Leopoldo Rodríguez
01dc30c06a
- contenido curado desde Drogheda via scripts/sync-docs.sh (procedimientos de banco, montaje/arneses, diagnostico remoto, guias git) - scripts/deploy.sh: mkdocs build --strict + rsync al servidor Helgen - README con setup de nginx + htpasswd para acceso privado de tecnicos Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01YJ4eAB6zjQT49RVwpgfKdt
103 lines
6.9 KiB
Markdown
103 lines
6.9 KiB
Markdown
---
|
||
title: EinarBoard 1.0 (rev 1.1) — Bring-Up Runbook & Anticipated-Trouble Register
|
||
tags: [project-drogheda, hardware, einarboard, bringup, testing, runbook]
|
||
project: Project Drogheda
|
||
author: Helgen CTO
|
||
date: 2026-07-08
|
||
status: ACTIVE — execute in order when boards arrive
|
||
---
|
||
|
||
# EinarBoard 1.0 — Bring-Up Runbook
|
||
|
||
**Order (FINAL — replacement of 2026-07-08):** original PCBA SMT026070860564 cancelled (2 BOM
|
||
lines lost at matching); replacement placed 2026-07-08, **live-verified complete: 53 lines,
|
||
116/118 designators assembled** incl. D2/DBK2/DBK1 and all six LEDs. Rotations confirmed by Leo
|
||
in the placement viewer. Deltas vs the original parts map: LEDs = **C2295 KT-0805R (RED — all
|
||
indicators red by choice)** · RS1 = **C2988654** (HoLLR1206-1W-8mR-1%, spec-compliant alternate) ·
|
||
100nF = C1591 · 22µF ×4 = C76637 · R51 = C22972 · 22k = C31850. **J1/J2 NOT assembled (the only
|
||
unselected parts)** → bench jig-solder; buy two 1×20 2.54 female sockets locally (the C50984 reel
|
||
is locked in JLC inventory for future PCBA). Setpoints re-derived from ordered values 2026-07-08: fan 4.99 V · modem 3.33 V ·
|
||
I_chg 2.42 A · input limit 3.12 A · PFI 4.61 V · backup 4.79 V — all on target.
|
||
|
||
---
|
||
|
||
## A. NOW — before production starts (the $479 gate)
|
||
|
||
- [ ] **DFM Analysis check (4–6 h after ordering, Order History → DFM Analysis).** Verify the
|
||
polarity dozen against the answer sheet: **U6 pin-1 TOP-LEFT · U2/U7 pin-1 BOTTOM-RIGHT ·
|
||
Q7 pin-1 TOP-LEFT · bands: D2/Dbk2/D5/D10 LEFT, Dbk1 TOP, D6 RIGHT · electrolytics "+":
|
||
C3 BOTTOM, C10/C43 LEFT · LED cathodes LEFT (D1, D4G1, DB1–4)**. The preview already caught
|
||
U6 and U2 turned 180° once — assume nothing.
|
||
- [ ] **Confirm Production File is set to Yes with 48 h auto-confirm** — an unwatched weekend
|
||
approves itself. Approve only after the DFM check passes.
|
||
- [ ] Answer every JLC review email **same day** (schedule protection).
|
||
|
||
## B. Reception (day of delivery)
|
||
|
||
- [ ] Visual vs the ordered renders: DNPs bare (P3/P13/H3, MOD1 field), all sockets present,
|
||
P1 wire-entry toward the edge, J9 opening at the edge.
|
||
- [ ] Polarity dozen again, on physical boards, one board, 5 minutes with a loupe.
|
||
- [ ] Measure J1↔J2 socket spacing on one board (decision record says 15.00 mm) and test-insert
|
||
a Blackpill — expect light preload if the module measures wider; note insertion force.
|
||
|
||
## C. Board #1 electrical bring-up — NO modules socketed, current-limited bench supply
|
||
|
||
1. **Resistance sanity (unpowered):** 24V_IN→GND, MAIN5V→GND, 3V3_MODEM→GND — no shorts (>1 kΩ).
|
||
2. **24 V path:** bench supply 24 V, limit 200 mA → P1 pins 1/2. Check: 24V rail after Q10 ≈ 24 V
|
||
(reverse-protection FET conducting), **5V_FAN = 5.0 V** (U2 running), fan header P6 pin 1.
|
||
3. **Vehicle 5 V path:** 5 V, limit 500 mA → P2. Check: **MAIN5V ≈ 5 V** (LTC4040 input FET
|
||
passing), SYS5V after P11 jumpered/switched ON, Blackpill socket pins J1-20/J2-18 = 5 V.
|
||
4. **Load switch default-OFF (invariant #1):** with no MCU, SW_5V must be **0 V**. Then short
|
||
J1-7 (PWR_CTRL) to 3V3 momentarily → SW_5V = 5 V, **3V3_MODEM = 3.33 V** (U7 up).
