Bearing test rig
Designed and built an Arduino-based rig to characterise ball-bearing performance empirically
Bearing test rig spinning up
Why measure bearings?
Wheel bearing friction, as my race model proved, is a large contributor to car performance. Previously, we characterised bearings using a compressed air jig spinning up a representative wheel and timing the spin-down time. This was not an ideal process as there was high variance in initial spin speeds and manual timing accuracy, and the output was only a number used to rank bearings that did not tell us anything more about the underlying part. Bearing performance varies greatly with material, type, lubrication, seals, and break-in, and we needed a more precise way to characterise them.
The rig
I built an automated test rig around an Arduino. A motor spins each wheel up, then releases it to spin down freely while the rig records angular velocity over time; a calibrated laser diode opposing a photodiode count the wheel's passing spokes to derive rotation speed and subsequently angular acceleration. Closing the measurement loop with a rotation speed sensor allowed for precise baselined measurements, by truncating each dataset to 12,000rpm regardless of motor inconsistencies, and selecting a lower-end stopping point for timing as well.
Bearing test rig data collection run
Early laser tripwire design testing. A photodiode was required over a light dependent resistor due to the responsiveness requirements of 650+ cycles per second.
Engineering parameters
From each spin-down curve I derived a bearing friction function as a function of angular acceleration; differentiating the rpm-vs-time relationship and applying the rotational form of Newton's second law. I fed that function into the race model, integrating physical testing with simulation. I ran a structured comparison across:
Bearing burnishing results based on material
Result
Even though removing the factory lubrication resulted in temporary higher performance in the bearings, across more runs the data pointed clearly to a sealed, factory-lubricated hybrid ceramic bearing. This gave over two minutes of freespin on a representative wheel — up from under a minute, against a worst case below 30 seconds on bearings we'd used previously. Beyond R&D, I used the rig as a production tool: burnishing, grading and ranking every bearing, then grouping them into matched sets so both race cars ran balanced, high-performance wheels.