Study guide · Solar / PV monitoring desk

Desk first. Then the drive.

Twelve modules for technical support and NOC-style PV monitoring work: prove what you can from the portal before anyone drives to site, then walk a safe, ordered field checklist when remote fails.

12 modules Remote-first On-site ladder Safety callouts Self-check quizzes

Pick any stop on the path, or scroll in order. Progress is saved only in this browser.

Remote before travelScope, read alarms, try soft reboots and config fixes from the desk. Write a one-page field checklist only when remote evidence is exhausted.

Symptom → cause → checkEach module has a table for the ticket language you will hear, what usually causes it, and what to verify next.

Safety stays first on siteHigh DC and AC are possible. Qualified people only for electrical work. CT secondaries and combiner fuses have hard rules.

Part A · Remote desk

Prove what you can from portal, alarms, and config before anyone drives.

Desk first — then the drive

MODULE 01 · Principle

The opening rule for PV monitoring support: exhaust remote evidence before you ask for a truck roll.

Most “inverter offline” tickets are not solved with a screwdriver first. They are solved by reading the last good data time, the alarm code, the modem signal, and the plant configuration — then deciding whether a soft reboot or a config fix is enough.

When remote fails, your job is not to improvise on the phone. It is to hand the field tech a one-page checklist: what already failed remotely, what to verify in what order, and which CT/phase/ID to look at.

Core ideas

Why remote-first matters

Time: a 10-minute portal check often beats a two-hour drive.
Evidence: soft resets without a written timeline make the next visit harder.
Scope: one inverter vs whole plant changes who you call and what you pack.

The support loop

Scope

One device or plant?

Read

Alarms & last data

Act soft

Reboot / sync

04Hypothesize

CT / RS-485 / LTE

Handoff

One-page checklist

Safety — read before any on-site work High DC and AC voltages are possible at inverters, combiners, and meters. On-site electrical work is for qualified personnel only.

Lockout order: device OFF → wait for DC voltage drop (SolarEdge guidance often cites <50 V) → DC disconnect → AC breaker.
Never open a CT secondary under load. Follow the manufacturer shorting / disconnect procedure.
Never pull combiner fuses under load.
Always use the model-specific wiring diagram — A/B vs D+/D− labels vary by brand.

At the deskAsk: “What can I prove without anyone on site?” If the answer is still “nothing,” write the field order before you dispatch.

How it breaks, and what to check

Symptom	Likely cause	What to check
Dispatch before reading portal	habit / pressure	Last good data, alarm code, device count offline
“Reboot it” with no notes	no timeline	Record time, who, what changed, before/after
Field tech arrives with no plan	weak handoff	One-page checklist: polarity, termination, IDs, CT

Check yourself

Answer out loud first, then open the card.

Why write a field checklist before dispatch?

So the visit tests a specific hypothesis in order — not a random cable tug. Remote work already ruled something out; the checklist carries that forward.

Name the lockout order before opening inverter covers.

OFF → wait for DC voltage drop (often <50 V on SolarEdge guidance) → DC disconnect → AC breaker. Qualified personnel only.

Next: 2. Scope the ticket →

Scope the ticket

RemoteMODULE 02 · ~5 min

One inverter or the whole plant? Night silence or a true outage? Weather and production context before you touch anything.

Scoping is the cheapest diagnostic step. “Site offline” and “inverter 3 offline” are different problems with different owners, spare parts, and urgency.

Core ideas

Questions that shrink the blast radius

How many devices? One inverter, one logger, or every device on the plant.
Last good data time with timezone. Compare to sunset and to known maintenance windows.
Night silence vs outage: many portals show zero production at night — that is not “offline.” Offline means missing telemetry / communication loss.
Weather / irradiance: cloudy day explains low kW; it does not explain missing heartbeats.
What changed? Firmware push, SIM swap, electrician visit, Wi-Fi rename, new meter.

