60 Seconds of Glory ended with two wins, and two losses, similar to it's AVC performance. The bot has a ways to go before it's really decently competitive, and failures continued to plague us at Moto '18.
For Motorama, we'd spent time improving securement and placement of the electronics. Previously at AVC, we'd resorted to zip ties, straps, and electrical tape to keep things attached to the frame. We added attachment points to the magnet housing to keep wires from moving too much, and reorganized our electronics to neatly fit in the one fully enclosed compartment in the frame.
We also zip-tied and taped all bullet connectors and xt60 connectors. Our most common failure in previous fights and testing were connectors popping loose. It's easy to underestimate the g-forces that the occur within a full body spinner. Not only did we see disconnects during big hits, but if we started to coin or vibrate violently, things would either bust free or unplug themselves.
These changes seemed to be for the better, with no connectors popping free during the matches.
Our first loss came at the hands of Megatron. Jamo was easily controlling us, and got a few big chops in. Our weapon stopped working after a few hits, and we tapped out. The drive was still fine, but the VESC controlling the weapon was flashing out the dreaded "DRVR" error. At first we thought that we'd fallen victim to the VESC gremlins that you hear about, even though we were only running at 8s and very low eRPM. Our hope had been that we wouldn't see driver chip failures due to not pushing the limits of the VESC.
Luckily (I guess), upon further investigation, it looks like one of Jamo's hits had physically popped a FET off of the VESC board. We've had a history of bad solder joints on the DIYSkateboard sourced VESCs, and this looks like it's possibly another case of a bad joint.
With not enough time to reconfigure our backup VESC to fit in the electronics bay, we reverted to strapping the VESC to the frame. We used a lot more tape, and had more options for zip ties, and this variation won a fight.
Our last fight was against Boom Boom, a multibot by the wacky guys at Bots FC. The fight was going well, until one of our drive sides stopped working. The other side stopped responding shortly afterward. We were able to use shell vibration to briefly show translational movement, but that too died, giving Bots FC the win.
Our first side seems to have split one of the MarkForged printed pulleys. We were cutting it close in terms of how much meat was left between the key way and pulley, and this finally caught up to us. There was also some interesting deformation on the teeth of another pulley, and we had keyways that seemed deformed by heat.
With only one side functioning, it appears we asked too much of the 50 amp Spider ESC on the other side, which eventually burst a few FETs.
The second VESC died a violent death. One of the retaining zip ties had popped off, which gave the VESC enough wiggle room to physically pull a motor lead off of the board, solder pad and all. Once that happened, several other components smoked on the board.
Another interesting thing we learned this competition was how oversized our batteries are. Originally we though we were cutting it close with 2200 mAh on the weapon battery, and 1300 mAh on drive. We never used more than 300 for the weapon, and 200 for drive. The plan is pick up smaller batteries that can still source appropriate current, and use the weight savings for bigger ESCs, and better mounting options.
In terms of the pulley failure, we'll at the very least add .05" to the width, which should make them a tiny bit stronger. We may also look at swapping to aluminum pulleys, if weight allows. I'd rather stick with printed if possible, due to the easy of customizing them, and the weight savings. We'll do a full tear down and see if any of the other pulleys show any hints of failure.
<Pictures to follow once I'm by a real computer>
Pete - Team Cosmos
Show Posts