|
||
5. **Ignition sense:** 12 V and 24 V through P1-3 → IGN_SENSE (J1-11) clamps ≈ 3.5–3.6 V; 0 V → <0.3 V.
|
||
6. **⚠ CHARGER NTC TRAP (this WILL look like a dead charger):** with no battery-harness NTC on
|
||
P4-3, the LTC4040's NTC pin is pulled to VIN by R55 → reads "battery too cold" → **charging
|
||
silently paused**. For bench charge tests, connect **10 kΩ from P4-3 to P4-2 (GND)** to emulate
|
||
25 °C. Do not debug the charger without it.
|
||
7. **Charger (with NTC emulation + real 1S4P pack or bench cell):** connect pack to P4 →
|
||
CHRG (J2-11) low while charging; charge current ramps ≤ 2.42 A; input current never exceeds
|
||
~3.1 A (watch the bench meter — this validates RS1/CLPROG behavior).
|
||
8. **Backup event (THE LTC4040 validation — this block never saw a proto):** system running on
|
||
P2 5 V with pack connected → pull P2 → **VSYS holds ≈ 4.79 V** from boost, board keeps running.
|
||
Measure: no back-feed voltage appearing on the disconnected P2 pins (validates Q7 orientation
|
||
fix). Restore P2 → seamless return + recharge. **Do this before any fleet commitment.**
|
||
9. **K1 fail-safe:** unpowered board: continuity K1 pin 1 ↔ pins 5/6 (NC closed = corrections
|
||
path exists with dead MCU). Powered + PB5 high: path opens to NO (pin 10).
|
||
|
||
## D. Module integration (board #1 only, then replicate)
|
||
|
||
- [ ] **MOD1 socket strips: solder using the simpleRTK2B Pro as the jig** (strips on module →
|
||
insert → solder → remove). Same jig trick for any stiff J1/J2 fit.
|
||
- [ ] Blackpill in (USB end toward board-left / VB-B12 corner at pads-1 — sockets don't key!).
|
||
Flash bench firmware via module USB.
|
||
- [ ] **Firmware flag: PROD `CHRG_EN` (PA8) is HIGH = charge OFF** — the proto TP4056 logic was
|
||
INVERTED (`CHRG_RUN` HIGH = ON). Port the flag before flashing, or units will disable charging
|
||
when they mean to enable it.
|
||
- [ ] Pro into MOD1 (antenna top-left, USBs bottom, per the validated orientation). **Old-rev
|
||
Pros: XBee switch → UART1. All revs: Pro's own XBee socket stays EMPTY; never plug the Pro's
|
||
"USB XBee" port in service** (its FTDI sits on the RTCM3 downlink on new revs).
|
||
- [ ] **F9P config: write to FLASH, not RAM** (Jun-20 finding — RAM-only config dies at power
|
||
cycle). Verify with a power cycle before shipping any unit.
|
||
- [ ] 4G NTRIP Master into J7/J8 (VCC row = J7). Telcel micro-SIM installed first. Board USB-C
|
||
(J9) enables the modem's USB only when a host provides VBUS — bench console via J9 is safe;
|
||
it does not fight the uplink.
|
||
- [ ] End-to-end: open sky, corrections from caster → **GGA quality 4 (RTK fixed)** — the
|
||
fleet's first-ever verified fix ≥4 per the Jun-20 gap.
|
||
|
||
## E. Known-accepted risks (decision register — watch, don't re-litigate)
|
||
|
||
| Risk | Decision/owner | Watch for |
|
||
|---|---|---|
|
||
| J1/J2 rows at 15.00 mm vs 15.24 reading of module | Leo, vernier-verified, 2026-07-07 | Insertion force; pin preload; use jig-solder trick if stiff |
|
||
| LTC4040 block fab'd without proto validation | Gate waived consciously (boards cheap) | Step C.8 is the validation — run it FIRST |
|
||
| C22 = 35 V part on 24 V node (spec said 50 V) | Accepted at order | Nothing expected; note for rev 1.2 BOM |
|
||
| C778721/C90142 ceramic bias derating | Within design margins | — |
|
||
| D4G1 modem LED driven from 1.8 V pin | Known-dim (audit L4) | Don't debug as a fault |
|
||
| Backup boost budget 1.93 A of 2.5 A at modem TX burst | Margin accepted; 104 µF VSYS bank | Brownout during TX only if budget exceeded — scope VSYS during TX test |
|
||
|
||
## F. Fleet replication (after board #1 passes everything)
|
||
|
||
Per-unit checklist: NTC harness populated (removes the C.6 emulation) · module orientations ·
|
||
switch-to-UART1 on old Pros · F9P FLASH config verified by power cycle · jig-soldered strips ·
|
||
GGA≥4 open-sky check · then HBOX integration per the enclosure docs (fit verified 2026-07-07).
|