction and this one proved a little bit trickier to solve. The cleats hadn’t worked as expected so I could either try and refine my cleat setup or move over to tyres. I started by testing new cleat designs, I designed and tested cleats made using arduino header pins, sewing pins, perforation wheels used for marking out fabric, sharpened versions of my original laser cut cleats and ultimately they all performed poorly. At this point I decided a rather big redesign was needed and I set about finding some suitable tyres.

I’d previously cast my own silicone tyres form shoveron however the whole process was fiddly and a little unreliable. I’m sure with more time I could have improved it but ultimately I decided there were probably better off the shelf options. After a fair bit of searching and chatting to a few people about options my uncle mentioned that he thought scalextric stuff might be about the right size. He was right and Ultimately I ended up settling on the Scalextric F1 tyres, they’re cheap, have incredibly high grip and can be easily mounted to a printed hub.

False starts and silver lining

With my new components ordered and in the post I set about redesigning Revron in CAD. This version had tyres all round, I moved the motors forward and ran a spur gear system to drive the front with belts to the rears. I also moved a standoff to in front of the front wheel to allow for modular armour packages that wrap around the standoff and slide in from the side. I also realized that with the weight saved by the new battery I should just be able to run an additional disk which seemed like it could be fun!

(Pic of V2.0 CAD)

(Assembly of one drive side for V2.0)

(Semi-assembled V2.0 alongside V1.0 and a Pnematic ant… Ignore that for now)

For those of you familiar with the bot the above CAD and pics will look pretty unfamiliar. This is because while V2.0 looked good in CAD it didn’t really work. I had some real issues getting it to handle and drive as expected. The extra gearset had a bit of backlash and the tyres were so grippy the belts on the 4wd had a tendency to slip before the bot would turn. Despite a lot of time spent tweaking and reprinting parts I just couldn’t get it to drive in a way I was happy with.

At this point I’d spent a lot of time developing but hadn’t really go much to show for it and SCAR 4 was starting to get a bit close. I decided that maybe it’d be best to only use the grippier tyres at the back and keep the cleats upfront. This would at least allow for a bit of extra pushing power but not at the expense of the speed of turning. That decided I went back to my original CAD, made some quick alterations spent some time I didn’t have messing with elastic bands and forks and built a couple of the V1.1 instead of V2.0.

At the time I was a little gutted to abandon V2.0 however much of the learning from it would be incredibly important for the beetle even if I didn’t know it yet.

(Finished CAD with fork setup)

(Weight check)

(PLA test chassis)

(Fully exploded view of final parts)

SCAR 4

SCAR 4 went very well with Revron winning 5 straight matches and proving to be an incredibly destructive bot. The new speed massively improved weapon bite which seemed to be the main contributing factor in the damage it was causing. It also “helped” Petrifeye win the most destroyed award separating the top and bottom half of the robot in pretty spectacular fashion in a great fight in the round of 8. The run ended after a pretty back and forth fight with Iceberg where their slightly longer weapon reach caught the side of my weapon motor and sent me out of the arena.

Unfortunately due to previous damage All Star had to forefit the playoff fight which gave Revron a third place finish. I also picked up a best engineered award which given all the time spent experimenting with the drive on this bot and working on its previous short comings I was really pleased with.

(Revron looking a bit battered after a good day of fights and earning some silverware)

In retrospect

While SCAR 4 had definitely been a success for the bot competition always shows areas for improvement and this event was no exception. In the box I couldn’t have asked for much more however in the pits was a different story. The bot is slow and fiddly to work on due to the sheer number of tiny parts, it also made keeping track of spares very difficult. A secondary problem was my use of directly threading into polycarbonate. After a few rounds of removing panels for battery changes the threads were starting to strip out and I realised the chassis was probably only going to be good for this event.

Sheffield roboteers events

This version of the robot also competed at a couple of the Sheffield roboteers events with mixed success. During the first event I accidentally swapped 2 motor phases during a weapon motor replacement and subsequently sent myself out of the arena multiple times before I realised the issue.

The second event I managed 7 straight wins before losing out in the final to an extremely well drive Paranoia to finish second.

For The next round of upgrades I set myself the following goals.

