Full Wiring 6 — Bed-battery combiner (BED board + DC Monitor)

wheel / pinch = zoom · drag = pan
fw6_combiner FULL WIRING SHEET 6 — BED-BATTERY COMBINER (§5.9 · §6.5, rev 3.55) contactor + pre-charge, pin-by-pin — state machine runs LOCALLY on the BED board (Option B); no truck-length bus in the safety loop WHY pre-charge: two low-ESR LiFePO4 banks + inverter input caps — slam the contactor at ΔV and only ~15 mΩ of cable resists → 30–60A+ spike, caps look like a dead short for ms → arcing / welded contacts / BMS trips. The 5Ω makes worst case a gentle 2.7A. · physical layout, glands, breather: Sheet 8 cluster_bank REMOVABLE BED BANK — 2× Dumfume 300Ah (~7.7 kWh) one chain per pack (×2), paralleled on the way to STUD A · each pack ~30 kg = liftable (§5.8) cluster_sb ANDERSON SB175 + handle (×2) contactor open → mates onto a DEAD circuit (only the DC-Monitor sense tap) = NO SPARK plug-in auto-detected: pack V appears on ch2, ~5s debounce → sequence runs, no user action cluster_power POWER SECTION — SEPARATE box/bracket (Sheet 8) contactor · resistor · bus bars — high current stays OUT of the control box cluster_cont CONTACTOR 150–200A · 12V coil · NORMALLY OPEN continuous-duty (White-Rodgers 120-105 / EV200 class) NO = FAIL-SAFE: bed board dead / power lost → coil drops → banks isolate cluster_box CONTROL BOX — Lost Creek medium IP67 (Sheet 8) logic only: bed board · DC Monitor · RLY-PC · small fuse block (5A feeds) pack BED PACK ×2 Dumfume 12V 300Ah LiFePO4 removable · store at 50–60% never charge below 0°C prot PER-PACK PROTECTION MRBF 100–150A AT the + post (fuse FIRST) 275A rotary disconnect (red key, ~$12) OFF = mate/unmate ALWAYS dead + service lockout, independent of software pack->prot 2 AWG sbn − contact pack->sbn 2 AWG sbp + contact prot->sbp 2 AWG studA STUD A (battery side) sbp->studA 2 AWG ×2 bedneg BED − BUS BAR negative NEVER fused, NEVER switched − trunk = the bed↔front ground (star, §5.9) sbn->bedneg 2 AWG studB STUD B (bus side) studA->studB CLOSED = mΩ path dcm MODBUS DC MONITOR (SKU 33931) 0–36V · 16-bit · 4-ch, on the bed board's LOCAL RS485 ch1 = front/bus V · ch2 = bed stud-A V → ΔV REPLACES the 100k:10k divider → ADC2 (rev 3.30) common-GND required — star ground = OK ⚠ current ±8A only → NOT the main shunt (WonVon stays) studA->dcm ch2 bed-V · 22 AWG + 1A fuse AT the tap bedbus BED + BUS BAR always tied to the front bus via the 125A-fused trunk (Sheet 1) → also = front-V studB->bedbus 2 AWG coil coil + / coil − 1N4007 flyback ACROSS the coil (stripe → coil +) — clamps the inductive kick, protects RLY-O5's contacts coil->bedneg coil − return res PRE-CHARGE RESISTOR 5Ω · 25–50W aluminum-housed wirewound, bolted to metal (heatsink for the fault case: 13.5V/5Ω = 2.7A · 36W) res->studA lands ON STUD A = bypass AROUND the contactor r5 RLY-O5 — contactor driver (Bosch, socketed) 86 ← fused +12V · 85 ← DO1 sink (via G1) 30 ← 5A fused feed · 87 → coil + waterproof socket next to the contactor r5->coil 87 → coil + · 16–18 AWG · ~0.6–1A bedbus->dcm ch1 front/bus-V esp BED ESP32-S3 BOARD (Waveshare 8DI-8DO) runs the combiner state machine LOCALLY (§6.5) board dies → DOs dark → contactor opens = FAIL-SAFE DO1 → RLY-O5 · DO2 → RLY-PC · RS485 → DC Monitor Ethernet to front = coordination/telemetry ONLY dcm->esp RS485 Modbus — local, in-box seq SEQUENCE (local, on the bed board) 0 ISOLATED — both DOs off. NO contactor = fail-safe. 1 REQUEST (plug-in auto-detect / app) — read ch1 vs ch2. 2 PRE-CHARGE ON (DO2) — current through the 5Ω: I = ΔV/5 .    Fixes in SECONDS: inverter caps (τ = 5Ω × ~10mF ≈ 50ms)    + surface charge. Can NEVER fix a real SoC gap    (0.1–0.4A into 300Ah+ ≈ 0.05%/hr) — ΔV still there    after ~15s = genuine SoC gap, waiting won't close it. 3 DECIDE ~15s, by ΔV (inrush ≈ ΔV / 15–25 mΩ loop):     ΔV ≤ 0.2–0.3V → CLOSE (~10–20A blip — the common    case: LiFePO4 rests 13.0–13.4V over most of SoC)    0.3–0.7V AND charge source ON → close = managed merge    >0.7V → STAY ISOLATED + alert (rule at right) 4 CLOSE (DO1) — nothing "switches over": the contactor's    milliohms simply parallel the 5Ω; current takes the metal. 5 +0.5s: DO2 off — resistor out of circuit. COMBINED. 6 SEPARATE — DO1 off (LVD / fault / request) → state 0. esp->r5 DO1 sink → 85 (out via G1) rpc RLY-PC — pre-charge relay (Bosch, in-box) 86 ← fused +12V · 85 ← DO2 sink 30 ← 5A fused feed · 87 → 5Ω (out via G1) esp->rpc DO2 sink → 85 front FRONT / CAB Switch A → front board + OptiPlex MQTT/PWA announce, telemetry NOT in the safety loop esp->front Ethernet (Cat5e) rpc->res 87 → 16 AWG via G1 plugnote PLUG / UNPLUG PLUG-IN: sparkless (dead circuit) → auto-detect ~5s debounce → COMBINED in ~15–25s, hands-off. UNPLUG: SEPARATE FIRST (app / override / rotary OFF), THEN pull the SB175 — while combined, ch2 reads BUS V through the closed contactor (a yank is invisible, and the connector would break load current = arc). rules RULES DEEPLY-DEPLETED (>0.7V band): the 5Ω cannot fill a dead bank (~0.2A into 300Ah+ ≈ forever). Combine ONLY under ACTIVE CHARGE: the charger lifts the bed toward charge V, banks meet there, THEN pre-charge + close. The SOURCE fills the bank, not the front battery. REBOOT (§6.3): restore_mode ALWAYS_OFF on DO1/DO2 — never restore relay state from flash (pack may have been swapped) · ~10s boot-settle before energizing anything · no DO1 until both voltages read + banded. A reboot mid- COMBINED drops the contactor (break-rated) → self-heals: re-read, ΔV≈0, band-1 recombine. parts PARTS (combiner subsystem) · 150–200A 12V continuous-duty NO contactor · 5Ω 25–50W aluminum-shell resistor · 1N4007 · 2× Bosch relay (RLY-O5 socket + RLY-PC in-box) · 2× MRBF 100–150A + holders (one per pack + post) · 2× Anderson SB175 + handles · 275A rotary disconnect · Modbus DC Monitor SKU 33931 (V-sense — no divider) · wire: 2 AWG battery/studs/bus · 16 AWG pre-charge   · 16–18 coil · 22 AWG sense (+1A fuse at the tap)

open the PDF version ›

‹ prevnext ›