The 30 N insertion force does not decide this sensor choice by itself. Select around the smallest defect-related force change, the direction in which it appears, and the signal noise on that axis. For a go/no-go gauge moving along tool Z, the FT-150 and FT-300 both list 0.5 N signal noise in that direction; the FT-300 becomes less attractive if the inspection must resolve small X- or Y-axis resistance because its listed noise there is 1.2 N.
Reject the quick fixes first
Do not choose the sensor with the larger name, treat a resolution number as a trip threshold, or set the reject limit equal to the noise figure. Those shortcuts can produce unstable pass/fail results.
| Quick fix | Why it fails | Correct action |
|---|---|---|
| Use the 30 N insertion force as the only selection value | The measurement challenge is the smaller resistance change that identifies a defect. | Determine the minimum pass-to-fail force separation on the active axis. |
| Use 0.5 N as the acceptance threshold | Normal gauge friction and signal movement can reach the decision boundary. | Measure good-part friction and set the threshold above its observed range. |
| Compare 0.2 N resolution directly with 0.5 N noise | Resolution and signal noise describe different behavior. | Compare current specifications using the same metric. |
| Filter until the trace looks smooth | Excess filtering can hide short resistance events or delay the stop decision. | Apply only enough filtering to stabilize classification, then test at production speed. |
The FT-150 instruction-manual value is 0.2 N effective resolution. Updated specifications replaced the different earlier terms with a single signal-noise specification of 0.5 N. That change means the two numbers must not be treated as interchangeable limits. Check: write down the active axis, smallest defect force to detect, and the current noise metric before selecting hardware.
Fix the measurement axis
Align the gauge insertion and retraction direction with tool Z when the mechanical design permits it. A force sensor reports forces in its own coordinate system; tool tilt, mounting offsets, cable pull, and side loading can move an axial contact into X, Y, and Z readings.
- Define the sensor coordinate system in the robot program.
- Jog the unloaded tool slowly in the insertion direction and confirm which signed channel represents entry into the hole.
- Apply a controlled axial contact and verify that Z dominates while X and Y remain secondary.
- Repeat during retraction because friction reverses direction while a mechanical bind may not be symmetric.
If the design later uses lateral motion inside the hole to find resistance, X and Y become measurement channels rather than alignment diagnostics. That change directly affects the choice because the FT-300 lists 1.2 N signal noise in X and Y. Check: accept the coordinate setup only when axial contact appears primarily in Z and the sign of insertion and withdrawal forces is documented.
Select the sensor from the force direction
| Selection point | FT-150 | FT-300 |
|---|---|---|
| Listed signal noise | 0.5 N | 0.5 N in Z; 1.2 N in X and Y |
| Earlier published measurement term | 0.2 N effective resolution in the instruction manual | No resolution value stated in the supplied specification information |
| Axial go/no-go insertion | Same listed Z-axis noise basis as FT-300 | Good fit when Z is the inspection axis |
| Fine lateral resistance measurement | Evaluate against the required X/Y discrimination using current specifications | Confirm that 1.2 N X/Y noise leaves adequate pass/fail separation |
| UR5 integration preference | Viable candidate | Better physical fit on the UR and more versatile for this application |
Signal noise is movement around the true force value and is similar to white noise. A zero-load Z signal therefore moves rather than remaining at exactly 0 N. The listed noise value alone does not define a guaranteed minimum detectable force because its statistical definition, bandwidth, and filtering must also match the measurement method.
Choose the FT-300 when insertion is reliably on Z and its UR fit or multi-axis versatility matters. Before using it for small lateral-defect detection, prove that the bad-part X/Y response is well separated from the 1.2 N noise and normal process variation. Check: test representative good and known-defective parts on every axis used for acceptance; do not release a selection based only on the 30 N nominal insertion force.
Mount, zero, and capture the baseline
- Mount the gauge, housing, and sensor in their final orientation. Keep cables from applying changing force or torque during motion.