1-Reduce unique part count
2-Improve service life
3-Maintain performance

[Jordan (Jordan Gell)]

Revron Upgrade Package 2

The Chassis

The primary objective with the new design was to reduce part count. An obvious starting point was the chassis design and the top and bottom panels. In the previous version, each panel had been secured by 7 M2 screws self-tapped directly into the plastic. I began by integrating the top panel with the chassis, thereby reducing the part count by 8 pieces.

Above the previous chassis version

For the bottom panel attachment, I recessed the mounting height of the panel by the thickness of the bottom panel. This adjustment aimed to decrease the likelihood of the panel’s edge taking a hit. Instead of tapping directly into plastic, I increased the hole diameters to accommodate brass inserts. This modification should prevent thread degradation over time, enhance serviceability, and provide increased strength. With this more protected design, I was able to reduce the fastener count to 5 M2 screws for the bottom panel.

Above completed chassis CAD

Another design change made to improve serviceability and chassis life was the side armour mounting. Previously, M3 screws tapped directly into the plastic, which tended to degrade over time. As a solution, I employed through holes backed with hex recesses, embedding aluminium M3 nuts. This method provided much more reliable thread engagement, and even if damaged, the nut could be replaced. I also decreased the fastener count for the side armor from 5 M3 screws per side to 4, saving some weight in the process. One noteworthy feature is the keyhole slot used for the N10 motor axles, allowing for angled motor installation from the top and easier motor removal.

Above printed chassis with heatserts

With the general shape fleshed out, any surplus material was removed, resulting in a compact enclosure for all the electronics weighing in at 16.5g when printed in polycarbonate.

The Front Armour

One area that consistently failed was the front armour mounting. M2 screws often stripped out of the chassis, leading to the necessity of gluing the front armour back on during later stages of events. Since I was already using stand-offs to mount the side armour panels, it made sense to utilize these for mounting the front armour as well, reducing the part count by 8 M2 screws.

Above the stand-off setup for the side armour and front armour mounting

To achieve this, I switched from machined polycarbonate sheet to TPU printed front armour. I’d seen this work very successfully on bots like Pantherk so I was confident it’s deal well with impacts.

Above armour fitted with side panel off

Above armour on with side panel fitted

The new design allowed for easy loading of different armour configurations, including forks, and increased durability by requiring four M3 fasteners secured into aluminium to fail for the front armour to be forcibly removed, as opposed to four M2 screws tapped into plastic.

Above fork and wedglet configs

After testing some wedglets, I quickly realized how much quicker it was to swap out these modules compared to removing and replacing the polycarbonate panels fitted with fiddly M2 screws.

The Complete Package

With all the aforementioned changes and a few other minor alterations, I managed to reduce the number of fasteners used by 19 and significantly reduce the time required to swap a battery or motor. Additionally, I removed enough weight (9.6g) to allow for the use of two 3mm steel forks borrowed from Shoveron if needed. Assembling the robot proved to be a bit of a challenge, but I finished this version and tested it at 1 am the night before my first ORCS event.

ORCS

ORCS was a really fun event however it could have gone better for Revron, failing to qualify for the tournament and achieving a 2:3 W:L record. However, the new upgrade package was blameless in this; the culprit for the first two losses was a wiring mishap caused by replacing an1806 motor one with a different phase order (again), resulting in the weapon spinning backwards (again) and launching itself out of the arena twice before I realized the mistake (note to self: avoid wiring motors past midnight before an event).

Although disappointed by the outcome, ORCS had a significant silver lining. The tournament finished ahead of schedule for the bot, allowing ample time for rumbles. This provided an opportunity to durability test the new design against at least five other spinners, including Delta V and Iceberg, repeatedly. The design coped admirably with this abuse, sustaining no major damage despite facing heavy hitters. The chassis remained intact and usable for another event, and the TPU front armor absorbed hits effectively. The brass inserts worked seamlessly, and swapping batteries became effortless.

The Rest of 2023

For the remainder of 2023, Revron underwent no major revisions, only minor part tweaks. The design proved to be quite effective, and I ended the year with a W:L record of 31:10 on rampage, having participated in approximately 60-70 total fights, including whiteboards, rumbles, robonerd, and other non-rampage events.

The chassis endured much better, with one lasting the majority of the year before retirement due to accumulated damage, and the second one still in use into 2024. Revron also achieved an event win at the Sheffield Roboteers End of Year Fight Night, going undefeated in eight straight fights.

