Hammer Throw Turn Tempo: An IMU Protocol for Rhythm, Not Just Peak Speed

A college hammer thrower texts her coach a screenshot: a personal-best peak turn-3 angular velocity from the training app, right next to a throw that landed nearly two meters short of her seasonal average. The number went up. The distance went down. That contradiction is common enough in hammer circles to retire the habit of judging a throw by one peak-speed reading — but it's still the default, because it's the easiest number off a gyroscope trace.

What separates a throw that carries through the circle from one that dies on the last turn is rarely a speed the athlete failed to hit. It's whether each turn built on the last one in a compressing rhythm — wider and slower on turn 1, tighter and faster by the release turn — or one turn in the middle dragged in double support and broke that compression. This guide lays out a sacrum-mounted IMU protocol for measuring that rhythm directly: turn-by-turn duration, the double-support-to-single-support split, and a consistency score that flags the one turn quietly costing a thrower distance.

Most turn-tracking setups report one headline number per turn: peak angular velocity, or a peak hammer-head speed reconstructed from wire tension and radius. Coaches watch that number climb across a session and read the trend as proof of improving technique — reasonable, and incomplete, since peak speed captures the top of a curve that has already risen and fallen more than once by the time it's logged.

Film analysis of elite hammer technique keeps landing on the same shape: within a single turn, the hammer doesn't accelerate continuously. It slows through the brief double-support section where both feet touch down, then speeds back up through the single-support orbit that follows. Two throwers can hit an identical peak reading on turn 2 — one because the whole turn accelerated cleanly, the other from a dragging double-support phase followed by a compensatory spike. Only the first pattern survives into turn 3.

For this protocol, turn tempo is the duration of one full turn cycle — from entering single-support orbit on one turn to entering it on the next — split into its double-support and single-support components. Dapena's (1984) analysis is the mechanical reason this split matters: hammer speed characteristically dips during double support and climbs during single support, so the balance between the two, not just the turn's total length, is what a clean turn looks like.

A single IMU worn at the sacrum, centered over the belt line, captures this through 800Hz gyroscope-accelerometer fusion reading yaw angular velocity. The sacrum sits close to the body's true rotational axis, so its signal tracks trunk rotation without the noise a wrist- or hammer-mounted sensor picks up from grip shifts and wire whip. The app marks each turn boundary at the angular-velocity trough that signals double support, then times the gap to the next trough — a proxy for turn timing, but the right tool for tracking whether an athlete's own tempo is compressing turn to turn.

Mount the IMU at the sacrum with a low-profile belt, sensor face aligned so its yaw axis reads vertical when the athlete stands square in the circle — a placement that stays clear of grip, wire path, and footwork.

Calibration Sequence

1. Neutral standing hold (3 seconds): establishes the zero-reference yaw orientation before winds begin.
2. Slow 360° walk-through: one full rotation at walking pace, confirming the gyroscope tracks within a few degrees of drift.
3. Three no-hammer practice turns: footwork only, confirming the app flags each double-support trough — a green check per turn, re-strap after two consecutive misses.
4. One full-speed calibration throw: confirms detection holds at competition angular velocity, since walking-pace accuracy doesn't guarantee it.

Run the sensor through a full throwing series, not one throw — tempo consistency is a within- and across-throw pattern. A standard 8–12 throw technical session works.

Per-Throw Steps

1. Complete calibration, then throw normally. The app auto-segments each throw into turns from the yaw-rate trough pattern — no manual tagging.
2. Discard flagged throws: false starts, dropped implements, or low-confidence segmentation (typically under 8% once the strap is seated correctly).
3. Record per-turn duration and the double-support:single-support ratio for each turn.
4. Compute the turn-to-turn delta (turn 1 minus turn 2, turn 2 minus turn 3, and so on) — in sound throws this delta is positive and roughly even in size.
5. Across the session, calculate the coefficient of variation (CV%) of these deltas — low CV% means the compression pattern repeats; high CV% means some throws stall unpredictably.

Log this against a baseline built from at least 4–5 prior clean sessions — turn count and circle time vary enough between throwers that a universal cutoff is close to useless.

These bands are practical field categories from coaching video review and pilot IMU sessions on 4-turn throwers, not a certified standard — a 3-turn thrower compresses the same shape into one fewer step.

The boundaries matter less than the shape: each row should run shorter than the one above it. A thrower whose turn 3 sits inside the turn 2 band — not faster, just similar — is showing exactly the stall this protocol exists to catch, even with a respectable turn 4 peak.

The bands below combine the mechanical logic in Dapena's (1984) analysis of hammer-speed fluctuation with turn-timing patterns reported by Gutiérrez-Dávila, Soto, and Rojas (2002) from the 1999 World Championship finalists in Seville. Neither study used a sacrum IMU — Dapena worked from cine film of one elite thrower, Gutiérrez-Dávila's team from video digitization across a finalist field. Treat the thresholds as a field translation, not a direct validation of this sensor placement.

Dapena (1984) found hammer speed dips through double support and recovers through single support, tying the dip to gravity's effect as the hammer moves from its low to high point — some dip is structural, not automatically a fault; the flaw is an oversized dip against a thrower's own pattern. Gutiérrez-Dávila et al. (2002), timing turns across World Championship finalists, found throwers whose durations shortened in a consistent, step-wise pattern tended toward higher release velocities than throwers with an uneven progression, even among athletes with comparable peak velocities on any single turn.

Two limitations matter. Dapena's dip magnitude comes from one elite thrower's film — a case study, not a population average, so treat any specific percentage as directional. Gutiérrez-Dávila's sample was one World Championship field of already-elite performers, so it speaks to what separates good from very good more than whether the pattern holds for a developing thrower.

Tempo tracking earns its keep run continuously through a training cycle, not as an occasional check.

• Every technical session: full protocol, updating the personal delta baseline.
• Competition sessions: live CV% tracking so a coach can flag a stalled turn between attempts, not after reviewing film days later.
• Technical changes (turn count, grip, entry footwork): expect CV% to spike for 1–2 weeks, then re-baseline once it settles.
• Heavy-implement weeks: tighten the threshold, since fatigue widens the double-support share on a thrower's weakest turn first.

Key References

• Dapena, J. (1984). The pattern of hammer speed during a hammer throw and influences of gravity on its fluctuations. Journal of Biomechanics, 17(8), 553–559.
• Gutiérrez-Dávila, M., Soto, V. M., & Rojas, F. J. (2002). A biomechanical analysis of the individual techniques of the hammer throw finalists in the Seville Athletics World Championship 1999. Journal of Human Movement Studies, 43, 1–16.