Blast-radius map

Single inverter offline → local RS-485/ID/power more likely.
Whole plant offline → logger, modem, Ethernet, SIM, upstream ISP.
Whole fleet region offline → vendor cloud or carrier incident (check status).

Worked example

Ticket: “Plant not producing”

Portal shows inverters online with 0 kW at 21:40 local — night. Not an outage.
Same ticket at 11:10 with last data 07:02 and logger offline → communication, not “no sun.”
Neighbor plant on same portal still updating → not a global cloud outage.

At the deskWrite one sentence: “Scope = logger + 4 inverters, last good 07:02 PT, night ruled out, peer site healthy.” That sentence drives the rest of the ticket.

How it breaks, and what to check

Symptom	Likely cause	What to check
“No production” after dark	expected night zero	Online status / last heartbeat, not kW
One inverter missing	local bus, ID, power	Siblings still reporting?
All sites in region quiet	cloud / carrier	Vendor status, other customers

Check yourself

Answer out loud first, then open the card.

Portal shows 0 kW at midnight but “online.” Outage?

Usually no — night silence. Confirm heartbeats and last good data time, not production alone.

One inverter offline, four siblings fine. Where do you start?

Local to that inverter: power/LED, RS-485 drop, unique Modbus ID — not the site modem first.

Next: 3. Portal & alarms →

Portal & alarm reading

RemoteMODULE 03 · ~6 min

Online/offline flags and alarm codes (comm timeout, meter fault, grid, fan, overtemp) — read them before you reboot.

Alarm text is a hypothesis, not a work order. “Communication timeout” does not automatically mean “reboot the modem.” It means start at the communication path and prove each hop.

Core ideas

What to capture from the portal

Device online/offline and last successful upload.
Active alarms with codes + first/last seen times.
String / MPPT / meter channels if shown (zeros vs stale).
Pending firmware or config that never applied.

Common alarm families

Comm timeout / RS-485 undetected: bus wiring, termination, ID, baud — see modules 9–10.
Meter fault: CT polarity, phase map, meter power, Modbus to meter.
Grid fault: utility / settings / actual AC issue — not a logger reboot.
Fan / overtemp: thermal path; reboot rarely fixes blocked airflow.

At the deskPaste the exact alarm string into the ticket. “Error 3x” without the vendor wording wastes the next person’s time. Don’t reboot blindly — match the code to a layer.

How it breaks, and what to check

Symptom	Likely cause	What to check
Comm timeout on one slave	wiring / ID / power on that node	Siblings OK? Unique ID? LED?
Meter fault with live production	CT/phase map	Import/export signs, phase order
Overtemp repeating	airflow / sun load / fan	Not a soft reboot candidate

Check yourself

Answer out loud first, then open the card.

Why is “reboot the logger” a weak first response to a meter fault?

Meter faults usually point at CT/phase/wiring or meter config. Rebooting the logger without reading the code loses evidence and rarely fixes polarity.

Next: 4. Remote soft actions →

Remote soft actions

RemoteMODULE 04 · ~5 min

Logger/modem reboot if available, remote inverter restart when supported, firmware/config sync, timezone and wrong meter type in the portal.

Soft actions are allowed only after you have a before-state. Capture screenshots or exports, then change one thing and wait long enough for the next heartbeat.

Core ideas

Safe remote moves

Remote logger / gateway reboot from the portal (or SMS/power-cycle tool if your process allows).
Remote inverter restart only when the vendor UI supports it and policy allows.
Push pending config / firmware that never applied — wrong meter model, wrong plant timezone, wrong baud in the cloud profile.
Correct plant timezone so “last data” and night/day logic make sense.

What soft actions do not fix

Swapped A/B, missing 120 Ω termination, star topology.
CT on the wrong conductor or reversed polarity.
SIM not seated, antenna dangling in a metal cabinet.

At the deskOne change → wait one or two reporting intervals → re-check. If it fails, leave the soft trail in the ticket and move to a field checklist — don’t loop reboots.

