Worked Example: Rugged Control Box
This is Exercise 1 worked all the way through: a small sealed control box with 24 VDC input, Ethernet, CAN, four sensors, one motor output, and a debug port. We'll run each interface through the decision paths, then turn the result into the kind of paperwork the templates expect.
Illustrative composite — not a case report
The box below is a teaching scenario built on Exercise 1, not a report of any single project. The reasoning is the point, and every connector fact is stated by reference to the deep dives and decision paths, which carry the sources.
Read this as reasoning, not a shopping list
Like the Selection Packet and the M12 example, this page teaches how to decide and document. Part numbers, counts, and lengths marked J- or TBD are placeholders. A real project replaces every one with an exact, datasheet-verified value. Don't carry any rating from this example into your hardware.
Scenario
A shoebox-class sealed controller mounted on a machine frame in a light-industrial bay: splash and dust, occasional washdown mist nearby, no submersion. It powers from the plant's 24 VDC supply, talks Ethernet to the line controller, sits on a small CAN segment with two other nodes, reads four DC sensors on the frame, drives one small 24 VDC gearmotor, and needs a service port a technician can reach without opening the enclosure. Cables are unplugged whenever the frame section is serviced — by whoever is on shift.
The quiet headline decision: every interface below lands on one circular family (M12-class), on purpose. One shell size, one seal system, one torque procedure, one cordset supplier list — the same standardization logic the M12 example uses, extended to a whole box. The engineering effort then goes where it belongs: keying, labeling, and documentation so that nine same-size connectors across six interface types can never be confused.
Requirements
| Interface | Requirement | Notes |
|---|---|---|
| 24 VDC input | Box supply, current TBD from the power budget (assumed within a power-coded M12 class below — verify against the exact datasheet) | Plant 24 VDC distribution; source-side protection TBD |
| Ethernet | One run to the line controller; rate TBD from the controller spec — 10/100 assumed below, to be confirmed | Shielded run, washdown-adjacent |
| CAN | Multi-drop segment, this box + 2 nodes; bus power TBD | Termination location must be designed, not discovered (§8.5) |
| Four sensors | Discrete DC sensors on the frame, per-circuit current TBD vs. exact part ratings | Swapped by shift technicians — unplug/replug service model |
| Motor output | One small 24 VDC gearmotor, stall current TBD from the motor datasheet | No separate feedback run in this scenario |
| Debug / service port | Reachable without opening the box; sealed when unused; low mating-cycle duty | Used a few times a year, capped otherwise |
| Environment (all) | Sealed-when-mated target per plant spec (exact IP class TBD); unmated state must also be protected | Caps are part of the design, not an accessory |
Connector selection table
| Interface | Requirement summary | Candidate families | Selected family | Status |
|---|---|---|---|---|
| J1 — 24 VDC input | Sealed DC power inlet | M12 power-coded; bare cable gland; DT-style | M12 power-coded (T- or L-coded per the confirmed power budget — §8.1) | Pending power budget |
| J2 — Ethernet | Sealed industrial Ethernet | M12 D-coded; M12 X-coded; sealed RJ45 | M12 D-coded, 4-pin (§8.1) | Pending rate confirmation |
| J3 — CAN | Sealed multi-drop fieldbus drop | M12 A-coded 5-pin; hardwired gland | M12 A-coded, 5-pin ("some CAN" rides A-coded — §8.1); topology per §8.5 | Selected |
| J4–J7 — Sensors | Four sealed DC sensor circuits | M12 A-coded 4-pin; M8; glands into a junction box | M12 A-coded, 4-pin (Industrial sensor path) | Selected |
| J8 — Motor output | Sealed small-motor DC power | M12 power-coded; M23-class; DT-style | M12 power-coded (same coding as J1 — one power-cordset type in the crib) | Pending stall-current check |
| J9 — Debug/service | Sealed, occasional-use service port | Sealed M12 A-coded 8-pin; USB-C behind a cover; Micro-D behind a hatch | M12 A-coded, 8-pin, with a chained screw cap (Debug/service path) | Selected |
Rejected connectors
| Interface | Rejected option | Why rejected |
|---|---|---|
| 24 VDC input | Cable gland + hardwired tail | Fails the service model — every box swap becomes a wiring job, done by whoever is on shift; a gland is the right answer only for a run that never unplugs, and this run does unplug during service |
| 24 VDC input | MIL-DTL-38999 | No program requirement to justify the cost/lead-time class (defense path logic in reverse) |
| 24 VDC / motor | M12 K-coded | AC power coding (§8.1) — wrong lane for a 24 VDC system |
| 24 VDC / motor | A-coded pins doing power duty | The ~4 A-class note in §8.1 exists for exactly this temptation — power belongs on a power coding the datasheet supports |
