Mainsheet Travellers Compared: Harken, Ronstan and Antal

This is an independent, objective comparison with no hardware partner — built on published specifications, documented construction and independent bench data, not our own hands-on test. On a one-design the traveller is fitted by the builder, so for an owner the live levers are servicing, race condition and spares. See the traveller and mainsheet system and deck hardware servicing.

On a 72m² square-top main the traveller is not carrying a clean vertical pull — it is carrying a leech load that swings off-axis as the boom moves, which is where car architectures actually diverge. A Melges 40 (11.99m, all-carbon, canting keel, 3,250kg lightship, published by the builder) loads its mainsheet hard and drops the car a decisive handful in every gust to depower without dumping leech tension. The engineering question is not "does the car roll" but "how much friction appears when the load vector tilts off the ball path." That splits cleanly into two design philosophies — single-race recirculating ball (Harken, Ronstan) versus twin-bearing ball-plus-Delrin (Antal) — and it is worth going through properly.

At a glance

The two architectures

Every serious mainsheet car here uses recirculating ball bearings — the balls run in a closed loop inside the car body so that as the car translates, balls carrying load at the front are returned to the back through an internal race rather than being shed off the end of the track. That is what lets a car roll continuously under a load that would seize a captive-cage bearing. Where the makers genuinely differ is in how many bearing systems the car carries and what each one does.

Single-race recirculating ball (Harken, Ronstan)

Harken's CB (Captive Bearing) and current CB+ cars, and Ronstan's Series cars, both put the car on twin rows of recirculating Torlon balls running in machined aluminium races. Harken's are Hardkote-anodised and Teflon-impregnated, which is the coating detail that matters: the race surface has to stay hard and low-friction after thousands of cycles of point-loaded balls, and an anodised, PTFE-loaded race resists the micro-galling that would otherwise roughen a bare aluminium groove. The balls are held captive by twin stainless-steel retaining clips (Harken) so the car can be lifted off the track for service without spilling bearings across the cockpit sole — a genuinely material advantage when you are servicing at the dock rather than on a bench.

Ball material is not incidental. Harken (and the industry) use Delrin balls for small-boat, light-load cars and Torlon for high-load cars — Delrin is a fine acetal for low ball stress but cannot carry the Hertzian contact stress of a loaded mainsheet car, so every car in the Melges 40 load class runs Torlon. One hard constraint follows from the recirculating design: ball cars require a dedicated grooved track and cannot run on plain T-track, because the balls need the special race section to recirculate. Track section is therefore a system-level decision, not a car-swap.

The weakness of a single-race car is off-axis load. The balls are optimised to carry load along one vector. When a square-top leech swings the load off the vertical, a single race relies entirely on toggle geometry to keep that load pointed near the ball path. This is why Harken mounts the sheet attachment low inside the car and quotes free rolling with mainsheet loads as much as ~40° off-angle — the geometry keeps the resultant close to the race so the balls keep rolling rather than being levered against the groove wall. It works well, but it is a geometric mitigation of a single-axis bearing.

Twin-bearing ball + Delrin (Antal)

Antal's ball cars solve the same problem with hardware instead of geometry: two separate bearing systems in one car. A lower race of Torlon balls carries the upward/vertical load, and a second groove of Delrin bearings in the upper part of the car takes the side and torsional load. In independent bench testing (Practical Sailor's midsize traveller comparison), this is precisely why the Antal car rolled with the least friction under a torsional load — the second bearing layer absorbs exactly the off-axis component that a single-race car has to fight through toggle geometry. For a boom that sweeps a wide arc under a big square-top, that is the most relevant single result in the category, and it is a real design distinction rather than a marketing line.

Antal also offers plain-bearing T-track (HS and FB extruded sections) as a separate line. Plain composite-bushing cars have far higher breakaway friction under load and belong on genoa leads and low-duty travellers, not an actively trimmed mainsheet — but the same T-track philosophy underpins Antal's structural/deck-track range. The relevant point for a mainsheet is the twin-bearing 4Race ball car, not the plain T-track.

The comparison, by axis

Breakaway friction under load

This is the number that matters at the trim position: the force to start the car moving when the sheet is loaded. Independent bench data puts it in perspective — under a 400 lb mainsheet load, none of the tested midsize cars needed more than ~12 lb to break the car away and start it rolling, i.e. the loaded car moves with well under a 3% friction coefficient. So all modern recirculating ball cars are "good enough" to move under load; the meaningful separation is off-axis (below), and how that low friction holds up as the races age and contaminate. A car that measures 12 lb breakaway new and 40 lb after a season of salt in the race has failed at the only thing that matters. That is a servicing story (race condition), not a spec-sheet story.

Off-axis / torsional capture

The Grand Prix differentiator. A clean vertical pull is the easy case; a square-top main under vang and sheet feeds a combined vertical + side/torsional load into the car, and it changes as the boom moves. Antal's twin-bearing car addresses this directly with a dedicated Delrin race for the side component and consequently measured smoothest under torsional load. Harken CB+ addresses it geometrically — low toggle mount, load vector kept near the ball path to ~40° off-angle — which is effective and proven across the one-design fleet. Ronstan Series relies on the same single-race-plus-geometry approach with a low-profile, precision-machined body. If your single priority is minimum friction with the boom swinging, Antal's architecture has the theoretical and bench edge; if you want the proven fleet-standard car with excellent serviceability, Harken's geometry is the reference.

