Plant room wallOcho Rios · rev 23 Oct 2026

Plant room wall elevation

23.43 kW PV · 2 × SPI-12K-UPRO · 48 kWh LiFePO4 · drawn to scale, dimensions in mm

Two things on this drawing can still stop the job

1 · The ceiling. Stacking the inverters above the battery cabinets needs more height than a standard 8-foot room has. The cabinets roll on their own casters, so there is no plinth — that bought back 100 mm. Bottom up:

1250 battery cabinet, standing on its own casters 500 cable zone — five containments and the bend radius of 3/0 723 inverter — SPI-12K-UPRO, §2.5 (the drawing allows 750) 300 heat clearance above ← NOT a manual figure, see below ──── 2773 REQUIRED · 9′-1″ ← this drawing, rounded to 2800 2550 absolute minimum, every clearance squeezed · 8′-4″ 2440 a standard 8′-0″ ceiling ← 110 mm SHORT of the minimum

If the wash room is 8′-0″, the inverters cannot go above the batteries. They go beside them instead, which needs about 4 m of wall — so the plant moves to two walls in an L, batteries on one and inverters on the other. Everything else here stays true; only the arrangement changes. Measure it, then the drawing gets redone.

Worth knowing about that 300 mm: the UPRO manual gives no clearance figures at all. §3.1 asks for “sufficient heat dissipation space” and nothing more. The 300 / 150 / 500 numbers on this drawing are trade practice carried over from the SEI manual, which is a different machine. They are sensible. They are not quoted.

2 · The nine-panel string. The UPRO ceiling is 500 V, not 550. Nine panels at 49 V open-circuit is about 455 V and fine; at 52 V it is 482 V and tight; at 54 V it is over 500 V and the string cannot be used. The panel Voc is still unconfirmed and it is the one number that can force a restring.

Circuits
Click a box, a trunking run or any circuit line · drag to pan · pinch or scroll to zoom

Elevation · looking at the wall · services enter top left, return wall at the right · grey channels are conduit and trunking, coloured lines are the circuits inside them

Building from this drawing

  • Chalk every box on the wall before one anchor goes in. All of it, including the containment runs. Twenty minutes, and it is the step people skip.
  • This inverter has no PV breaker and no battery breaker. Corrected 23 Oct. The earlier version of this drawing said every port had its own chassis breaker — that was the SEI manual, a different machine. The UPRO has an ON/OFF rocker and one internal breaker for AC-input overcurrent. §4.8 step 1 is “close the circuit breaker of the battery”, meaning an external one. §4.2 wants battery 3P-300 A, PV 2P-25 A per MPPT, AC in and out 3P-63 A. Fit them before anything gets landed.
  • It is IP20 and indoor-only, and it is going in a room with washing machines. §3.1 says do not install in a humid environment. The answer is a sloped drip shield above both units, open at the sides, braided steel hoses on the machines, a leak alarm on the floor, and an extractor on a humidistat rather than a switch — it has to run at 3 a.m. when the room hits dew point. It is not the US$980 upgrade to the IP65 SEI: an IP rating covers solids and liquid jets, not vapour, which is Sunpal's own point — “water vapor will enter the inverter from the bottom.”
  • Nothing passes behind a surface-mounted enclosure. Conduit runs to a box and enters through its own glands. The anchors go into the wall exactly where a hidden cable would be.
  • Fixings are supplied — do not buy anchors. §3.3: drill four holes, two M6×60 expansion screws above carry the weight, two M5 screws below hold it flat. All of them are in the carton, along with the SC95-8 battery lugs, both comms cables and two screwdrivers. The earlier M8 figure came from the wrong manual.
  • Minimum 120 mm between enclosures on this drawing, 150 mm each side of an inverter for airflow, 100 mm between the earth drop and the transfer switch. Those gaps are for your hands and a screwdriver, not for looks.
  • Everything enters and leaves at the left-hand end — four strings, the JPS feed, the house feed, the array bond and the rod conductor, all through the ceiling and the cast-in chase. No penetration anywhere in this wall.
  • Everything bound for the main breaker travels in T6 — one vertical trunking at the far left carrying four circuits: the JPS feed to the grid cabinet, the utility source for the transfer switch, the house feed out, and the earth bond. One route off this wall, one set of saddles, one place to look when something needs tracing.
  • Where a cable rises past a trunking it does not belong to, it goes through a grommeted knockout in that trunking's side wall, or in front of it on its own saddles. Decide which at set-out, not with a drill in your hand — and never let a 1000 V PV conductor share the inside of a trunking it is only passing.
  • The earth bar sits directly under the SPD box, high and on purpose. The arrester earth lead is the one length in this system that genuinely has to be short, and from there it is about 100 mm. The bar ends up around 2.4 m up — awkward once at install and never again.
  • Keep the 3/0 short. That pair carries 520 A, which is why the combiner sits between the inverters and gets its own short trunking, and the batteries are the ones that reach. 85 mm² per §4.2 — not the 67 mm² the earlier drawing showed.
  • Equal runs to both inverters — same length, same gauge, on grid, load and battery. §4.9.2 says so three times. Cut both to the longer one and coil the slack.
  • Four strings means four independent circuits. §4.9.2: each inverter's array must be independent, and PV1 and PV2 on one inverter must be independent of each other. So one surge arrester per MPPT — four, not two — and no string fuses anywhere, because a 22 A MPPT ceiling against ~18.5 A of panel short-circuit current makes paralleling impossible in the first place.
  • Each cabinet weighs around 160 kg and rolls. Check the door width, the corridor and the corners before three of them arrive on a pallet, and check the floor is level. Casters on a sloping screed put the load on two wheels. Immobilise them once positioned — locking casters or levelling feet.
  • No plinth means no flood margin. The casters only lift the base 40–50 mm. Keep the condensate line away from above them, and find out where the floor actually drains.
  • The grid line never reverses, and here that is structural. The UPRO has no on-grid export mode at all — its modes are Inverter, Mains and Off-Grid. There is no setting to disable and nothing for anyone to switch back on later. That is the one way the cheaper machine is the better machine for this house. The CT point is drawn and left unused; the CT port on the inverter is marked “reserve” by the maker anyway.
  • The generator port is spare, and it has a use. §5.5: set item 81 to AC OUT and it becomes a secondary load output that sheds automatically on overload or low battery. That is the obvious home for the water heater.
  • The earth bar is not a substitute for equipment grounding conductors. Every feeder carries its own EGC inside its own trunking — that is the fault path. The bar collects what has no feeder: the array bond, the arrester earths, the enclosures.
  • Bond to the main panel AND put a rod at the plant. Roughly 1 kV of potential rise per metre per kA/µs means a 6 m bond sits tens of kV above the service earth during a strike — fine at 50 Hz, nearly useless at lightning speed. A local rod with no bond is the one arrangement that is genuinely dangerous.
  • Label both ends of every cable before it goes in the trunking, and write the circuit on the lid at each opening. T2 and T4 each carry two bundles that look identical and are not.