How it breaks, and what to check

Symptom	Likely cause	What to check
Reboot “works” for 20 minutes	thermal / flaky power / weak LTE	Note recurrence; plan site power/modem check
Wrong daily totals	timezone / meter type	Plant TZ, meter model in portal
Firmware stuck pending	device offline mid-update	Connectivity first, then retry per vendor guide

Check yourself

Answer out loud first, then open the card.

Name two portal misconfigs that look like “bad data” but need no truck.

Wrong plant timezone and wrong meter type / profile in the cloud config.

Next: 5. Remote network clues →

Remote network clues

RemoteMODULE 05 · ~5 min

Modem signal (RSRP/RSRQ when shown), SIM state, Ethernet vs LTE, Wi-Fi SSID and APN notes for the field tech.

When the whole plant disappears, the upstream path is the first remote target. Many portals expose modem diagnostics — use them before guessing RS-485.

Core ideas

What to read if the portal shows it

RSRP / RSRQ / RSSI — rough LTE quality. Weak or fluctuating signal → antenna placement, metal cabinet, carrier coverage.
SIM state — registered, not registered, PIN locked, data blocked.
IP / WAN type — Ethernet vs cellular vs Wi-Fi. A site that “always ran on LTE” suddenly on Wi-Fi (or the reverse) is a clue.
APN / ICCID — for carrier tickets and SIM swaps.

Notes that save the truck roll

Write the expected SSID and whether the cabinet is metal (needs external antenna).
Prefer Ethernet when LTE is chronically weak and a drop is available.
SolarEdge and others publish cellular / Ethernet troubleshooting guides — match symptoms to their LED and status tables (see sources).

At the desk“RSRP poor, SIM registered, last data 3h ago” already tells the field tech: reseat antenna / move to external / test Ethernet — not “rewire every inverter.”

How it breaks, and what to check

Symptom	Likely cause	What to check
Whole plant offline, logger LED on	WAN / modem / SIM	Signal metrics, SIM register, Ethernet link
Intermittent at midday only	heat + weak LTE / power brownout	Correlate with production; note for field
Modem not detected	wrong module / seats / firmware	Vendor cellular guide; modem type match

Check yourself

Answer out loud first, then open the card.

One inverter offline vs whole plant offline — which points at the modem?

Whole plant (or logger) offline. A single inverter usually stays on the local RS-485 side.

Next: 6. Wrong readings →

Wrong readings from the desk

RemoteMODULE 06 · ~6 min

Negative production, swapped import/export — hypothesize CT polarity, phase mismatch, or meter not at the PCC. Write which CT/phase to verify on site.

Communication can be perfect while energy numbers are nonsense. That is still a monitoring ticket — and often a wiring ticket waiting for a precise field note.

Core ideas

Patterns you can spot remotely

Negative production while the site is clearly generating → CT polarity reversed or phase map wrong.
Import/export swapped → CTs or meter orientation vs vendor convention.
One phase wild, two sane → that phase CT on the wrong conductor or wrong meter input.
Production looks like load → meter not at the point of common coupling (PCC); CTs after a branch.

What to write for the field

Which meter serial / which CT (A/B/C).
“Verify phase match first, then polarity” (Solis-style guidance — phase association before flipping arrows).
Confirm CTs on the correct feeders before load/inverter branching when measuring grid exchange.

Worked example

Desk note that helps

Site Sunnydale / Meter M2
Observation: export reads negative while inverter AC power > 0
Hypothesis: CT L2 polarity reverse OR phase B mapped to meter input A
Field: 1) confirm phase order vs inverter grid port
       2) confirm CT arrows per Sungrow/Solis/Enphase brief
       3) do not move CTs off PCC

At the deskNever ask the field to “check the CTs” with no which/where. Name the meter, phases, and expected arrow direction from the brand brief.

How it breaks, and what to check

Symptom	Likely cause	What to check
Negative PV while AC power > 0	CT polarity / phase	Named CT + phase order vs grid port
Export never appears	CT not at PCC	Installation photo vs one-line
Only phase B absurd	swapped B CT	Meter input B vs conductor B

Check yourself

Answer out loud first, then open the card.