- Read the product documentation for allowable force and moment limits before applying the 30 N insertion load. Stop if the combined tooling load, contact force, or off-axis moment approaches an unspecified or unverified limit.
- Zero or bias the force reading with the tool clear of the part and in the same pose used at the start of each test.
- Run multiple good parts through insertion, dwell if used, internal motion, and retraction. Record the Z baseline, normal friction band, peak insertion force, and withdrawal force separately.
- Repeat after the tool and fixture have reached normal operating condition. Mechanical seating, gauge wear, contamination, and thermal drift can shift the baseline.
A good gauge does not produce zero resistance in real hardware. Sliding contact, alignment error, surface condition, and seal or fixture forces create a nonzero normal signature. A threshold placed at 0.5 N can therefore reject good parts even when the sensor is operating correctly. Check: the zero reading must be stable, cable movement must not change classification, and every sampled good-part trace must remain inside the proposed acceptance band.
Program separate insertion and inspection decisions
Do not reduce the cycle to one instantaneous comparison. Separate successful insertion from resistance detection, and evaluate force over the motion segment where each condition matters.
- Approach the hole without contact and establish the local zero.
- Insert along Z while monitoring signed Z force. Use the verified 30 N process requirement as the commanded application target only if that is what the mechanism requires; keep the rejection threshold as a separately commissioned value.
- Declare insertion successful only when the programmed position or travel condition is reached without exceeding the validated force boundary.
- Move the gauge through the required inspection path. Monitor Z for axial binding and X/Y if the gauge is deliberately moved laterally.
- Retract while monitoring the opposite signed force and classify withdrawal resistance independently.
- Stop motion on a reject condition and retain the peak values and the motion stage that triggered it.
Set limits from distributions of good and defective parts, not from sensor noise alone. The decision boundary needs margin above normal friction, noise, drift, and repeatability while remaining below the smallest verified defective-part response. If these ranges overlap, improve alignment, fixturing, gauge condition, motion profile, or signal processing before production release. Check: the program must identify insertion failure, internal resistance, and retraction resistance as distinct outcomes.
Prove the complete cycle
- Run repeated no-contact cycles to expose zero drift and cable-induced signals.
- Run representative good parts and confirm no false rejects across insertion, internal movement, and retraction.
- Run known-defective or controlled challenge parts and confirm that each defect crosses the intended threshold on the expected axis.
- Repeat at the production speed. Confirm that any filtering does not suppress a short force event or move the trip until after damaging contact.
- Introduce credible setup disturbances one at a time, including small alignment changes and normal gauge seating variation, then verify that classification remains stable.
- Record the sensor model, active axes, zeroing condition, thresholds, filter configuration, robot motion conditions, and acceptance results for maintenance.
The final proof is separation: the highest good-part response, including normal variation, must remain below the lowest reject response with usable margin on every acceptance channel. Check: release the station only after it detects every challenge part and passes every qualified good part without operator interpretation.
FAQ
Why does the force reading move when nothing touches the gauge?
Signal noise makes the reading move around the true value. The listed value is 0.5 N for the FT-150, and 0.5 N in Z plus 1.2 N in X and Y for the FT-300.
Why does a 0.5 N reject threshold fail on good parts?
The gauge has real insertion and sliding friction in addition to sensor noise. Measure repeated good-part traces and place the threshold above their observed range while retaining separation from defective parts.
Why does FT-300 selection depend on the inspection axis?
Its listed signal noise is 0.5 N in Z but 1.2 N in X and Y. It suits a Z-axis insertion test when the required discrimination fits that noise, but small lateral resistance must be proven with representative parts.
Stop here if the good- and bad-part force ranges overlap, the active axis cannot be established, the current noise definition is unclear, or the 30 N load and resulting moments have not been checked against the sensor ratings. Contact official product support for the current specifications, mounting limits, signal-noise definition, and integration guidance before resuming commissioning.