WIRING SCHEMATIC
PHOTONS → BUSBAR · v8 BUYLIST-SYNCED · 2026-11-12
RUN 58VDC/210A · companion to BUYLIST.SYS v11.0 · phone-first Kafkas · gPV 20A before panels — NO EXCEPTIONS
FULL DC TRAIN — TOP TO BOTTOM
Array → combiner/SPD → 30m twisted bundles → wall boxes → MPPTs → shared bus → packs + Studers
OVERCURRENT ROLES — THE MENTAL MODEL
Hardware quantities, SKUs and prices: the dedicated SHOPPING LIST — BUY ORDER v10.5 page is the source of truth
Why the chain looks the way it does: every layer has ONE job, and the jobs get more brutal as you move toward the battery
THE THREE JOBS — learn these and every device on the schematic snaps into place
| Job | Device family | What it actually does | Why it can't do the other jobs |
|---|---|---|---|
| EXECUTE | fuses — gPV 20A · NH00 80A aR · NH1 250A aR | melts open when current exceeds its rating for long enough. aR class melts in milliseconds — before the copper heats — because semiconductors and fat battery shorts give it no time to spare | a fuse dies doing its job. It can't be reset, can't be switched daily, and can't tell a fault from a thunderstorm-day surge unless the rating is chosen lean (that's sizing, below) |
| ISOLATE | switches — SD63 racks & isolators · Studer DC disconnects | opens a circuit on YOUR command, for service or emergency. It carries current all day, interrupts it cleanly (DC-rated contacts), and never acts on its own | a switch has no fault detection and no interruption muscle for a dead-short — asking a breaker to clear a 48V bank short means watching it arc until it burns |
| CLAIM EARTH | surge gear — SPD T2 · the ELEMKO rod | gives lightning a designated exit: the SPD clamps a spike and dumps it down the 7px earth bus to ONE rod, so it never routes through your MPPTs instead | does nothing against overcurrent — a surge is over in microseconds; fuses barely twitch. Different monster, different weapon |
THE LADDER, WALKED — trace one fault and you know the whole system
1 · a string faults at the panel: its gPV 20A fuse at the combiner melts — ONLY that string opens, the other 11 keep harvesting. The fuse sits AT the source because a fault must be stopped before it enters the 30m wire run.
2 · an MPPT leg faults: the NH00 80A aR in box #2 clears in milliseconds — fast enough that the 25mm² leg never warms. The blue SD63 next to it doesn't even notice; its job starts only when a human walks over and pulls it for service.
3 · the bank itself shorts: three 314Ah packs can deliver thousands of amps. This is NH1 territory: the 250A aR main is the ONLY device on the page rated to interrupt that monster without a sustained arc — it sacrifices itself instantly, and the 95mm² trunk (good for ~400A+) never leaves its temperature corridor.
4 · a pack needs service: no fuse moves. You open the SD63 rack / disconnect, the pack goes quiet while the NH1 stays armed protecting the other two. Executing and isolating never borrow each other's uniforms.
THE SIZING LOGIC — why each number is that number
- 20A gPV on a 12.94A string: ~1.5× the string's working current — above any cloud-edge spike the panels can produce, below the wire's tolerance. The classic lean window.
- 80A NH00 on a 60A MPPT leg: 75% loading — headroom so full harvest never nuisance-trips, yet 8× quicker to melt than the wire's own limit would demand. aR class because the leg feeds electronics, not heaters.
- 250A NH1 on a 210A BMS bank: the BMS is the pack's own first guardian at 210A; the NH1 backs IT up — 250A means "bank running flat-out is legal, but a dead short is not."
- Order on the wall: fuse closer to the SOURCE, switch after it. The protective device must always see the monster first; the switch is the door, not the shield. That's why box #2 draws orange above blue, and why the NH1 sits between busbar and packs.
- Cascade test — the only question that matters: when the smallest fuse opens, has it protected everything upstream? Each rung only faces faults smaller than itself. If a rung can meet the full battery, it must be an aR fuse — full stop.
INTERIM MODE — SINGLE PACK OPERATIONS
Until packs #2 & #3 land — same hardware, reined-in limits
- ARRAY: 12 PANELS / 3 MPPTs. One string pulled per the interleaved plan — leave combiner #4's breaker OPEN, coil the 2 spare strings. Peak charge ≈ 100A vs. the 120A single-pack BMS ceiling — full 16-panel fleet would push ~128A and flirt with nuisance trips at solar noon.
- 16 kWh bank, same-day recovery. ~100A charge rate takes the pack from empty to full in ~3.5 sun-hours — the baseline array's whole point survives the interim.
- LOAD DISCIPLINE. One pack = 210A BMS ceiling vs. ~500A of Studer potential. Set Studer power limits, stagger motor starts — a BMS trip mid-mix is an on-air dropout.
- RESTORE = 15 MINUTES. When pack #2/#3 arrive: top-balance, bolt to busbar (equal-length leads), flip combiner #4's breaker, hang the last 2 strings. Hardware unchanged — this is a breaker + 2 strings undo.
CORRECTIONS FOLDED IN — V8.0 → V8.1 STRING ACCOUNTING (12 Nov 2026)
Rule 28 append · fleet math checked against the 150/60-Tr specs: nominal 3,440W @48V · max PV Isc 50A — "higher current may damage controller"
- Baseline grows to 24 panels / 12 strings: 3 strings (6 panels, 2,700W) per MPPT = 78% of nominal — inside the 80% line. Output ~39A: legs, NH00 80A aR and 25 mm² all still correct.
- 32-panel verdict REVOKED: 4 strings × 13.7A Isc ≈ 55A short-circuit vs the 50A damage ceiling. The old "hard ceiling, doable" line was a 150/45-era number. New law: beyond 24 panels, ADD MPPTs (2× 150/60 more = 36 panels) — never stack a 4th string.
- At 3-string expansion: combiner main breaker 25–32A → 50–63A (bundle ~39A); gPV count becomes 12× 20A. The 15A gPV minis in combiner #4 are spliced to 20A on this page.
- SVG still draws the 2-per-combiner baseline — visual redraw queued; all texts already speak 3-per.