Why fix phase association before flipping CT polarity?

If the CT is on the wrong phase, flipping the arrow “fixes” the sign on the wrong channel and can still leave energy allocation wrong. Match phase first, then polarity (Solis-style verification).

Next: 7. Handoff →

Address clashes & ticket handoff

RemoteMODULE 07 · ~6 min

Duplicate Modbus IDs and baud mismatches show up as multi-device drops. When remote is done, hand off a one-page field checklist.

If several devices vanish together after a new inverter is added, suspect address collision or serial settings drift — and prepare a one-by-one power-on test for the field.

Core ideas

Remote clues for ID / baud trouble

New device commissioned → older neighbors go intermittent.
Portal shows duplicate Modbus IDs or flickering discovery.
Works at night / low traffic, CRC errors rise with production (also shield/common — still list both on the checklist).

Field test plan to attach

Power only master + one slave; confirm unique ID; add devices one at a time.
All nodes: same baud/parity (often 9600 8N1 — confirm per brand).
Document final ID map in the ticket.

One-page handoff

Minimum fields

Time (with TZ):
Scope: plant / logger / inverter list
Alarms (exact text):
Last good data:
Remote steps already tried:
Hypothesis (ranked):
Field order:
  1) Safety lockout
  2) Power & LEDs
  3) RS-485 port / polarity / daisy-chain
  4) 120 Ω ends only · shield one end · common
  5) Unique Modbus IDs · matching serial settings
  6) CT / meter (named)
  7) Modem / antenna / Ethernet
  8) Strings / combiner / thermal
Contacts / cabinet access:

At the deskA handoff without “what we already tried” causes the field to repeat soft reboots and burn the visit.

How it breaks, and what to check

Symptom	Likely cause	What to check
Two devices share behavior oddly	duplicate ID	ID map; single-device power-on
Gibberish / all timeouts after swap	baud/parity mismatch	Every node same serial profile
Field repeats desk reboot	handoff missing	Checklist includes tried steps

Check yourself

Answer out loud first, then open the card.

How do you find a duplicate Modbus ID on site?

Leave only the master and one slave powered, confirm communication, then add devices one at a time until the clash reappears. Correct IDs before restoring the full bus.

Next: 8. Power & LEDs →

Part B · On-site

Qualified personnel. Safety lockout first. Walk the ladder — don’t skip to hard reset.

Power & LEDs

On-siteMODULE 08 · ~4 min

After lockout and restore procedures per site rules: logger, modem, and inverter power — breakers, adapters, and status LEDs.

Many “comms down” tickets die at a tripped breaker or a dead wall wart. Confirm power before you re-terminate an RS-485 bus.

Core ideas

Visual pass

Logger / gateway LEDs per vendor legend (power, link, activity).
Modem LEDs — power, network registration, signal.
Inverter display / LEDs — fault vs sleep vs producing.
AC breakers and DC disconnects in the expected state for the test.
Adapters and DIN supplies: seated, correct voltage, no burn marks.

SafetyFollow the site LOTO procedure. Do not energize with covers off unless your qualification and the procedure allow it. Measure only with rated tools.

In the fieldPhoto the LED state before changing cables. Compare to the vendor LED chart in the truck folder or phone.

How it breaks, and what to check

Symptom	Likely cause	What to check
No LEDs on logger	supply / breaker	Adapter, outlet, DIN fuse
Inverter dark, logger fine	inverter AC/DC	Local disconnects; not the modem
LEDs on, still offline in portal	WAN or bus	Continue to modules 9–12

Check yourself

Answer out loud first, then open the card.

Why photograph LEDs before unplugging anything?

You preserve the as-found state for the ticket and avoid losing a clue (e.g. modem registered but no PPP) when power is cycled.

Next: 9. RS-485 physical →

RS-485 physical layer

On-siteMODULE 09 · ~8 min

Correct COM port, connector click, A/B or D+/D− polarity, daisy-chain (not star), 120 Ω at both ends only, shield one end, common/GND when required.

