Tethered velocity-based training systems were the first tools to make bar speed measurable outside a motion-capture lab, and for a strict vertical pull, they're genuinely accurate. That's the problem. The accuracy is conditional on an assumption that almost never holds once you leave the smith machine: that the load travels along a single, fixed axis.
I want to walk through why that assumption breaks, because it isn't a minor edge case. It's the majority of real training.
What a tethered system is actually measuring
A cable-based sensor sits on the floor, tethered to the bar, and measures how fast the cable is paid out. That's a proxy for vertical displacement, not a direct measurement of the bar's motion. The system implicitly assumes the cable stays parallel to the true direction of travel for the entire rep.
On a strict vertical pull, that assumption is close enough to true that the error is negligible. The cable and the bar's path are effectively coincident, so cable length rate tracks bar velocity well.
Where the assumption fails
The moment the bar's path curves, even slightly, the cable is no longer measuring the true trajectory. It's measuring the projection of that trajectory onto a straight line from the anchor point to the bar. Any component of motion that isn't aligned with the cable simply disappears from the measurement.
A few concrete cases where this shows up:
- Bench press: the bar doesn't travel in a perfectly vertical line, it arcs slightly toward the chest and back out. A tethered system underreports the true path length and distorts the velocity curve at the top and bottom of the rep.
- Squat and deadlift variations: any lateral drift, and there's always some, gets silently discarded. The reported velocity is systematically lower than the true velocity, and the error isn't constant, so it can't be corrected with a simple calibration factor.
- Anything that isn't a straight pull: cable work, sled pushes, rotational or diagonal patterns. These aren't edge cases for an athlete's program, they're a normal part of it, and a single-axis cable simply can't represent them.
Why this matters more than it looks like it should
The failure mode here is worse than random noise. It's a systematic bias that depends on how much the bar path deviates from the cable's fixed line, which changes rep to rep, athlete to athlete, and exercise to exercise. That means:
- Two athletes doing the "same" lift with slightly different bar paths get inconsistent velocity readings, even though the load and effort are comparable.
- The same athlete's velocity numbers can drift across a training block, not because their performance changed, but because their bar path changed slightly as fatigue set in or as load increased.
- Any load-velocity profile built from that data inherits the distortion. Estimated 1RM, velocity-loss thresholds, and autoregulated load adjustments are all downstream of a number that was quietly biased before it ever reached the app.
None of this is visible to the coach. The system doesn't throw an error when the bar path drifts. It just reports a number, and the number is wrong by an amount you can't see and can't easily correct for after the fact.
Why I didn't build ORCA around a tether
Once I understood this failure mode, a tethered architecture was off the table for me. It solves the strict-vertical case well and fails, silently, everywhere else. I wanted a system that measures the bar's actual motion through space rather than its projection onto an assumed line, which is what pushed me toward a free, untethered sensor that reconstructs the true 3D path directly from onboard inertial data rather than from cable geometry.
That's the difference between a device built for a controlled test and one built for how people actually train.
Léo Fortin Dionne designed the advanced measurement systems behind ORCA Strength Systems.