Seatbelt Retractor Spiral Spring: 65Mn, Constant Torque, 200K-Cycle Verified
A seatbelt retractor spiral spring is not just a strip of steel rolled into a coil. It decides how firmly the webbing pulls back across the entire spool travel, it sits inside a safety-critical assembly that must never bind, and it has to behave identically on piece number one and piece number fifty thousand. Miss the torque curve and the driver feels it every single time they buckle up.
Project Background & Challenge
A Tier-1 automotive safety supplier arrived with a seatbelt retractor program that had stalled. Their previous spiral spring supplier could hold a torque number at one point on the curve, but torque fell away as the spring unwound, so webbing retraction weakened near full extension. The belt felt progressively slacker the further it was pulled, drivers noticed, and field complaints were climbing just as a launch gate closed in. Re-sourcing a safety-critical spring that late usually means slipping the program - unless somebody can fix the curve itself rather than just reproduce the part.
The part is a 65Mn spring-steel spiral (clock-type) spring about 30mm in outside diameter, running several turns of travel inside the retractor spool. A retractor works by trading spring torque for webbing tension, so if torque drops with deflection, retraction force drops with it. The requirement was near-constant torque across the whole working range, not a peak figure at a single angle. The old parts shed 15 to 20 percent of their torque from the wound state to the extended state, which made the belt feel stiff on the first pull and lazy near the end of its travel.
The second pressure was fatigue. A retractor spring cycles every time the belt is pulled out and released - hundreds of thousands of times across a vehicle's life, plus misuse cases where the belt is yanked hard. The customer wanted 200,000 cycles with torque loss inside a defined band and no cracking at the inner end where the spring anchors to the arbor. Both requirements had to be met at 50,000 pieces a month, on a schedule that could not move.
BQUQ Process Solution
Our answer: treat the torque curve as the deliverable, not the geometry alone, and control winding, heat treatment and stress relief so the curve stays flat and repeatable from part to part.
Torque-Curve Metrology First
We rebuilt the spring around a torque-curve target instead of a single-point specification. Strip width, thickness and the number of active turns were rebalanced so torque stays inside a tight band across the full travel. Every design iteration was wound to drawing and run on a torque tester, with the curve plotted at several angles of deflection, well before any production tooling was cut. Working from a curve instead of a number is what let us see the falling-torque behaviour and design it out, rather than discover it again in production. Two iterations moved the curve from a steep fall to a nearly flat line; the third locked the shape in.
Coiling and Heat Treatment
The spiral is formed from 65Mn spring steel, then austenitized, quenched and tempered to a hardness that balances torque retention against fatigue life. Tempering temperature is locked and logged, because it is the single biggest lever on how much torque drifts over the life of the part. After hardening, every spring is preset beyond its working range, so it takes its permanent set in the factory rather than in the field, which is exactly what keeps the first pull and the thousandth pull feeling the same.
End Forming and Burr Control
Spiral springs fail at their ends. The inner and outer hooks are formed first, then tumbled and deburred under magnification, because a burr at the inner end becomes a crack initiation site once you multiply it by 200,000 cycles. Ends are inspected one by one before the spring goes into the retractor assembly, and the anchor geometry is kept to drawing so the spring sits square in the spool and does not scrub against the housing.
Key Specifications
| Item | Specification |
|---|---|
| Material | 65Mn spring steel strip, quenched and tempered |
| Torque output | Constant-torque curve, ±5% part-to-part within a batch |
| Torque flatness | Torque variation ≤8% across the working travel |
| Fatigue life | 200,000 cycles verified, torque loss ≤5%, no cracking |
| Dimensions | φ30mm outer diameter, 0.6mm material thickness, 4 active turns |
| Surface finish | Deburred ends, black oxide finish |
| Volume | 50,000 pcs/month |
| Inspection | Torque curve testing per batch, fatigue rig sampling, CMM on critical features |
| Delivery | Samples in 5 days, first 50,000 pcs in 18 days |
Quality Control & Delivery
Every production batch is sampled and its torque curve plotted against the approved curve - if a batch drifts outside the band, it does not ship. Critical features are measured on a CMM, and the fatigue requirement is proven on a cycling rig with periodic torque checks through the run. First-article approval covers the full curve, not a single point, so there is no gap between the sample the customer approves and the parts they receive in volume. Trend data is reviewed batch to batch as an early warning against tooling wear or drift in the heat-treat furnace.
Result: the redesigned spring held torque within the target band across the full travel, passed the 200,000-cycle test with under 5 percent torque loss, and went into volume at 50,000 pieces per month, with samples in 5 days and the first batch in 18 days. Monthly rolling schedules held the launch gate, and the complaint stream on retraction force stopped. The work runs under an ISO9001:2015 quality system.
Related Products & Resources
To see more wound and formed parts like this one, browse our coil spring and torsion spring collections:
For a closer look at how we prove spring fatigue, read the case study on Automotive High-Load Compression Spring Fatigue. More manufacturing Q&A lives in our FAQ Center.
FAQ
What does "constant torque" actually mean for a retractor spring?
It means torque stays inside a narrow band from the wound state to full extension, rather than peaking and then falling. On this part we held variation to 8 percent across the working travel, so the belt feels the same whether it is barely out or almost fully extended, which is what the previous parts failed to do.
How do you verify 200,000 cycles without over-running every part?
We run a sample from each batch on a cycling rig and check torque at intervals across the full 200,000 cycles, with a pass band of 5 percent torque loss and no cracks allowed. The rig data is logged and ships with the inspection report for the customer's own records.
Why does the inner end matter so much on a spiral spring?
Because it carries the sharpest stress concentration. A small burr or a rough formed edge becomes a crack starter once it is cycled hundreds of thousands of times, and it is usually hidden inside the assembly. We form, tumble and then inspect every end under magnification before assembly, and keep the anchor geometry to drawing.
Can you hold ±5% torque and still hit 50,000 pieces a month?
Yes. Torque is controlled by locked heat-treat parameters, presetting and per-batch curve testing, not by hand-sorting good parts from bad. Steady output runs at 50,000 pcs/month, with samples in 5 days and the first 50,000 pieces in 18 days, so the curve and the volume are not a trade-off.