Most intermittent CRC and “undetected device” tickets that survive remote reboot die here. SolarEdge’s RS-485 undetected guide and ABB’s RS-485 guidelines both hammer the same physical rules.

Core ideas

Connector and polarity

Use the documented COM / RS-485 port — not a random terminal block.
Hear/feel the connector click; tug-test gently.
A/B or D+/D− must be consistent on every node. Labels differ by brand — follow that device’s diagram, not a generic blog color code.
If the whole bus is dead and IDs/baud look right, try swapping the pair once as a controlled test and document it.

Topology, termination, shield, common

Daisy-chain end to end. Remove star taps and long stubs.
120 Ω termination at the two physical ends only — not on middle devices.
Shield grounded at one end (typically master/logger) to avoid ground loops.
Common / signal ground when the vendor requires it — missing common shows up as CRC/timeouts that worsen under high production / EMI.
Prefer shielded twisted pair rated for the run.

Worked example

CRC rises when inverters ramp

Night: bus looks fine. Midday: timeouts spike.
Find a tee splitter in a junction box → convert to daisy-chain.
Confirm only end devices have 120 Ω; middle inverter had an extra terminator — remove it.
Bond shield at logger only; add signal common per inverter manual.

In the fieldChange one physical factor at a time. Retest discovery after polarity, after termination, after shield — so you know which fix worked.

How it breaks, and what to check

Symptom	Likely cause	What to check
No devices detected	open cable / wrong port / polarity	Continuity, COM port, A/B swap test
CRC under high production	EMI, missing common, extra terminators	Shield, common, terminator count
Only far devices fail	stub / bad joint mid-chain	Walk the daisy-chain joints

Check yourself

Answer out loud first, then open the card.

Where do 120 Ω terminators belong?

Only at the two physical ends of the daisy-chain — not on every device and not on mid-span nodes.

Why ground the shield at one end only?

To drain noise without creating a ground loop between cabinets at different potentials.

Next: 10. Modbus IDs →

Modbus ID & serial settings

On-siteMODULE 10 · ~5 min

Unique slave IDs, matching baud/parity across the bus, and the one-by-one power-on method to find clashes.

Physical wiring can be perfect and the bus still fails if two slaves share an address or one node runs 19200 while the master runs 9600.

Core ideas

Rules that keep discovery clean

Every slave on one master segment needs a unique Modbus ID.
Baud, parity, and stop bits must match the master (commonly 9600 8N1 — verify the brand).
After changing an ID, power-cycle or apply per manual so the node advertises the new address.
Record the final map: ID ↔ serial ↔ role (inverter, meter, battery).

One-by-one power-on

Power master only; confirm it is healthy.
Add one slave; set/confirm ID; verify in logger/portal discovery.
Repeat until the bus is complete or the clash reappears.

In the fieldSchneider Conext and other Modbus notes emphasize consistent serial parameters — don’t mix “almost the same” baud rates on one cable.

How it breaks, and what to check

Symptom	Likely cause	What to check
Wrong inverter data on an ID	duplicate address	One-by-one power-on
No response any ID	baud/parity or wiring	Serial settings, then module 9
Works until device #N added	ID conflict on #N	Readdress #N; update map

Check yourself

Answer out loud first, then open the card.

Master 9600 8N1, one inverter 9600 8E1. Symptom?

That inverter (and possibly the bus) fails to communicate cleanly — parity must match on every node.

Next: 11. CT & meter →

CT & meter wiring

On-siteMODULE 11 · ~7 min

Polarity marks, phase match, CTs on the correct PCC conductors, meter phase order equal to inverter grid port order.

Bring the desk hypothesis (module 6) and prove it with the one-line diagram on the wall. Solis, Sungrow, and Enphase publish CT orientation guidance — use the brand in front of you.

