Two different things happen to the baseline of a recorded trial. The device tares (zeros itself at whatever load it currently sees). The platform separately runs a baseline-correction step on the recorded signal (subtracts a measured rest offset from every sample, so every metric is computed against "force above resting state"). The two operations are independent, and understanding both helps you read the chart correctly, retare when it matters, and ignore the tare state when it doesn't.
There's also a counterintuitive point worth stating up front: on a tension-gauge dynamometer (which is the most common kind in clinical strain-gauge work), a small amount of resting pretension on the strap is favorable, not a problem. It makes countermovement at onset visible on the chart. A perfectly zeroed baseline can hide a countermovement entirely.
What the device tare does
Pressing the tare button on the dynamometer records the device's current reading and subtracts it from every subsequent reported value. The next sample reads zero (or close to zero), and all subsequent samples are reported relative to that. The device firmware doesn't know what state it was tared in; it just knows what offset to apply.
This is the same operation on every BLE dynamometer, isokinetic export, or load cell on the market. It's not platform-specific behavior.
What the platform's baseline correction does
When a trial is recorded (or uploaded), the platform computes a trimmed-median offset across the pre-onset rest window of the signal, ignoring the top and bottom 10% of samples to suppress any spike outliers. That offset is then subtracted from every sample on the trial.
The result is that every metric is computed against "force above this trial's actual resting state," regardless of what the device thought zero was at the moment of tare. If you tared and then the strap settled slightly under tension before the patient pushed, the platform's correction step takes care of the small drift. The metrics don't care what the device's offset was; they care what the deflection was.
This means small tare offsets between trials don't matter much for the reported values. A few Newtons of pretension or a few Newtons of drift, the correction step absorbs both.
Why some resting pretension is favorable on tension dynamometers
This is the load-bearing point of the article. On a tension dynamometer (the strain gauge senses pulling force through the strap; the Tindeq Progressor, the PitchSix Force Board, the VALD DynaMo and FightTech in a tension setup, and most rehab-oriented BLE dynamometers), the trace can go in two directions during the moment before onset:
- At zero baseline (perfectly tared, strap loose): the trace sits at zero. If the patient does a countermovement (relaxes the limb briefly before pushing), the strap is already at zero and cannot read negative tension. The countermovement is invisible on the chart.
- With a small amount of resting pretension (the strap is slightly loaded, the trace sits at maybe 5–15 N before the patient pushes): the countermovement unloads the strap. The trace dips below the pretension level, clearly visible as a downward bump before the rise.
The dip is the diagnostic. A countermovement before maximum-effort onset shifts the auto-detected onset late and contaminates the early-window RTD, even with the platform's baseline correction (the correction subtracts a constant offset; it doesn't unwind a transient dip). The earlier you see the dip, the earlier you can decide whether to override the onset, exclude the trial, or re-cue the patient. Pretension is what makes the dip visible.
The published methodology for identifying contraction onset relies on being able to see the baseline-noise envelope clearly enough to identify the last trough before the deflection (Tillin et al. 2013). A perfectly zeroed baseline collapses that envelope toward zero on one side; pretension lifts it into a regime where the trough is legible.
Practically, this means: when you're setting up a trial, you do not need to obsess over getting the strap perfectly slack at zero. Five to fifteen Newtons of resting tension on the strap before the patient pushes is fine, and on tension dynamometers it's actively useful.
Compression dynamometers behave differently
If you're using a compression-based dynamometer (a force board the patient pushes down on, a hand-grip dynamometer the patient squeezes), the trace is unidirectional. Force can only go up from baseline. A countermovement (the patient releasing before pushing) would only show as a return to zero, which already happens whenever the patient isn't pushing. There's nothing to gain from pretension on compression-mode setups because there's no signed countermovement to surface. Default to a zero baseline.
The distinction matters because several devices read in both directions, and the rule above follows the setup rather than the device. The Tindeq Progressor, VALD DynaMo, FightTech and Muscle Meter all read signed force and can be rigged either way.
Two devices cannot. The PitchSix Force Board is tension-only, and the ActivForce 2 is compression-only. Check the direction of your rig against the device before you build a protocol around it, because a tension-only sensor loaded in compression does not fail loudly. It clamps and still returns a number, which is the worse failure of the two. ForceIQ detects the clamping and flags the session, but the trial is spent by then.
When to retare mid-session
Small offsets are handled by the platform. The cases where retaring mid-session is worth doing:
- Visible baseline drift over the course of a session. If the pre-onset baseline on trial three is meaningfully different from trial one (the strap has migrated, the patient has shifted in the chair, the dynamometer has settled), retare to bring it back to a sensible starting point.
- Switching limbs. If your bilateral protocol involves moving the dynamometer from the left side to the right side, the resting load on the strap is going to be different by the time you've re-strapped. Retare on the new side.
- After a disconnect. Some devices reset their internal calibration on disconnect; a brief re-tare after reconnect is good practice.
Retaring is not necessary between trials within the same set as long as the baseline looks stable. The platform's correction handles a few Newtons of drift; you only need to physically retare when the drift is large enough to make the chart hard to read.
When the baseline is broken in a way tare cannot fix
A few patterns to recognize on the chart that aren't fixable by retaring:
- The baseline is noisy throughout. Pre-onset, mid-trial, post-release, the trace is fluttering rather than holding steady. Usually a loose strap, a patient who is fidgeting on the seat, or BLE interference. Re-strap, reposition, or reconnect.
- The post-release baseline doesn't return to the pre-onset baseline. The patient released, but the trace settled at a different level than where it started. Usually the strap migrated under load. Retake the trial after re-setting the strap.
- The baseline ramps slowly throughout. Drift, often thermal or from strap slippage. Retake and watch for the same pattern; if it persists across multiple trials, the issue is the setup, not the recording.
For cases where the baseline is broken on a trial that's already been recorded, see when to exclude a trial.
What to do next
- On tension dynamometers, embrace a small amount of resting pretension. Five to fifteen Newtons of strap load before the cue is fine and actively useful for spotting countermovement.
- Trust the platform's baseline correction for small offsets. You do not need to perfectly zero the device before every trial.
- Retare when the baseline is visibly drifting, not on a fixed schedule. Visible drift is the signal; the schedule is the noise.
- If pretension is unavoidable because of strap setup, that's fine. If a perfectly zero baseline is unavoidable for the same reason, that's also fine. The platform is forgiving on the absolute baseline; what matters is that the trial's own rest period is quiet enough for the onset detector to see clearly.
References
- Tillin NA, Pain MTG, Folland JP. Identification of contraction onset during explosive contractions. Response to Thompson et al. J Electromyogr Kinesiol. 2013;23(4):991-994. doi:10.1016/j.jelekin.2013.04.015