Dual Force Plates: When Crosstalk Fakes an Asymmetry

A club volleyball setter has shown up on the weekly CMJ screen with a 'weak left leg' for six straight sessions - left plate reading 8 to 11% lower peak force than right, every single week, tidy as a metronome. The strength coach has already added single-leg work to her program by the time anyone asks the two questions worth asking before touching a program based on a number like that: what happens if she stands on the plates mirrored, and what does the left plate read with absolutely nothing on it? Both answers come back wrong, in the same direction. The left leg was never weak. One of the two plates had been reading a piece of somebody else's force the entire time.

A dual force-plate rig is, mechanically, two independent sensors sharing one testing surface, and the sharing is exactly where the trouble starts. Crosstalk is any signal that shows up on one plate's channel because of a load that was actually applied to the other plate, not to the one reporting it. It travels by two routes, and most field setups are exposed to both without anyone ever isolating which one is doing the damage.

Mechanical crosstalk moves through whatever the two plates are physically tied to - a shared subframe, a shared plywood deck, a shared rubber anti-vibration mat, or a floor not stiff enough to keep one plate's deflection from reaching its neighbor. Load one plate hard enough, a drop landing, a max isometric pull, and the structure underneath both plates deflects by some fraction of a millimeter. The neighboring plate's load cells register that deflection as force, even though nothing touched it.

Electronic crosstalk is rarer on modern systems but still shows up on older or poorly grounded amplifiers: a strong signal on one channel bleeds into an adjacent channel's wiring or shared power supply, adding a small, force-proportional offset to a channel that should be sitting near zero. Either route leaves the same signature. The 'quiet' plate's trace does not stay flat while the other plate is loaded - it moves in step with it, usually in the same direction, at some fraction of the loaded plate's magnitude.

Portable dual-plate systems bought specifically to measure bilateral asymmetry are, ironically, among the setups most exposed to this. A single in-ground plate poured into its own pit and separated from the surrounding floor by an expansion gap has no neighbor to leak into. Two portable plates set side by side on a gym floor, bolted to a shared aluminum sled for easy transport, or resting on the same rubber gym-flooring tile have exactly the shared structure mechanical crosstalk needs to travel through.

Two setup habits make it worse without anyone noticing at the time. A tare or zero routine that zeroes both plates simultaneously from one shared 'zero' command, rather than confirming each plate independently reads zero with nothing on it, bakes a small residual offset into every session that follows and looks identical to genuine asymmetry. Uneven leveling compounds it - a plate resting even a few millimeters higher on one edge redistributes part of the vertical load into a shear component the plate's calibration matrix was never built to cleanly separate from vertical force, and that redistribution gets worse the closer the tilt sits to the shared edge between two plates, which is precisely where an athlete's feet land in most single-leg and split-stance protocols.

None of this shows up as an obvious fault reading. It shows up as a number sitting inside the range coaches already expect from real athletes, which is what makes it dangerous. Bishop, Turner and Read (2018, Journal of Sports Sciences) pooled interlimb asymmetry data from dozens of studies in a systematic review and found reported values for jump and strength tests spanning from under 5% to more than 20%, depending on the test, the population, and - the review flagged this directly as a limitation of the literature it was synthesizing - the calculation method and measurement protocol used in each lab. A crosstalk-inflated asymmetry of 8-12% does not stand out inside that spread. It reads as an unremarkable athlete.

Exell, Irwin, Gittoes and Kerwin (2012, Journal of Sports Sciences) measured step-to-step variability in lower-limb asymmetry during sprinting and found within-athlete asymmetry swinging widely from one step to the next, in some cases exceeding the average between-limb difference the same athletes showed overall - evidence that a single trial's asymmetry score is not reliable on its own, regardless of where the noise comes from. Their protocol measured sprint kinetics rather than dual-plate CMJ or isometric testing, so it does not isolate crosstalk as a specific cause, but the core finding transfers directly: a number that moves this much trial to trial cannot be trusted from one reading, and mechanical crosstalk is simply one more noise source stacked on top of the biological variability their data already documented.

Beckham, Suchomel, Mizuguchi, Sole and Stone's (2014, New Studies in Athletics) review of force plate use in performance testing likewise flagged mounting rigidity and independent calibration as recurring, underreported sources of error in field-based setups - a caution written with single-plate installations in mind that applies with more force, not less, once two plates share a mount.

Run this before trusting any asymmetry number a dual-plate system has produced, and repeat it any time the rig is moved, re-leveled, or a cable is disconnected and reconnected.

1. Confirm both plates are individually zeroed with nothing on either one, and record each plate's baseline for a full 5 seconds - a baseline that drifts or never settles near zero on its own is a separate problem worth fixing before testing anything.
2. Have the athlete stand with full weight on Plate A only, the other foot physically lifted clear of Plate B, not just unweighted.
3. Read Plate B's output during that 5-second hold. A properly isolated system should sit within noise of zero, typically under 1-2% of Plate A's loaded value by most manufacturers' isolation specifications.
4. Repeat with full weight on Plate B only, checking Plate A.
5. If either 'empty' plate reads more than roughly 3-5% of the loaded plate's force, treat every asymmetry number this rig has produced as suspect until the mounting is addressed.

The foot-off check catches gross bleed-through, but a subtler version of the same problem survives it: a small, consistent bias baked into one plate's calibration or leveling that only shows up under real athlete loading, not an empty-plate test. The swap-sides test catches that version, and it needs nothing more than one extra trial.

1. Run the protocol as normal - left foot on Plate A, right foot on Plate B - and record the asymmetry index (smaller limb minus larger limb, divided by the larger limb, times 100).
2. Have the athlete physically swap position: left foot now on Plate B, right foot now on Plate A. Repeat the same protocol, same warm-up state, same day if at all possible.
3. Compare the two readings. If the asymmetry follows the limb - the same leg reads weaker no matter which plate it stands on - that is a real, biomechanical asymmetry.
4. If the asymmetry follows the plate instead - whichever leg happens to be on Plate A reads stronger, regardless of which physical leg that is - the rig itself is producing a systematic difference between channels, and no conclusion about the athlete's limbs should be drawn until that is fixed.
5. A result that partially flips, shrinking or reversing sign without fully mirroring, usually means both are present at once: a real underlying asymmetry, partly masked or amplified by a plate-level bias. Fix the rig first, then re-test to size the genuine number.

A club volleyball program running weekly CMJ asymmetry screens on a portable dual-plate rig flagged a setter's left leg at roughly -10% for six straight sessions, a value sitting comfortably inside the normal range reported in the asymmetry literature and consistent enough week to week that nobody questioned it. A reference check on a single validated plate, testing each leg alone with the same known load, put her true asymmetry at just -2.6%, unremarkable for a healthy athlete. The swap-sides test on the dual rig settled the discrepancy in one extra trial.

Plate B tracked the reference value almost exactly in both orientations. Plate A under-read by roughly 8% regardless of which leg stood on it, dragging down whichever limb happened to be assigned there that day. The asymmetry did not follow the athlete's body - it followed the hardware. Re-shimming Plate A onto its own isolation pad instead of the shared aluminum transport sled it had been bolted to closed the gap to within 1.5% on the next foot-off check, and asymmetry readings across the following month settled around the true -2 to -3% baseline.

Once a foot-off check or swap-sides test confirms crosstalk is present, the fix is almost always mechanical, not software.

None of these fixes require replacing hardware in most cases - they require treating two force plates as two separate instruments that happen to sit next to each other, not as one combined sensor with two output columns.