New GPU, Prop Testing, V3 Reactor Progress
Beat by the $5 Helicopter...
My gearbox beat the small rotors but the helicopter gearbox won by a mile
This all really makes me appreciate the engineering effort that went into making those cheap helicopters fly. The helicopter gearbox also weighted the most so in reality the small black prop is probably the most efficient. The gearbox I made was just as loud and didn’t produce as much thrust so I have some work to do.
Some Initial Testing:
- Small Black Prop (4.2V) = 8.2g
- My gearbox/3d printed prop (~4V? Prop Broke) = 9g+
- Helicopter Gearbox (4.2V) = 49.6g!
For the V2 model I might just do anything necessary to get it flying. I might end up using a smaller (and noisier) prop or just modify the helicopter gearbox. I’m not giving up on the larger props yet but I want this model to fly.
I designed a gear box using helicopter gears and one that uses rubber bands for a belt drive. I had a misprint with the belt gearbox that prevented it from working properly but the gearbox I designed did work. The 9+ grams I got from it before the prop exploded surprised me. It seemed relatively efficient too but the gears were too loud. That is why I wanted to try the belt drive method
That's Production Ready for Sure
The heatsink is coming soon but I wanted to make sure its working
This V100 GPU had two missing capacitors but still worked without issue. I was suppressed because there were so many failure points. The heatsink is tied on with three layers of sewing thread (the extra two are a safety margin), It stays cool at idle but quickly heats up if I run anything on the card.
It is very fast for image generation because it has no issue loading the entire model into its 32GB of VRAM. It only took a few seconds for it to reach 80 degrees though so I’ll need a proper heatsink to do any real testing. I’ll probably be using this card for vision tasks once I get that part of the project figured out.
I Really Don't Like ABS
My Ender 3 is rock solid with PLA but I struggled to get good results with ABS
The last batch of parts got brittle and I accidently epoxied over the small metal holding groove so I had to scrap them. PLA dissolves in the sodium hydroxide solution so I needed to print the parts in ABS at home. The prints kept warping and coming off the print bed so eventually I resorted to gluing them down. This worked but then the layer adhesion started failing and the parts would split in half and curl up!
Eventually I did get a good set of parts but the tolerances were off. I reworked the CAD model and I’m printing those parts for next week.
This Week's Gallery:
Plans For Next Week:
I really want to get the circuit board and the reactor done
I’m not done testing the props and there is still more I need to learn to get that right. I originally wanted to get the reactor done first so I could know how much weight the balloon could handle. I’m very close to getting that done so that will probably be the headline for next week. I’m sure I’ll also have time to solder up a circuit board for testing too.
Todo List:
- Finish testing some props
- Rebuild the reactor
- Assemble a flexible PCB/Test with Esp32