Core ideas

Verification order

Phase association: CT on L1 feeds meter input L1 (etc.).
Polarity: arrow / H1 marking per vendor (toward grid or toward load — read the brief).
Location: for grid import/export, CTs at PCC before load/inverter branches.
Meter ↔ inverter: meter phase order matches inverter grid port order.
Never leave a CT secondary open under load — shorting rules per manufacturer.

After the fix

Compare portal import/export signs to a known production period.
Capture as-left photos of CT orientation and terminal labels.

In the fieldIf production is positive in the inverter UI but export is negative in the portal, treat it as instrumentation until proven otherwise — not as a “panel problem.”

How it breaks, and what to check

Symptom	Likely cause	What to check
Negative export	reversed CT or phase map	Phase first, then polarity
Meter fault alarm	open CT / wiring / power	Secondary continuity per procedure
Numbers disagree with utility	CT not at PCC	One-line vs installed location

Check yourself

Answer out loud first, then open the card.

Can you open a CT secondary while the primary is energized?

No — dangerous voltage can appear. Follow the manufacturer shorting / disconnect procedure. Qualified personnel only.

Next: 12. Close-out →

Modem, strings, thermal & close-out

On-siteMODULE 12 · ~8 min

Modem/antenna/SIM/Ethernet, combiner and string basics, thermal/fan checks, recommended field order, interview self-check, and manufacturer sources.

Finish the physical path the desk could not: restore WAN if needed, don’t ignore repeating fuses, and leave thermal issues for hard reset last — not first.

Core ideas

Modem / antenna / SIM / Ethernet

Correct modem type and seating/pins for that logger.
SIM fully seated; wait for registration when the carrier is slow.
Antenna connector tight; metal cabinets need an external antenna.
Chronically weak LTE → move to Ethernet when a drop exists (SolarEdge Ethernet guide is a good pattern).

Strings, combiner, thermal

Check Voc expectations, MC4 seating, fuse type/rating.
A fuse that blows again has a root cause (short, water, bad connector) — don’t just upsize.
Airflow, direct sun on the enclosure, fan cable/dirt; overtemp alarms are not fixed by portal reboot.
Hard reset last, after power/LED → cable → polarity → termination → IDs → CT → modem → strings.

Recommended field order

Safety lockout

OFF → DC drop → DC disconnect → AC breaker

Critical

Power & LEDs

Logger, modem, inverter supplies

Start

Cable & port

Correct COM, connector click

Start

Polarity

A/B or D+/D− consistent

Bus

Termination · shield · common

120 Ω ends only; shield one end

Bus

Unique ID + serial

One-by-one power-on if needed

Config

CT / meter

Phase then polarity; PCC

Meter

Modem path

SIM, antenna, Ethernet fallback

WAN

Strings / thermal

Voc, fuses, airflow — hard reset last

Last

Interview principle“I prove what I can from the portal first — scope, alarms, soft actions, network clues — then I hand the field a ranked checklist: safety, power, RS-485, IDs, CT, modem, strings.”

Interview self-check

Answer out loud first, then open the card.

Walk me through a plant-offline ticket without driving yet.

Scope (plant vs device), last good data vs night, exact alarms, peer site health, modem signal/SIM if shown, soft reboot once with notes, config/timezone/meter type, then a one-page field checklist if still down.

Where do RS-485 terminators go, and why do CRC errors rise at noon?

120 Ω at both ends only. Midday EMI/production noise exposes missing common, bad shield, stubs, or extra terminators.

Negative production while the inverter shows positive AC power — your move?

Instrumentation: phase map then CT polarity at the named meter; confirm PCC placement. Not a module replacement first.

CT secondary safety rule?

Never open under load; use manufacturer shorting procedure; qualified personnel only.

Sources

Vendor documents (verify current revision for your model). Huawei SmartLogger pages may be geo-blocked — check current vendor docs for your model.

Scope: This is a portfolio study guide for technical support interview prep, not a substitute for site electrical codes, LOTO policy, or manufacturer manuals. Always follow the diagram for the installed model.

Back to the roadmap ↑