String vs Microinverter vs Optimizer Retrofit Decision Matrix

Why this matters

When a residential system is being designed or retrofitted, the inverter topology choice locks in the next 10-25 years of performance, monitoring, and serviceability. String inverter, microinverter, and DC optimizer each have a defensible best-fit case and a defensible worst-fit case. Choosing string for a shaded multi-orientation roof loses production and frustrates the customer. Choosing micros for an unshaded south roof in a high-temperature attic shortens unit life with no production gain. The matrix below sets the rule, the exceptions, and the NEC compliance posture that comes with each.

Symptom presentation - the design question

Three real-world scenarios drive the choice:

• New build: blank slate, customer wants best long-term value.
• Retrofit on an existing array: string inverter failed, customer is deciding whether to replace in-kind or upgrade.
• Adding panels to an existing array: integrate to current architecture or change topology.

The decision depends on roof geometry, shade, climate, monitoring expectations, NEC rapid shutdown obligations, customer technical sophistication, and budget.

Quick checks before the matrix

1. Roof solar resource. Single-orientation south roof at low pitch with no shading is a string-friendly case. Multi-orientation, complex shade, or dormer obstructions push toward MLPE (module-level power electronics: micros or optimizers).
2. Shade profile by month, modeled in PVsyst or HelioScope. If any string sees more than 5 percent annual shading by edge or chimney, MLPE recovers most of the loss.
3. Attic temperature in summer. Hot attics (above 140 F) shorten microinverter electrolytic-capacitor life. String and optimizer chassis live in cooler garage or exterior locations.
4. NEC code year adopted by jurisdiction. 2017 NEC introduced 690.12 module-level rapid shutdown, 2020 NEC retained, 2023 NEC restated. Microinverters and optimizers comply natively; string-inverter-only designs without RSD-compliant components do not.
5. Customer monitoring expectation. Per-panel data drives many residential customer expectations; string-only systems give per-string data.

The three topologies

String inverter

Single (or dual) central inverter, panels wired in series strings, MPPT at the string level. Examples: SMA Sunny Boy, Fronius Primo, Solis 4G series.

Strengths:

• Lower upfront cost per watt at scale.
• Single device to inspect or replace.
• Lifecycle 10-15 years; inverter swap is straightforward.

Weaknesses:

• One shaded or soiled panel reduces output of the entire string.
• No per-panel diagnostics.
• Requires separate rapid shutdown initiator and module-level RSD-compliant device (per NEC 690.12-2017 and later) - typically Tigo TS4 or equivalent. This adds cost back.

Microinverter

Inverter mounted under each panel, AC trunk cable runs to combiner. Examples: Enphase IQ7+/IQ8/IQ8H, APsystems QS1, Hoymiles MS-A.

Strengths:

• Per-panel MPPT - shaded or soiled panel does not drag the rest.
• Per-panel monitoring via vendor app.
• NEC 690.12 rapid shutdown natively satisfied at module level.
• AC trunk simplifies array expansion later.
• Lifecycle 20-25 years per vendor; warranty 20-25 years.

Weaknesses:

• Upfront cost premium per watt.
• Roof penetrations include unit; service of any micro requires roof access.
• Temperature-derated electronics under hot attic; Enphase rates IQ8 at 65 C ambient max.

DC optimizer plus central inverter

Optimizer per panel, central inverter on the wall. Examples: SolarEdge HD-Wave / Genesis / Home Wave with P370/P401/P505 optimizers.

Strengths:

• Per-panel MPPT and monitoring.
• NEC 690.12 RSD natively satisfied at module level.
• Single high-voltage inverter on the wall is easier to service than micros on the roof.

Weaknesses:

• Two-vendor lock-in: optimizers and inverter must match.
• If the central inverter fails, the whole array is offline until replaced.
• SolarEdge has had specific reliability windows on certain models (HD-Wave Gen 1 capacitor batches); customer should know.

Decision drivers - matrix

Driver 1: Shade

• No shade, single orientation: string is acceptable.
• Light shade or multi-orientation: optimizer or microinverter.
• Heavy shade (10-plus percent annual): microinverter outperforms; optimizers are second.

Driver 2: Roof complexity

• Simple single-plane: string fine.
• Multi-plane or dormers: MLPE.
• Need to expand later: microinverter trunk is the easiest to extend.

Driver 3: Monitoring expectation

• Per-panel data: micros or optimizer.
• String-level data acceptable: string.

