ECSS Crimping — Crimping Operation (space-systems/ecss/q7026-crimping-operation)
Use when the Process clause of ECSS-Q-ST-70-26 is the task: running the
crimp cycle itself, with the tooling, the setting and the completion of
the cycle all decided before the handles move and none of them
recoverable afterwards.
Domain quick reference
- The cycle is not adjustable. Everything that decides whether the joint
is a cold weld or a mechanical clamp is settled by the setup, so the
grading here is mostly a grading of the setup.
- The die and the positioner belong to the contact part number. The die
closes a profile and the positioner decides where along the barrel it
closes, so a die from a neighbouring contact produces a crimp whose
height the band does not describe at all.
- The selector setting belongs to the wire gauge inside that contact,
not to the contact alone. One setting carried across a mixed-gauge
harness over-compresses the fine wire and under-compresses the heavy
one, and both crimps come out looking finished.
- Calibration has two clocks. A tool is due on a date and on a cycle
count, whichever arrives first, and a busy tool reaches the cycle
limit long before the date. Either lapsing makes the compression
applied unknown rather than wrong, which is why the crimps taken
afterwards have no disposition rather than a bad one.
- The ratchet is the only thing enforcing a full cycle. A cycle released
by hand applied part of the compression and left a crimp that looks
complete, which is exactly why the tool has a ratchet instead of a
stop.
- A barrel takes one cycle. A second pass work-hardens the material
along a different profile, and more compression is not better
compression.
- Crimp height fails in two opposite directions. Below the band the
barrel has been driven into the strands and cut them; above it the
barrel never closed and the joint is mechanical.
- A failed crimp is cut off, not adjusted. What separates a remake from
a reject is therefore the wire length that is left, not anything about
the crimp.
Workflow
- Validate the setup table: a die and positioner per contact part, and
per gauge inside it a selector setting, a height target and a
tolerance that stays below the target.
- Resolve the (contact, gauge) entry, refusing a contact the table does
not carry and a gauge the contact does not accept rather than
borrowing a neighbouring setting.
- Authorize the cycle before it runs: die and positioner against the
contact, selector against the gauge, calibration against both its
date and its cycle interval. Collect every blocker rather than
stopping at the first.
- Grade the cycle: completion, exactly one crimp on the barrel, and a
ratchet that was actually fitted.
- Grade the compression: the measured height against the band either
side of the target, naming over-crimp and under-crimp separately
because they are opposite process errors.
- Take the worst disposition, then convert a remake into a reject where
the wire margin left is short of what a re-termination needs.
- Roll the run up: accepted, remake and rejected counts, first-pass
yield, the number of cycles that ran unauthorized, and the crimps the
record never covered.
Pitfalls
- Selecting tooling by wire gauge alone. The die and positioner are
properties of the contact, and the gauge only chooses the setting
inside it.
- Carrying one selector setting across a mixed-gauge harness. The
setting that is right for the heavy wire crushes the fine one.
- Checking the calibration date and not the cycle count. A tool in a
production cell reaches its cycle interval months before its date, and
the date sticker still reads current.
- Treating a lapsed calibration as a failed crimp. The crimps taken
afterwards have unknown compression, which is a records problem with a
different remedy from a bad crimp.
- Releasing a ratchet by hand to free a jammed contact and keeping the
crimp. The compression that was applied is whatever the handles
reached.
- Running a second cycle to correct a light crimp. The barrel is already
work-hardened and the second profile does not undo the first.
- Reading crimp height as a single-sided limit. A check written one way
round passes every barrel that never closed.
- Offering a remake without looking at the wire. In a harness whose
routed length is already cut, the contact can come off and nothing can
go back on.
- Comparing a crimp height with a band edge, or a cycle count with its
interval, by bare arithmetic. A height specified to land exactly on
the limit can evaluate a few units in the last place past it after a
unit conversion, so the comparison absorbs that representation error
while the limit stays untouched.
Behavior contract (gate 3)
The setup table validation, the (contact, gauge) resolution and its
refusals, the die and positioner match, the selector check, the two-clock
calibration check, the pre-cycle authorization with its collected
blockers, the cycle completion and single-crimp rules, the two-sided
compression grading, the remake-to-reject conversion on wire margin and
the run rollup with first-pass yield are exercised by the gate 3 contract
test: scripts/test_q7026_crimping_operation.py against
scripts/q7026_crimping_operation_logic.py (stdlib unittest, offline).