The machine, as the manual describes it

Only SPI-12K-UPRO Manual (Sunpal, 23 Oct 2026) applies. The SEI-UP, HESP48120U200-H and HEBP-Pro documents in 05-datasheets/ describe a different machine and were the source of the figures corrected on this revision.

Install mode — the whole job, in order

Plan view

Plan · looking down · the working space is as much a part of the design as the wall is. The inverter is now 150 mm deep rather than 240, so the battery cabinets govern the room depth on their own.

Containment schedule

RefCarriesSizeRunWhy separate
T1 Battery 3/0 · 85 mm² — combiner to both inverters 100 × 75 trunking800 mm, local 520 A. Short, heavy, and nothing else should share its bends.
T2 Battery 2/0 — three cabinets to combiner, plus the three cabinet earths 100 × 75 trunking1,290 mm Same system as T1 but three times the cable count. The earths ride with their own circuit.
T3 PV DC, four strings at up to 500 V each 1¼″ conduit2,020 mm High-voltage DC does not share a raceway with 51 V DC or 240 V AC. Different voltage class, different enclosure.
T4 AC — grid in and load out, 240 V 100 × 75 trunking1,830 mm Grid and load may share; both are 240 V AC. Label the two bundles at every opening.
T5 Battery BMS CAN, inverter parallel link ¾″ conduit1,280 mm Furthest from the 3/0. A CAN bus run alongside 520 A of switching DC is a fault you will spend a week finding.
T6 Everything to and from the house main panel: JPS feed in, transfer-switch utility source, house feed out, earth bond 100 × 75 trunking, vertical1,160 mm The only channel that leaves this wall. Four circuits, one route, one set of saddles.

Questions for the electrician

Ten things this drawing cannot answer on its own. Some block purchases that are otherwise ready to order; the rest decide how something gets installed rather than whether it can be bought. Question 0 comes before all of them — it is the only one that can change the shape of the system rather than the contents of a box.