| Ethernet | M12 X-coded "to be safe" | X is for GbE/10G-class; D-coded is the honest 10/100 answer, and §8.1 says don't default to X — revisit only if the rate requirement changes |
| Ethernet | Sealed/rugged RJ45 | Workable, but breaks the one-family standardization and adds a second seal/latch system to maintain (Rugged Ethernet path checklist concerns) |
| Sensors | M8 | Viable technically; rejected as a standardization decision — one shell size across the box (same reasoning as the M12 example) |
| Sensors | Glands into a junction box | Service model again — sensor swaps become enclosure work inside a sealed box; the M12 example rejects glands on the same grounds |
| Motor output | M23-class | Right family for a real servo axis with feedback; oversized for one small DC gearmotor — becomes the answer if the motor grows (see "What would change") |
| All sealed runs | DT-style sealed automotive | Good parts in their lane (budget path); rejected because the industrial cordset ecosystem here is M12 — fighting the ecosystem means custom pigtails |
| Debug/service | Bare USB-C on the panel | The debug path allows USB-C only behind a cover; a bare consumer port on a washdown-adjacent panel fails §12.4's boundary |
Pinout
Assignments below are illustrative structure only — every pin function, and especially every wire color, is verified against the exact device/cordset datasheet and recorded in the ICD before release. Wire-color conventions are example-only, never design authority (Source Notes).
| Connector | Pin | Signal | Direction | Notes |
|---|---|---|---|---|
| J1 (24 VDC in) | 1…4 (+FE if L-coded) | +24 V / +24 V / 0 V / 0 V (paired per the coding's datasheet) | In | Contact pairing and FE per the exact power-coded part (§8.1) |
| J2 (Ethernet) | 1–4 | TX+ / RX+ / TX− / RX− per the D-coded cordset datasheet | Bidir | Shield continuity per §5.7; category per the cordset spec |
| J3 (CAN) | 1–5 | Shield / bus power / bus return / CAN_H / CAN_L per the device and protocol documentation | Bidir | Multi-drop: this port is one drop — T-piece and end-termination per §8.5 |
| J4–J7 (sensors) | 1–4 | +V / signal / 0 V (+ spare or second signal) per each sensor datasheet | In | Same discipline as the M12 example |
| J8 (motor) | 1…4 | Motor + / Motor − on paired contacts per the power-coded datasheet | Out | Stall current vs. the exact part's rating — §4's derating discipline |
| J9 (service) | 1–8 | UART TX/RX, boot/enable strap, logic ground, reserved | Bidir | Logic-level only — no power export; capped when unused |
Cable drawing
One row class per run; the cable drawing template carries the full per-cable version with evidence columns.
| Wire | Gauge | Color | Pair / shield | End A | End B | Notes |
|---|---|---|---|---|---|---|
| J1 power conductors | Per cordset spec (sized to the power budget) | Per cordset — recorded, not assumed | Unshielded power | J1 plug (molded) | Plant 24 VDC drop | Verify current at temperature vs. the exact assembly |
| J2 Ethernet | Per cordset category spec | — | Shielded, pairing per spec | J2 plug (molded) | Line controller | Shield termination both ends per §5.7 |
| J3 CAN drop | Per CAN cable spec | — | Shielded pair + power | J3 plug | Bus T-piece | Drop length within protocol limits (§8.5) |
| J4–J7 sensor runs | Per cordset spec | — | Unshielded | Sensor M12 | J4–J7 | Off-the-shelf molded cordsets, stocked lengths |
| J8 motor run | Sized to stall current | — | Per motor/EMC needs | J8 plug | Motor leads | Anchored against gearmotor vibration |
- Cable jacket: oil-resistant per the plant environment spec; drag-chain-rated only where a run actually moves (none in this scenario — record that, too)
- Labels: every cable labeled at both ends with the J-number; label spec in the ICD
- Length tolerance: per the cable drawing template's tolerance field, set from routing reality, not guessed
- Test requirements: continuity + hipot/IR per the harness inspection checklist and the acceptance standard the program names; design-level qualification evidence, where the program requires it, structured per the qualification plan template
Backshell / cap notes
Molded cordsets carry their own strain relief and sealing, so this box needs no field backshells — the accessory budget goes to caps and torque instead.
| Interface | Accessory | Notes |
|---|---|---|
| All receptacles | Sealing screw caps, chained | The unmated state is part of the sealing design (§8.3); an uncapped receptacle is an open hole |
| J9 service port | Cap plus a log discipline | The port is capped 360 days a year — the cap is the interface most of the time |
| All | Torque per each part's datasheet | Coupling-torque values are manufacturer-specified (§8.3); finger-tight is not a spec |
| CAN segment | Terminating M12 plug at each bus end | Termination lives at the bus ends, not "wherever" — §8.5 |
ICD entry
One worked entry (J2) in the ICD template's shape; the real package carries one per interface.