Load rating and track section

Makers publish maximum working load (the load at which the car still functions without excessive friction, conventionally set at half the breaking load) and breaking load (structural failure when new). Concrete published figures: Harken's 27mm CB+ midrange car is rated MWL 1800 lb standard / 2300 lb high-load with a 5000 lb breaking strength; the 32mm CB Big Boat car covers 9.5–15m (to ~70ft with coupled cars). Ronstan quotes application LOA rather than a single headline load per car (Series 26 →40ft, Series 30 →60ft, Series 42 →24m/78ft, Series 55 running 12.7mm/½" balls →36m/120ft), with a Series 19 car SWL of 250 kg as a lower-end reference point.

The car rating is only half of it. The extruded aluminium track sets deflection and pull-out under load — and critically, under off-axis load the track sees a side component that tries to peel the car outward. A lighter section deflects under a heavy sheet load, and track deflection pinches the car and raises friction, so section stiffness directly affects the very friction number the whole exercise is about. In the independent test, Harken's track showed the least deflection under the 400 lb load with Antal's the greatest — a reminder that "the car rolls beautifully on a bench" and "the car rolls beautifully bolted to a deflecting section under full sheet" are different claims. On a Melges 40 the builder has specified a big-boat-class section for exactly this reason; the loads on a 72m² main in a 12m hull are keelboat loads, not sportsboat loads.

Corrosion and galvanic isolation

The traveller lives loaded, exposed and continuously washed with salt — and it mixes aluminium car/track with stainless fasteners, toggles and clips, which is a galvanic couple. Antal is the one maker that explicitly isolates the aluminium from the stainless with nylon bushings and washers, which over seasons is the difference between a car that services cleanly and one where a seized stainless fastener has corroded into an anodised alloy body. Harken relies on anodised races and stainless retaining clips; Ronstan on precision alloy with corrosion-protected fasteners. All three survive a hard racing life, but the isolation detail is a genuine longevity edge for Antal, and worth knowing before a fastener bonds itself in.

Serviceability and spares

Because on a one-design the car is already fitted, this is the axis an owner actually lives with. Harken's captive-clip design lets you pull the car and inspect or replace Torlon balls (sold as service packs) without losing bearings — a real advantage doing a race-morning check on the dock. Ronstan Series cars are serviceable but the alloy end caps are, by independent account, harder to remove. For an Australian campaign, Ronstan's local manufacture and distribution is a concrete spares-availability advantage — a replacement car or ball pack is a domestic order, not an import. Practically, race condition beats brand: the dominant in-service failure is a dry, salted or grit-contaminated race raising friction, so a fresh-water rinse regime and periodic ball/race inspection (see deck hardware servicing) matters more to the trimmer's feel than the badge on the car.

Our take

With no hardware partner, the honest engineering read: these are three genuinely capable systems that diverge on one axis that matters for a square-top main — off-axis/torsional friction — and on servicing.

• Antal has the most interesting architecture and the independent bench edge: a dedicated Delrin race for the side/torsional load on top of the Torlon vertical race, which is why it measured smoothest under torsional load, plus explicit galvanic isolation. If minimum friction with the boom swinging is your single priority, it has the strongest technical case.
• Harken CB+ is the fleet reference for good reasons that are engineering, not reputation: Hardkote-anodised, Teflon-impregnated races, captive Torlon balls you can service without spilling, a low toggle mount that keeps the load on the ball path to ~40° off-axis, the stiffest track (least deflection under load) in independent test, and published ratings (27mm 2300 lb MWL / 5000 lb BL). Its off-axis solution is geometric rather than a second bearing, but it is proven and the serviceability is best-in-class.
• Ronstan Series is a true equal on core merit — twin-row recirculating Torlon, precision-machined, a clear LOA ladder to Series 55 — with the decisive practical advantage for us of Australian manufacture and spares.

On a one-design the traveller is already specified and fitted, so the owner's real levers are race condition, servicing discipline and spares logistics — not brand selection.

Who each is best for

• Harken (CB+) — the proven fleet-standard car: best serviceability (captive balls, dock-side inspection), coated races, stiffest track, geometric off-axis capture to ~40°.
• Ronstan (Series) — technical equal with twin-row Torlon and the strongest local spares/support for an Australian campaign.
• Antal (4Race twin-bearing) — the pick when off-axis/torsional friction is the priority: a dedicated Delrin side-load race plus explicit galvanic isolation.

All three are race-grade; on a new build or refit any is defensible provided the car, track section and fastener pattern are matched to the rig loads — which, on a Melges 40, are keelboat loads in a 12m hull.

The takeaway

Mainsheet travellers stop being about "does the car move" and start being about friction when the load is off the vertical — and that is exactly what a square-top main delivers. Antal attacks it with a second (Delrin) bearing race and measures smoothest under torsional load; Harken attacks it geometrically with a low toggle mount to ~40° off-axis on Teflon-impregnated races, and pairs it with the stiffest track and the best serviceability; Ronstan matches the core ball-car engineering and adds local spares. Match the track section to the deflection and off-axis loads, keep the races clean and inspected, and on a one-design remember the car is already chosen — so the useful question is "am I servicing and setting it up well" (see the traveller and mainsheet system and deck hardware servicing). Field notes to follow.

Our pick: for a square-top Grand Prix main, Harken CB+ is the default on the strength of its Teflon-impregnated captive-ball cars, ~40° off-axis toggle geometry, stiffest track (least deflection under load) and best-in-class serviceability — with Antal the sharper choice if minimum off-axis/torsional friction is the single priority (its Torlon-plus-Delrin twin-bearing car measured smoothest under a torsional load) and Ronstan a genuine equal that wins on Australian spares and support. On a one-design the traveller is already fitted, so race condition, servicing and spares decide the outcome, not the badge.