Above Revron at Sheffield roboteers rocking some very non-cube acetate

What’s Next?

Ultimately, I am now quite satisfied with the design as it stood at the end of 2023. For 2024, I’ve decided to branch the design into two variants: one with only minor changes from the 2023 spec, featuring cleats and the traditional HDPE machined side panels. This will retain the Revron name. The alternate version is a more printer-friendly version made with more TPU and with tires instead of cleats, which I’ll give a new name and will likely have a different driver.

Above printed variant with skirt

I’ve been wanting to open-source the CAD for this robot for some time; however, the machined parts make it somewhat inaccessible to build. Therefore, the goal is to produce a version that doesn’t require a CNC mill or lathe. Hopefully, this new design will make its debut at AWS70, and its performance will be assessed. If it proves successful, I’ll share the CAD and BOM along with a proper build guide. I’d love to see how the design fares in the US fairy scene or the Aus Ant scene so if anyone is interested in building one let me know!

[Jordan (Jordan Gell)]

2024 Forking designs

With the printable design complete in CAD I set about getting one put together ready for AWS 70 as well as rolling out the new fork design to the older version as they’d worked quite well in testing.

Revron Spung fork test

(Above prototype sprung forks)

I got a batch of forks laser cut some from 304 Stainless and some from Aluminium. The printable design features enough weight saving to run the steel forks however the older version required aluminium forks to get in under 150g. The forks Pivot around the wedgelet mounting standoff and are sprung using a small piece of elastic passed through a small hole at the back of the fork.

(Above laser cut forks)

The chassis features a few notable changes including the removal of one pair of stand-offs for weight saving and the implementation of a quick change weapon system that would allow for modular weapons in the future. The quick change system uses two square profile section of print that the weapon uprights slide over before being locked in place with four M3 bolts. I also put the motor on bullet connectors so weapon setups could be swapped in a couple of minutes maximum.

(Above cutaway view of quick change system)

The chassis was printed from PLA-ST as I hadn’t had time to source polycarbonate prints and in terms of durability I’ve seen little difference between the two at ant scale. It also made the design more representative of what most builders would have access too as polycarbonate is not an easy material to print with.

(Above ride height check of the new chassis, the 11mm tyres look so tiny)

The side skirts were printed from a fetching pink TPU and the electronics and assembly went pretty smoothly. The one differences worth noting is the 2200RPM Turnabot motors I’d been using for a while had gone out of stock for the foreseeable future so the new build got 3000RPM instead. The speed upgrade wasn’t really needed but it did make it a lot of fun to drive. Secondly the new design was running a Ranglebox prototype battery to save a little weight and to test it out in competition.

The bot was ready for its new driver and name. Gage would be taking the reigns of the new design and he christened the Pink vert Wev-won fitting in well with the name of his control bot UwU.

With Wev-won complete surprisingly early before AWS 70 there was time for some testing and by luck the Sheffield roboteers society had an event the week before which we popped across to.

Wev-won picked up second in that event getting thrown out of the arena in the finals after a pretty close back and forth with a bot called Fusion (a modified version of the broken link robotics derive) and it caused some respectable damage along the way. One lesson learned from this event is the ability to change out the forks quickly is vital for horizontal spinners so I made some tweaks to the skirts separating the fork mount from the rest of the skirt to improve serviceability.

(Above old and new side by side ready for AWS)

Wev-won (Revron V2.0) weapon, drive and fork test

(Above Wev-won first full systems test)

AWS 70

AWS 70 was a great day out and a really fun event. Both bots performed well however the day was not stress free with Revron binding a drive side in its first match, burning out one of the precious few 2200 RPM motors I have left and also damaging the malenki in such a way that ESC braking stopped functioning. That left it fighting though the losers bracket for the majority of the day and also made control a bit less precise than normal. Revron got in some great fights before eventually going out to Relativity in the the top 6.

Wev-won also had a pretty brilliant run with some great fights, it didn’t quite make the final stage bracket loosing out to Ablation but overall we were pleased. The design proved to be reliable and the forks worked very well in most fights.

The bots didn’t come home without silverware with a best design award and a best fight award for Revron vs Stewie.

(Above Revron back home after a long fun day with a few additions to the shelf)