Close

Truck prototypes

A project log for 3D printed truck

3D printed running aid

lion-mclionheadlion mclionhead 12/30/2020 at 05:280 Comments

When lions started using RC cars as personal trainers, they had no functionality.  Range was unknown.  They ran on AA's.   They were controlled with the stock 2 pawed controller.  Seem to recall running at least 5 miles until the AA's went dead, holding the controller with 2 paws, then run/walking 5 miles back while carrying the RC car.  Another range test entailed running laps closer to the apartment.

It took an eternity to develop  a power supply better than the stock AA's.  That entailed discarding the car's convenient battery holder, building a large battery tray, & mounting a lipo on top, making it top heavy.

The custom truck was a step back from that to more primitive stages.  The previous vehicles were proven toy designs.  It wasn't known if the custom motors would make enough power, randomly lose alignment & die, how much power they used, if the H bridges would overheat, how long the PLA structure would last. 


The 1st drive entailed much paralysis by analysis, unit testing, & redesigning.  The mane board got redesigned 3 times.  The no load tests showed 1/2A.  It burned 300mAh going nowhere.

The direct drive motors were a big investment.  They needed sensors for slower speed, large diameter for torque, & to be outrunners to bolt wheels on.  They don't make large outrunners with sensors.  Ordinary 42mm diameter motors were rewound with 20 turns of 26 gauge to increase torque.

Making sensored motors out of unsensored motors required very precise hall effect sensor placement.  The sensors had to be analog with sample rates above 50khz.  Digital sensors could only sample at 1khz which wasn't fast enough.  Then, the sensors needed to face a magnet holder on the motor shaft.  It was pure luck that modern motors have a shaft on 1 end for mounting magnets & a rotor on the other end for mounting a wheel.

Motor encoding was a hard problem that required plotting the sensor voltages for all shaft angles, hoping the parts didn't shift.  It was a bit harder than a standard motor sensor, since slow starting required the position in the 360 rotation to be known rather than which phase to light up.  Sensing the 360 degree angle with a custom magnet holder required only 2 hall effect sensors instead of 3.  The motor controller still has many bugs & doesn't turn them smoothly.

Then came custom hubs which could mount a tire on a motor.

A cutoff of Chinese exports & runaway inflation dried up the entire supply of cheap tires of suitable dimensions.  Tires would have to be 3D printed out of TPU.  They had to be narrow to reduce power consumption.  The cost of TPU could actually be  competitive to what Chinese tires cost.  Several mathematically generated tire models were created, with curved spokes being the cheapest.


These tires were like rock.  The hope was for the tires to be soft enough to not need shock absorbers.   TPU can be very hard, depending on the placement of arches & structural members.  The needed width of the tires remanes unknown, as it depends on how wide the suspension becomes.

Battery placement, cargo placement, electronics placement, wire routing all came together over several months.  Chinese tires were slowly replaced with 3D printed tires as each tire took 12 hours to print.

Motor control was based on the tried & true L6234 of brushless gimbal fame.  It has a thermal shutoff, but is harder to cool.  

The prototype got a container from a past vehicle.  Radios & user interface were put in a separate enclosure.

The steering section was a difficult problem.  The suspension has to be as low as possible & not flexible enough for the tires to rise up too far.  Modern vehicles are all based on ackerman steering which entails odd angles.  Ball links were 3D printed since those too can't be imported in the right size.

The PLA steering linkages broke after only .25 miles.  The tires were too noisy.  They would need longitudinal treads.  The steering section was redesigned to use TPU for the load bearing.  TPU would absorb some of the impacts & be stronger than PLA, but this design ended up too flexible.

It was just good enough to get more data, but still not very robust.

Discussions