Run:
python3 scripts/test_q7026_crimping_operation.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and
paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
1---2name: q7026-crimping-operation3description: "Perform and grade one crimp cycle on a high-reliability electrical connection. Use when a crimp is being made and the setup has to be defensible before the handles move: match the die and positioner to the contact part number, take the selector setting from the wire gauge inside that contact, refuse a tool whose calibration has lapsed on date or on cycle count, confirm the ratchet released only on a completed cycle and that the barrel took exactly one, then grade the resulting height against its two-sided band and split remake from reject on the wire left. Trigger: ecss, q-st-70-26-crimping, crimp-die-and-positioner-match, crimp-selector-setting-by-gauge, crimp-tool-calibration-currency, crimp-ratchet-full-cycle, crimp-over-and-under-compression."4license: Apache-2.05compliance: STANDARDS-REF6standards:7 - id: ecss8 reference-only: true9gated: false10domain: space-systems11pack: space-systems12compatibility: "agentskills.io SKILL.md; any SKILL.md host (Claude Code, Hermes, OpenClaw)"13metadata:14 source: https://skillmd.com/skills/ashfordeou/q7026-crimping-operation15 domain: space-systems16 subdomain: ecss17 tags: [ecss, q-st-70-26-crimping, q7026-crimping-operation, crimp-die-and-positioner-match, crimp-selector-setting-by-gauge, crimp-tool-calibration-currency, crimp-ratchet-full-cycle, crimp-over-and-under-compression]18 version: 0.1.019 author: Aero Agent Skills20---2122# ECSS Crimping — Crimping Operation (space-systems/ecss/q7026-crimping-operation)2324Use when the Process clause of ECSS-Q-ST-70-26 is the task: running the25crimp cycle itself, with the tooling, the setting and the completion of26the cycle all decided before the handles move and none of them27recoverable afterwards.2829## Domain quick reference3031- The cycle is not adjustable. Everything that decides whether the joint32 is a cold weld or a mechanical clamp is settled by the setup, so the33 grading here is mostly a grading of the setup.34- The die and the positioner belong to the contact part number. The die35 closes a profile and the positioner decides where along the barrel it36 closes, so a die from a neighbouring contact produces a crimp whose37 height the band does not describe at all.38- The selector setting belongs to the wire gauge inside that contact,39 not to the contact alone. One setting carried across a mixed-gauge40 harness over-compresses the fine wire and under-compresses the heavy41 one, and both crimps come out looking finished.42- Calibration has two clocks. A tool is due on a date and on a cycle43 count, whichever arrives first, and a busy tool reaches the cycle44 limit long before the date. Either lapsing makes the compression45 applied unknown rather than wrong, which is why the crimps taken46 afterwards have no disposition rather than a bad one.47- The ratchet is the only thing enforcing a full cycle. A cycle released48 by hand applied part of the compression and left a crimp that looks49 complete, which is exactly why the tool has a ratchet instead of a50 stop.51- A barrel takes one cycle. A second pass work-hardens the material52 along a different profile, and more compression is not better53 compression.54- Crimp height fails in two opposite directions. Below the band the55 barrel has been driven into the strands and cut them; above it the56 barrel never closed and the joint is mechanical.57- A failed crimp is cut off, not adjusted. What separates a remake from58 a reject is therefore the wire length that is left, not anything about59 the crimp.6061## Workflow62631. Validate the setup table: a die and positioner per contact part, and64 per gauge inside it a selector setting, a height target and a65 tolerance that stays below the target.662. Resolve the (contact, gauge) entry, refusing a contact the table does67 not carry and a gauge the contact does not accept rather than68 borrowing a neighbouring setting.693. Authorize the cycle before it runs: die and positioner against the70 contact, selector against the gauge, calibration against both its71 date and its cycle interval. Collect every blocker rather than72 stopping at the first.734. Grade the cycle: completion, exactly one crimp on the barrel, and a74 ratchet that was actually fitted.755. Grade the compression: the measured height against the band either76 side of the target, naming over-crimp and under-crimp separately77 because they are opposite process errors.786. Take the worst disposition, then convert a remake into a reject where79 the wire margin left is short of what a re-termination needs.807. Roll the run up: accepted, remake and rejected counts, first-pass81 yield, the number of cycles that ran unauthorized, and the crimps the82 record never covered.8384## Pitfalls8586- Selecting tooling by wire gauge alone. The die and positioner are87 properties of the contact, and the gauge only chooses the setting88 inside it.89- Carrying one selector setting across a mixed-gauge harness. The90 setting that is right for the heavy wire crushes the fine one.91- Checking the calibration date and not the cycle count. A tool in a92 production cell reaches its cycle interval months before its date, and93 the date sticker still reads current.94- Treating a lapsed calibration as a failed crimp. The crimps taken95 afterwards have unknown compression, which is a records problem with a96 different remedy from a bad crimp.97- Releasing a ratchet by hand to free a jammed contact and keeping the98 crimp. The compression that was applied is whatever the handles99 reached.100- Running a second cycle to correct a light crimp. The barrel is already101 work-hardened and the second profile does not undo the first.102- Reading crimp height as a single-sided limit. A check written one way103 round passes every barrel that never closed.104- Offering a remake without looking at the wire. In a harness whose105 routed length is already cut, the contact can come off and nothing can106 go back on.107- Comparing a crimp height with a band edge, or a cycle count with its108 interval, by bare arithmetic. A height specified to land exactly on109 the limit can evaluate a few units in the last place past it after a110 unit conversion, so the comparison absorbs that representation error111 while the limit stays untouched.112113## Behavior contract (gate 3)114115The setup table validation, the (contact, gauge) resolution and its116refusals, the die and positioner match, the selector check, the two-clock117calibration check, the pre-cycle authorization with its collected118blockers, the cycle completion and single-crimp rules, the two-sided119compression grading, the remake-to-reject conversion on wire margin and120the run rollup with first-pass yield are exercised by the gate 3 contract121test: scripts/test_q7026_crimping_operation.py against122scripts/q7026_crimping_operation_logic.py (stdlib unittest, offline).123Run:124python3 scripts/test_q7026_crimping_operation.py125126## Compliance127128- ECSS standards are freely downloadable (ESA); cite the source and129 paraphrase per standards-map.yaml.130- compliance: STANDARDS-REF, gated: false.