← all reports
caliper session operator measurements 2026-08-21

Measurements In

Five readings came back off the real hardware. Two internal cross-checks passed, one reading was revised, and the result overturns what the field-test document said was wrong with the gripper — the crossing fingers were never a hole-position problem.

The measurements check out against themselves

Two bolt positions were read independently — R_in at (27, 2.5) and R_out at (36.7, 5.0), both relative to the fixed jaw's inner bolt with the fixed jaw's bolt line as the x-axis. The distance between them computes to 10.017 mm against a known bolt spacing of 10.0 mm0.017 mm out. Nothing about the points, the frame or the caliper is wrong.

01 — Where the points are

One blade first

Drawn to scale, and the outline is lifted straight out of Blade_v4_final.stl — not a sketch. The blade stands 89.4 mm from the bottom of its mounting foot to the tip.

内侧螺栓 外侧螺栓 P1 / P4 装配脚底面 最外侧那个角 = 离另一片爪最远的那个角 内侧角 (不要量这个) TL / TR 刀片最顶端的那个尖 89.4 mm 脚底到尖端 ← 这一侧朝向 另一片爪 (抓取面)
A single blade in side view. The side facing left is the grip face — the one that meets the other blade. P1 / P4 is the corner of the mounting foot's bottom face furthest from the other blade; TL / TR is the point at the very top.
中心线 左爪(固定) 右爪 TL TR L_in L_out R_in R_out P1 P4 T — 两个尖端 C — 两个内侧螺栓 ✅ 已量 (27, 2.5) B — 两个外侧螺栓 A — 两个脚外角
The same blade mirrored and closed against itself, with the four spans marked. C is the one already measured. Note that the innermost point of each blade is in the middle of the protrusion, not at the tip — which is why the tips do not touch on the current geometry.

02 — What came back

ReadingMeasuredPredictedStatus
R_in — right jaw inner bolt(27, 2.5)baseline
R_out — right jaw outer bolt(36.7, 5.0)(36.8, 4.4)✓ matched
Right pair spacing (derived)10.01710.000✓ self-check
θ — angle between bolt lines14.45°10.79°! 3.66° out
Official blade: outer bolt → protrusion face22–23 mm23.52✓ matched
Official blade: bolt centre → bottom face2.5 mmnew
Official pair, fully closedjust touchingthe spec
Previous printed pair, fully closedcrossingthe defect
!

One reading was revised, and one prediction was wrong

The outer-bolt-to-face distance first came back as 19 mm and was then revised to 22–23 mm. The first value produced a confident conclusion of HOLE_DX = −3.163 mmretracted. Nothing was changed in the repo on the strength of it.

The operator also caught an error in the prediction that accompanied it: the outer bolt is on the far side from the protrusion face, so its distance to that face must be larger than the hole-pair centre's, not smaller. The predicted 13.52 mm should have been 23.52 mm. With the sign fixed, the prediction and the revised measurement agree.

03 — The crossing was never about the holes

Measured on the two meshes directly

BladeOuter bolt → protrusion faceBehaviour when closed
Previous version — printed, field-tested26.27 mmcrosses
v4 — never printed21.77 mmunknown
Official22–23 mmjust touching

Both versions carry their bolts at the same x positions — −0.857 and 9.143, 10 mm apart. The holes never moved. What differs is the depth of the blade body: the previous version reaches 3–4 mm further toward the other jaw than the official one, and that is what makes the fingers cross.

v4 already changed this. At 21.77 mm it sits at the edge of the official 22–23 mm band — between 0.2 and 1.2 mm short. So requirement 2B's instruction to shift the hole pattern inward was written against a blade whose real problem was elsewhere, and applying it to v4 would likely push it the wrong way.

Recommendation: leave HOLE_DX at 0 for now

The remaining discrepancy is 0.2–1.2 mm, which is inside the precision of a "22 to 23" reading, and TPU is compliant enough to absorb a few tenths. Changing it in the wrong direction costs more than leaving it. If a reading to 0.5 mm comes back — a single number, not a range — then HOLE_DX = reading − 21.768 settles it in one step.

04 — One change that is clear

The hole sits 0.25 mm too low

BladeBolt centre → bottom faceHole edge → bottom face
Previous version3.500 mm2.100 mm
v42.250 mm0.850 mm
Official2.500 mm1.100 mm

The operator's note on why this matters: without it the holes will not line up with the bolt openings on the mounting bracket. Raising HOLE_Z by 0.25 mm matches the official part and improves the material left below the hole from 0.850 to 1.100 mm — the source comments call 0.865 mm the practical floor for TPU, so v4 is currently just under it. This one is a clear improvement in both directions and has no downside.

05 — Still open

ItemWhat it needs
θ disagreeing with the URDF by 3.66°Most likely the bolt line is not perpendicular to the jaw's long axis, in which case the URDF figure simply cannot be used to predict bolt geometry. The measurement wins either way — but the closure page's prediction needs qualifying.
FLATTEN sign — mating face angleNeeds A (foot outer corners) and T (tips). Decides whether the mating face should be flattened to the blade axis or left following the rail.
HOLE_DX to 0.1 mmA single-number reading of the official blade's outer bolt → protrusion face, to 0.5 mm.
Sleeve jaw openingThe sleeve costs 18.68 mm of opening. Whether a 90 mm cup still fits has never been checked, and it runs the gripper motor so it needs the operator present.