#AskWhat it decides
0 Where exactly do the solar feeds tap into the service, and what does the transfer switch load actually feed? Ask this first; it can change the shape of the system. This drawing shows two feeds leaving the house main panel (the grid cabinet feed and Source 2) and one arriving at it (the switch load). The panel cannot be upstream and downstream on one busbar: if all three share a bus, then with the switch on SOLAR the inverter output returns through the feed breaker into the inverter's own AC input — it would see “grid present” from itself and never go to battery. The panel is off this wall, through T6, so its internals were never resolved here. Two ways out. (A) A new service disconnect ahead of the panel: meter → disconnect → splits to the grid cabinet feed and Source 2; switch load → existing panel, which becomes 100% backed up. Right for a new build, and the feed breaker then lives in that new disconnect, so its brand stops depending on question 1. (B) A new backed-up sub-panel holding only chosen circuits, with the main panel left on raw utility — cheaper, but solar then covers part of the house and the circuits must be picked now.
1 What brand and model is the house main panel, and how many spare spaces does it have? Blocks three purchases. The 100 A feed breaker (if question 0 leaves it in this panel), the AC surge arrester, and the 2-pole breaker a wired-in arrester would need. A Siemens breaker does not fit a Square D panel and plug-in arresters are brand-specific, so there is no brand-agnostic way around this. The feed breaker needs 2 spaces and a plug-in arrester usually wants 2 more.
2 What size should the earth bus conductor be? No document in this project gives a figure. Eleven 6 mm² tees hang off one conductor running along the top of this wall — SPD box, transfer switch, both cabinet ground bars, both inverter grounding screws, the combiner body. It has to be at least as big as the largest of them and is almost certainly bigger. A real gap, not a drawing simplification.
3 Feed breaker — confirm 100 A (and that it is not 70 A) §4.2 specifies 3P-63 A per inverter. Two inverters is 126 A against a 125 A busbar — one amp over, so the grid cabinet sits at its ceiling by design. The feed breaker protects that bus, so it cannot be sized to the 70 A branches. Mains charge current (item 28, default 60 A, max 120 A — not item 07, which is total battery charge current) can be turned down to buy margin.
4 6 AWG or 4 AWG for the two inverter AC branch runs? 6 AWG is a 65 A conductor. 63 A continuous × 1.25 = 79 A. The runs are about 15 m each and the cable is not yet bought, so this is still free to change. Ask before the cable order goes in, not after.
5 Are local earth rods at the plant worth it here? The 4 AWG bond to the house main panel is required and is not in question. A rod is supplementary. A conductor rises roughly 1 kV per metre per kA/µs, so 6 m back to the service earth puts this earth bar tens of kV above it mid-strike. Never a rod without the bond — that version is worse than no rod at all. Two spare positions are left on the earth bar for this.
6 Is the plant room wall solid/filled concrete or hollow block? Picks the masonry anchor for everything on this wall. Sleeve and wedge anchors need something to expand against; in a hollow web they feel tight and pull out under load. One test hole settles it — if the bit punches into a void after about 30 mm it is hollow, and the heavy enclosures need toggles or a chemical anchor with screen tubes instead.
7 Will the inspector want the backfed breaker retainer, and does ECMBR2 satisfy it? NEC 408.36(D) requires a plug-in backfed device to be secured by a fastener needing more than a pull to release. Two genuine Siemens ECMBR2 are bought for the two backfed Q270. One thing to verify: that kit lists Ultimate / ES / PL load centres and ours is the EQ series E0816ML1125S. The panel's own installation sheet names the right part.
8 Does the mounting system provide listed bonding, or do the modules need jumpers? Decides whether 40 × 4 mm² bonding jumpers and 80 M5 ring terminals get bought at all. Look for toothed grounding mid-clamps, WEEB washers, or a UL 2703 marking on the rail. If the system is listed for bonding, the frames bond through the rail and none of that hardware is needed.
9 Do we need a DC isolator at the array, on top of the four C25 breakers already in the SPD box? Decides a purchase that is currently unordered. The four DZ47Z-63 C25 breakers in the SPD box are 1000 V DC rated, one per string, and lock off — that is already a disconnecting means, indoors and accessible. An array-side isolator adds one thing the breakers cannot do: it makes the cast-in chase dead. Open the C25s and the roof-to-wall conductors stay live whenever the sun is up, because a panel has no off switch. Against it: an enclosure on a roof in a hurricane zone, C4 salt air and full UV is a new failure point and new penetrations. If yes, it is 8 poles — four strings × two conductors — so two 4-pole units split by inverter (S1+S2, S3+S4), never two 2-pole.

One visit answers 0, 1, 3, 4, 6, 8 and 9 — and 0 should be answered standing at the meter, not at this wall. Questions 2 and 5 are design calls they may want to come back on, and 7 is for whoever signs the install off.