Track tensioner

The brief

Rocket cars race along a taut wire track. The task was to design the tensioner, not the car: a portable, hand-operated tool that can hold, apply and release tension on the wire so races run quickly and safely, without modifying the cars or the wire loops. It had to work outdoors, on sites such as tennis courts, and pack into a 25-litre rucksack.

The work came in two submissions: an initial portfolio in November 2024 (need, requirements, concepts and sizing) and a final portfolio in January 2025 (detail design and drawings).

Requirements

I turned the need into ten requirements:

Concepts and selection

I sketched three mechanisms:

• A: ratchet and pawl with a fixed anchor. A handle winds the wire onto a toothed wheel and a spring-loaded pawl locks it one tooth at a time.
• B: pulleys with a hanging load that sets the tension by gravity.
• C: a spring-loaded tensioner, similar to a turnbuckle.

(opens full size)

I scored them against ten weighted criteria whose weights sum to 1.0. Manual operation, portability, safety and weight carried the most weight, at 0.12 each.

Sizing

Tension

F = mg = 2 kg × 9.81 m/s² = 19.6 N, rounded up to 20 N.

Set-up time

Assuming each stroke adds about 1 N, 20 strokes at about 2 s each take 40 s, inside the 2-minute limit.

Grip and torque

A 12 cm grip circumference gives a radius of about 2 cm. Even if the full 20 N acted at a 5 cm lever, the torque would be only 1 N·m, which is comfortable by hand.

(opens full size)

Detail design and drawings

The final design has six parts: a base, a ratchet wheel, a handle, a shaft and a pawl, all in ABS-PC, and a bought-in spring. I modelled the parts and assembly in SolidWorks and produced four sheets in third-angle projection: an assembly drawing, a part drawing with section C-C and thickness limits, a bill of materials with a ballooned isometric, and a 1:1 hand drawing.

I bought in the spring rather than designing it: a bought spring is made to a known specification, and buying one costs less than the time to design and prototype it. I chose ABS-PC for the made parts because the blend is tough, heat-resistant, easy to mould and drill, and gives a good surface finish.

(opens full size)

(opens full size)

(opens full size)

Looking back

The process held up. The requirements were specific and measurable, and the concept choice is robust: with all ten criteria weighted equally the ratchet still wins, 0.96 against 0.81 and 0.80. The drawings follow the conventions, with a section view, a ballooned BOM and title blocks. But reviewing the work now, the analysis was thinner than the drawings suggest.

• The stroke count was assumed, not derived. In a ratchet winder, the tension gained per click depends on how much wire each tooth winds in and on the wire’s stiffness and slack. With 26 teeth on a 13 mm hub, each click winds in about 1.6 mm of wire. On a 1 mm steel wire 20 m long, that would add about 12 N per click once the slack is out. So the real problem is fine control near 20 N, not the number of strokes.
• I added torque across strokes, which isn’t meaningful. What matters is the holding torque (tension × drum radius) and the force at the hand (torque ÷ lever length). If the wire wound onto the 13 mm hub, holding 20 N would need only about 0.13 N·m.
• The materials changed without a check. My initial scoring chose aluminium for the handle and carbon steel for the ratchet and pawl. The final design uses ABS-PC for every made part, including the teeth, the 5 mm shaft and the pawl, with no strength, wear or creep check against the 50-use and no-slip requirements. The scoring was also shaky: copper scored full marks for “lightweight”, but it is denser than steel. And copper makes a poor spring because of its low elastic limit; spring steel or phosphor bronze would be the normal choice.
• The drawings don’t fully define the mechanism. The wheel has symmetric, gear-like teeth rather than saw teeth with a locking face. There’s no wire drum or clamp, no pawl release and no ground anchor, although the tool had to hold, apply and release the wire 2 cm above the ground. The final handle is a flat bar shorter than the 5 cm grip I’d specified, and the pawl tip is drawn with a zero radius, against my own “no sharp edges” requirement.
• The tolerances don’t suit the process. Limits of ±0.02 mm on 5–10 mm thicknesses are machining-grade, far tighter than moulded ABS-PC can hold, and the three CAD sheets don’t agree with each other. The features that control engagement, such as the positions of the shaft and pawl-pivot holes, have no geometric tolerances.
• Nothing was verified. I never worked out mass or cost from the CAD or tested a prototype, so my conclusion that the design met every requirement wasn’t backed by evidence. A rough estimate puts the parts at about 0.1 kg, so the 3.5 kg limit would have been easy, but I should have shown it.

What I’d do differently

• Specify the wire first (material, diameter and length), then size the drum and tooth count from the tension gained per click.
• Use asymmetric ratchet teeth, and add a pawl-release lever, a wire drum with a clamp, and a ground anchor that sets the 2 cm wire height.
• Check the teeth, shaft and pivot pin with a safety factor, by hand or in SolidWorks Simulation. Make the pawl and shaft in metal or glass-filled nylon, and specify a steel spring with a stated rate.
• Tolerance each part for how it will be made, and add geometric tolerances to the hole positions, with a detail drawing for each part.
• End with a requirement-by-requirement verification table: mass and cost from the CAD, and a prototype test of set-up time and holding.