- Interface name:
J2-RCB-ETH— control box to line controller, industrial Ethernet - Connector P/N:
TBD— M12 D-coded 4-pos panel receptacle, front-mount, shielded - Mating connector P/N:
TBD— molded D-coded cordset, shielded, length per routing - Contact P/Ns: n/a (molded assembly) — field-repair kit P/N
TBD, temporary-article rule per the M12 example §6 - Backshell P/N: n/a (molded); panel-side sealing per receptacle datasheet
- Dust cap P/N:
TBD, chained - Keying / polarization: D-coding is the key (§8.1); no same-coding neighbor on this panel face
- Electrical limits: per the exact receptacle/cordset datasheets — recorded here with revision, never quoted from memory
- Shielding: 360° continuity through the receptacle to the enclosure bond point (§5.7)
- Environmental assumptions: sealed-when-mated per the exact assembly's tested rating; capped when unmated; plant spec
TBDgoverns - Mating cycles: per datasheet (A4 carries the family-level orientation)
- Torque / assembly notes: coupling torque per datasheet; torque tool
TBD; record value + tool in the build record - Cable requirements: category/shield per the cordset spec; bend radius per datasheet; no drag-chain duty
- Test / inspection requirements: harness inspection checklist + program acceptance standard
Risks and mitigations
| Risk | Mitigation |
|---|---|
| Nine same-size circular connectors (six interface types) on one small box — wrong-port mating attempts | Codings differ where it matters most (power vs. D vs. A); among the A-coded ports, position counts differ (4/5/8) — and verify cross-mating behavior between position counts against the manufacturer rather than assuming; add color bands + J-labels at both ends; photograph the panel in the service manual |
| A-coded pins drafted into power duty during a "quick fix" | The rejected-connectors table records why power lives on the power coding; the ICD's electrical-limits line makes the boundary auditable (§8.1) |
| CAN termination forgotten or duplicated | Terminating plugs are BOM line items with J-numbers, not accessories; topology drawing in the package (§8.5) |
| Receptacles left uncapped after service | Chained caps; capped-state photo in the close-out checklist; the service port's cap logged like a tool |
| Motor stall current outgrows the selected power coding | Stall check is a release gate (requirements table); escalation path pre-named (M23-class — see below) |
| Ethernet rate requirement changes after release | The D-vs-X decision and its trigger are recorded (§8.1) — a rate change reopens J2, not a debate |
| Cordset vendor change alters pinout/wire colors | Colors are example-only by rule; the ICD pins functions to pin numbers with datasheet revisions (Source Notes) |
What would change if…
- The power budget outgrows the M12 power class — J1/J8 move up a family (the high-current path takes over, derating curve first), and the one-family standardization is consciously traded away, in writing.
- The motor becomes a servo axis — feedback appears, and the whole J8 question transfers to the motor + feedback cable path: drive-ecosystem cordsets, M23-class connectors, and the EMC discipline that comes with them.
- The box moves into direct washdown — re-verify every mated and unmated rating against the plant spec (§8.3) and re-run the sealed enclosure feedthrough path for the panel itself.
- A defense/aero customer appears — the requirement set, not preference, moves the external interfaces toward 38999-class hardware, and the Selection Packet shows what that documentation grade looks like.
- Someone proposes one big multipole for everything — that's the removable machine module path's problem statement; run it honestly before deciding: use the path for the service reality, and §5's anatomy for the blind-mate and float mechanics a docking interface leans on.
Documentation bundle
What "done" means for this box, per the templates:
- Requirements table with sources and status closed out
- Comparison matrix rows for the contested decisions (J1 power coding, J2 D-vs-X)
- Pinout + cable drawing per run, evidence columns filled
- ICD entry per interface (J1–J9), datasheet revisions recorded
- Design review checklist pass — the caps, torque, and termination rows are the ones this box lives or dies by
- Harness inspection checklist on the built cables
note
These examples illustrate structure and process, not a released design. The datasheet, applicable standard, and program requirements decide the actual values.