autosaves as you type
the markxworkshop

forming — ESG05 Calibration

measurement campaign CAL-260716 · unloading section, final row · 16.07.2026 · Proliner

REFERENCE STATEonce, before run 1 — this settles the 600/595 question class

Drive all nine motors to ONE commanded value and take points 1 + 2 (below) on all nine motors. This single state separates, for every later run: instrument frame and bed planarity (they vary by motor) from the machine’s zero point (uniform across motors). If the last-but-one row is reachable, repeat points 1 + 2 there at the same commanded value — that one repetition separates machine-wide zero point from per-row variation. Nothing is recalibrated on the machine until that decomposition is on paper.

RUN 1 41930B U1 · lite 1.1 · long side along travel

one bend, identically commanded on every row — measure the LAST row AND last-but-one if reachable: the repeatability pair across rows. This run carries the calibration. 15 rows · 9 values per row = motors 1–9 · semicolon · integer mm · leading edge first.

lower section only — the unloading section has no upper wires
RUN 2 41930B U2 · lite 1.1 · long side along travel

one bend, identically commanded on every row — measure the LAST row AND last-but-one if reachable: the repeatability pair across rows. This run carries the calibration. 15 rows · 9 values per row = motors 1–9 · semicolon · integer mm · leading edge first.

lower section only — the unloading section has no upper wires
RUN 3 41930B U3 · lite 1.1 · turned 90°

the bend runs ALONG travel, so the reachable row is nearly level by design. The last row carries file row 1 (leading edge) — confirm at load. This run confirms the turned release file runs. 9 rows · 9 values per row = motors 1–9 · semicolon · integer mm · leading edge first.

lower section only — the unloading section has no upper wires
RUN 4 41930B U4 · lite 1.1 · turned 90°

the bend runs ALONG travel, so the reachable row is nearly level by design. The last row carries file row 1 (leading edge) — confirm at load. This run confirms the turned release file runs. 9 rows · 9 values per row = motors 1–9 · semicolon · integer mm · leading edge first.

lower section only — the unloading section has no upper wires

Points per measured row (each motor)

  1. Housing corner — the agreed corner as on 15.07: toward motor 1, toward the second-last row. The same corner on every motor. (The top face carries the motor — there is no tappable centre; the centre is computed downstream from the corner, as in the 15.07 preparation.)
  2. Hinge front face, pin centre — the extension carrier. Tap the two END motors (1 and 9) twice — in the 15.07 file the last pin’s height was lost to a file artefact; a second tap makes the ends robust.

Plus, once per run:

  1. a sweep along the formed wire (the glass-contact profile).

The nine housing corners ARE the row line — no separate capture. No plumb/level capture is needed: every result on this campaign is built from same-motor distances and the row’s own plane, so nothing consumes gravity.

Wire of the measured row: the 15.07 commissioning profile took the forming steel on this row past its elastic range — the wire may carry a permanent set. This does NOT affect the motor measurements (housing/pin points) or the reference state. Wire-sweep readings on this row are not pristine until the wire is checked or replaced — the machine side’s call. Today’s four profiles are gentle (radius ≈ 3.4 m, far inside the elastic range) and cannot add to the set.

What comes back

One Proliner export per run (same preparation as the 15.07 file). Results return as: per-motor commanded vs measured extension · the uniform offset (zero point) separated from per-motor residuals (frame / planarity) · shape agreement along the wire. The 600/595 datum question is answered by the reference-state decomposition.

Source of record: job CAL-260716 (runsheet in 02 working/, articles in 03 output/) · articles = 41930B release set 260715-1415 · engine E-003 · generated 2026-07-16 by kernel/verify/gen-calibration-sheet.ts.