Electric go-kart v2

Two ClearPath servos and 10:1 planetary gearboxes, one per rear wheel — and the RMS current limit that ended it

Version 1 got the kart moving on electricity, and then taught me what happens when you put eighty lithium cells in series with no BMS. This is the rebuild: lower voltage, no chain, and a motor per wheel.

The Thunder Kart in its original state, a rusty steel frame with a white vinyl seat, steering wheel and four knobby tyres, parked on a garage floor — The donor, back to bare rolling chassis

Measuring what was already there

The plan was to build a complete rear drive as one assembly and put it on the existing frame, so the first job was finding out exactly how much room there was and what the wheels expected to be bolted to.

The go-kart chassis upside down on jack stands with the rear axle, sprocket and brake band still fitted — Everything comes off before anything goes on

The same chassis with dimensions written on the photo in marker: 19.5 inches, 16.5 inches, 10 3/4 inches and 0.190 — Marker on a photo is an underrated CAD package

Close view of the rear frame rail showing the band brake, threaded axle stub and yellow throttle cable — The band brake and cable that were about to become irrelevant

The bare rear section of the frame on jack stands with the axle removed — Rear frame, cleared out

A greasy 60-tooth sprocket with a wheel hub bolted through it, beside a second bare wheel hub, on a workbench — The old drive: one sprocket, one live axle, both wheels

A white painted four-bolt wheel hub with a bearing pressed into its centre, lying on a workbench — The wheel hub is the one part with a fixed interface — everything else was negotiable

The same hub with measurements written on the photo: 0.615 inch bore, 1.375 inch bolt circle and 2.5 inches across — Bore, bolt circle, overall

The hub photographed edge-on and held in a hand, annotated 0.57 and 1.873 — And the depth, which is what sets how far the shaft has to stick out

The hub bolted to the sprocket and held edge-on, annotated 3 3/8 inches — The stack height I no longer needed to make room for

Two motors, no differential

Each rear wheel gets its own motor: a Teknic ClearPath CPM-MCVC-3436D-RLN. These are integrated servos — motor, encoder and drive in one housing, running off DC, with no separate amplifier to mount or wire. The MCVC variant is the velocity-control one: it takes a speed command on its inputs and closes the loop around it itself. For a go-kart that is exactly the right division of labour — I did not want to write a motion program, I wanted something that means “go”.

That also means there is no live axle and no differential. Each wheel sits on its own stub shaft with its own bearings and its own brake disc, driven inline by its own motor.

The gearbox

Between each motor and its wheel is a SureGear PGCN34-1050 planetary gearbox — 10:1, with the NEMA 34 input flange the ClearPath bolts straight onto. A ClearPath makes its power at speed rather than at torque, and a wheel wants the opposite, so the reduction is what makes the two compatible: ten times the torque at the wheel, a tenth of the speed, and the motor spending its time in the part of its curve where it is actually good.

The planetary layout is the other reason it works here. Input and output are concentric, so motor, gearbox, coupling, bearing and wheel all sit on one straight line — which is why the finished drive is a bar rather than a bracket full of offsets.

A ClearPath motor bolted to a gearbox and a machined mounting bracket, on a workbench — Motor, gearbox and mount as one rigid unit

The same assembly from the output end, showing the gearbox output shaft in an orange jaw coupling — Jaw coupling on the gearbox output

A complete drive assembly laid out on a workbench: motor, gearbox, coupling, shaft in two pillow blocks, brake disc and wheel hub — One complete drive, end to end

Two identical drive assemblies side by side on a workbench — And then the same thing again

The subframe

Both drives went onto a subframe built up as one unit on the bench, then welded to the back of the kart frame where the axle used to be. Doing it that way meant the whole rear end could be aligned and squared at working height instead of upside down on jack stands.

Both drive assemblies mounted facing outward on a rectangular steel frame, on a workbench under a brick wall — Both motors mounted back to back, facing outward

A square-tube subframe clamped in a bench vise with one drive assembly fitted at the far end — Building it up on the bench

The subframe on a workbench with both motors and gearboxes mounted in the middle — Motors in the middle, output to each side

The subframe clamped in a vise with both drives, shafts, couplings and brake discs installed — Shafts, bearings and discs on

The subframe with both wheels fitted, clamped in a bench vise — Wheels on, and it is a rear end

The go-kart on grass with the subframe attached to the back of the frame in place of the original axle — Welded on where the axle used to be

Wiring and control

The high-voltage lesson from v1 stuck: this one runs on a 48 V LiFePO4 pack, with a 15 A fuse in each motor’s supply line.

Control goes through an Arduino Uno rather than straight from the throttle to the motors. A linear potentiometer on the throttle produces 0–5 V, the Arduino reads it, and the Arduino generates the PWM the ClearPaths take as their speed command. A switch picks forward or reverse. Each motor’s eight-pin connector carries the two input lines, an enable pair, and an HLFB pair coming back out — the motor’s own feedback, so the kart can tell me when a drive has faulted rather than just going quiet.

Putting a microcontroller in the middle is what makes the two-motor layout worth having. The wheels do not have to get the same number: with each motor commanded separately and HLFB coming back from both, the same box that reads the throttle can also do traction control.

Wiring schematic showing two ClearPath motors with eight-pin control connectors, a 0-5V throttle potentiometer, a forward/reverse switch, breakout boards, 15 A fuses and a 48 V LiFePO4 battery — Power and signal runs. The Arduino sitting between the throttle and the motors is not drawn here

A grey sealed enclosure open on a bench, containing a small breakout board on a mounting plate with red and blue battery leads and an inline fuse — Breakout board and fusing in a sealed box, because this thing lives outside

The subframe clamped in a vise with both wheels off the ground and power cables run to a wall disconnect — Wheels off the ground for the first power-up

The go-kart parked on grass with a laptop propped on the rear subframe, connected to the motors — Tuning in the yard

Where it stopped

On the bench, the drives were excellent. Smooth, quiet, instantly responsive, and far better behaved than anything I had run in v1. With the wheels off the ground, or with the subframe rolling under no real load, it did everything I asked.

Then I put the kart on the ground and tried to accelerate it with me in it.

A ClearPath protects its windings with an RMS current model — the drive tracks current over time and shuts down when that running average reaches 100% of what the motor can carry continuously. Peak torque was never the problem; the motors would pull hard from a stop. But getting a loaded go-kart up to speed takes that torque for long enough that the RMS figure climbs to 100% before the kart is moving properly, and the motor shuts down. Every time.

I reached out to Teknic about it and came away without a way around it. As far as I could establish there is no setting that raises or defeats the limit — it is winding protection, not a tuning parameter, and a motor that let you turn it off would be a motor that lets you burn it up.

Which is a fair design decision on their part and a dead end on mine. The mechanical side of this build is sound and I would do it the same way again; the drives just need to be something that can hold that torque long enough to get the thing rolling.