Team Darklight - Something Completely Different (Beetleweight Meltybrain); Retro

Despite having run Something Completely Different (SCD) at four Beetleweight events in 2018, I never did write up a build diary for it. Given it's 'retired', it makes sense to do a bit of a write up of the events it attended.

The build
SCD was built initially as a 'can I build a Meltybrain' rather than taking it particularly seriously.

The Mark 1 chassis was a criss-cross made out of 10mm HDPE, with a 5mm thin strip heated and wrapped around a mould to make the outer casing. The lids were made out of 2mm HDPE, which were initially made on a home-made CNC router as an experiment. We've not used this method since and generally outsource water cutting instead.

Attachment 8630

The Meltybrain circuit design itself is discrete, built on veroboard from an AtMega 328P plus supporting parts, ADXL 377 200G accelerometer, and a bunch load of transistors and LEDs. The ESCs are F-20A from Hobbyking, reflashed with SimonK for bidirectional use, but also with some settings tweaked as needed to get it to behave nicely in use for a Meltybrain. As the workload is essentially two weapon motors on full power for the whole match, the ESCs have to be overspecced from normal Beetleweight drive ones to stop overheating.

Initially in Mark 1, the motors were the standard 1000rpm 25mm gearboxes matched to DYS 1806 2300 brushless motors, recycled directly from our first Beetleweight, Knight of Ni (yes, we did have a Beetleweight before SCD!)

Firmware wise, it did start life using OpenMelt, an open-source Meltybrain implementation. However over the lifetime, a lot of things got changed, modified and refactored so it's kind of a frankenstein firmware bearing little resemblance to it at this point: Brushless mode didn't worked correctly at first so that was fixed, there were a few bugs relating to upper/lower speed thresholds for braking which made brushless just spin but not translate, timers fixed for newer AVR platforms, changes to support the ADXL 377 as the Accelerometer that OpenMelt specifies is no longer made, addition of direct-drive (regular 2wd non-melty mode, selectable via Aux/CH6 on the transmitter) and POV display. We've made a bunch of other changes I've forgotten about at this point as well.

I included the POV style display using blue LED lights that say '< SCD >' across the sides as it rotates as a cool factor, but this only ever worked fully once, as the LED board had a resistor / transistor fail on one of the LEDs at some point that I simply never bothered to fix during it's lifetime.

SCD Mark 1
Attachment 8620

Robots Live Burgess Hill

SCD Mark 1 debuted at Robots Live Burgess Hill, where it didn't do brilliantly, but it did actually work!

The main trouble was the motors, with the pinions continually falling off the motors and having to be re-attached between fights. These were eventually Red Loctite'd to the shafts, which lasted a bit longer, but it wasn't quite enough really.

Attachment 8631

The one any only time the Meltybrain core failed was in this event, where the controller reset into calibration mode mid-fight after getting hit. I never did find out why and this particular fault never happened again ever.

This was the first meeting of SCD and Sir lance-a-frog, which as you'll see, becomes a bit of a theme!

Video of the first fight against Snappy (Skip to 1:26)
https://youtu.be/XqjVrzc_ukk?t=86

SCD Mark 2
Attachment 8621

After Burgess Hill, I decided that the electronics were fine, but the chassis was too unbalanced to make the rotation speed improvements that I think it needed. So, I redesigned in CAD and got water cut a brand new sandwich chassis from 2x 20mm HDPE parts, along with 3mm hardox hook-style blades.

Attachment 8626

Drive was changed to direct-drive 2836 950KV brushless motors, the top speed and tracking was significantly improved via firmware tweaking, and I think this version was where it went from experiment to potential contender.

Next to Mark 1, dimensionally it's the same size, but much better balanced and solid!

You'll notice that in this photo, the Mark 1 chassis has the direct drive motors fitted. This was a configuration never actually used, as it was swapped into the Mark 2 chassis very shortly thereafter.

Attachment 8625

Bristol Bot Builders BBB3

This was the first outing of Mark 2, and as a result of the changes, it did much, much better! The direct-drive mode was still a bit wonky, and the translation speed needed upping a notch as it was quite slow, but otherwise, it worked correctly.

Though it didn't get far in the competition, it won the Gladiator by judges / audience decision, and was given the Best New Entry award!

Attachment 8627

My favourite fight of the event was against Sir lance-a-frog (spot a theme emerging here?)

https://youtu.be/N0MMdyVBNk8?t=1861

Bugglebots

SCD competed at Bugglebots after moving from being initially a reserve to a full competitor. In preparation, we revised the blades from the narrow-ended hook ones we had at BBB3, to more angular hard edged ones. We also used the leftover weight allowance to add more weight to the outer edges for heavier hits in theory. Credit goes to Sam Price for the suggestions post BBB3.
Attachment 8629

We also tweaked the firmware further for a higher top speed (as it was limited in BBB3), improved direct-drive mode (though it was still a bit wonky, just less so), and faster translation.
Attachment 8624

As Bugglebots has not been released yet as of 8th December, 2018, I can't reveal any details of how well SCD did. You'll just have to wait to see!

FRA Beetleweight Euros

In preparation for the Euros, only one further change/tweak was made, which was to change the motor mount brackets for something more substantial. Up till this point I had been using the low grade mounts provided with the motors, but after two full events, they had all bent out of shape.

Attachment 8622

SCD unfortunately did poorly at the Euros, but I did get to have a fourth and final redemption / grudge match again Sir lance-a-frog. As this was not a competition fight, we agreed not to drop the pit in this match, as this would have ended it quickly had SCD been shunted into it, the most common way up to now of beating it. Sir lance-a-frog is built really, really solidly, taking hit after hit over a full three minutes.

https://youtu.be/ECjoG7cIMi4?t=5508

As SCD was still working at the end, I entered it into the Gladiator, but this is where the journey ended. X-301 hit it really hard, which went through the chassis and mangled the lid. It did still sort of spin after the hit, and mechanically still worked, but the chassis is very broken, only held together by some tape around the layers added earlier in the event.

Attachment 8628

---

Here ends SCDs journey, but it's been fun!