Driver 4: NEC rapid shutdown obligation

NEC 690.12 sets a PERFORMANCE requirement for systems on or attached to buildings, not a product requirement. Controlled conductors outside the array boundary (or more than 3 ft from the point of entry into a building) come down to 30 V within 30 seconds; conductors inside the boundary, or within 3 ft of the penetration, come down to 80 V in the same 30 seconds. Note which limit goes where, because it is commonly stated backwards.

There is more than one way to meet it, and the 2020 cycle added a second path:

• Micros - natively compliant (each module's output is already AC at the module).
• Optimizers - natively compliant when paired with their own inverter.
• String with no MLPE - add a module-level RSD initiator. Cost and complexity come back.
• A listed PV hazard control system (UL 3741) - the alternative the 2020 NEC introduced, which achieves the inside-boundary requirement at the system level instead of per module. Do not tell a customer module-level electronics are the only legal answer.

In jurisdictions on 2014 NEC or earlier (rare today), array-level RSD was acceptable; the inside-the-array-boundary limit that drives module-level shutdown came in with 2017 and has been the requirement since. Confirm AHJ code adoption year.

Driver 5: Attic and ambient temperature

• Phoenix, Las Vegas, attic above 140 F summer: derate microinverters per spec; string inverter mounted in shaded garage is the cooler option. Optimizers are roof-mounted but lower-power devices and tolerate heat better than full inverters.
• Mild climates: any topology.

Driver 6: Customer technical sophistication

• Tech-savvy customer who wants the Enphase app: micros.
• Customer who wants set-and-forget: any, with monitoring tied to a service contract.

Driver 7: Retrofit specifics

• Replacing a failed string inverter under warranty: replace in kind.
• Replacing a failed string inverter out of warranty on an aging system: consider optimizer or micros if the array could go another 10-15 years and the customer wants better monitoring.
• Adding panels to an Enphase trunk: stay with Enphase, mix only within the compatibility matrix.
• Adding panels to a SolarEdge: stay with SolarEdge optimizers and the inverter's input voltage envelope.

Driver 8: Battery future

• Customer wants battery now or later: confirm topology compatibility. Enphase systems pair with Enphase IQ Batteries. SolarEdge systems pair with the Energy Hub plus LG or BYD batteries. String-only systems usually need an AC-coupled battery system (Tesla Powerwall) for retrofit.

Decision rules - quick call

Run the drivers in this order and stop at the first one that answers. They are not equal weight.

1. Existing topology on a retrofit or addition wins unless there is a reason to break it. Matching what is on the roof keeps one monitoring platform, one warranty channel, and one set of spare parts. Break it only when the array is being substantially rebuilt or the existing platform cannot accept the addition.
2. Shade and orientation decide the new-build call. More than one plane, or measurable annual shading on any string, and it is MLPE. Do not try to engineer around real shade with string sizing.
3. Rapid shutdown obligation is a floor, not a tiebreaker. If the AHJ is on a code year requiring module-level shutdown, string-only is not a candidate until you have added compliant module-level devices, and once you have added them the cost gap that made string attractive is mostly gone.
4. Ambient temperature is the argument against micros. A roof-mounted full inverter in a genuinely hot climate is a shorter-lived device than the same electronics in a shaded garage. If the site is hot and the roof is simple, string or optimizer.
5. Battery intent is the argument for picking a family early. Decide the battery family before you pick the inverter, not after. Retrofitting a battery onto the wrong family means AC coupling, a second app, and a more expensive install.
6. Everything else is preference, and preference goes to the customer once you have told them the tradeoff plainly.

Quick calls that cover most residential work:

• Simple south roof, no shade, mild climate, no battery plans: string, with whatever module-level shutdown the code year requires.
• Complex roof, several planes, partial shade, customer wants per-panel data: microinverter.
• Mostly simple roof, some shade, hot climate, customer wants per-panel data: optimizer plus central inverter.
• Failed inverter still under warranty: replace in kind, full stop. Changing topology voids the argument you are trying to win.
• Failed inverter, out of warranty, array with a decade of life left: this is the one genuine open call. Price both the in-kind swap and the MLPE upgrade and let the customer choose against the monitoring and serviceability difference.
• Adding a handful of panels: match the existing platform and verify the addition against the compatibility matrix and the inverter's input envelope before you quote it.

The failure mode to avoid is quoting a topology because it is what your crew installs fastest. Every one of these three is the right answer somewhere, and the roof, the code year, and the customer's plans tell you which.

References

• NEC 690.12 Rapid Shutdown of PV Systems on Buildings (2017, 2020, 2023 editions).
• NEC 690.7 Maximum Voltage and 690.8 Circuit Sizing.
• UL 1741 Inverters, Converters, Controllers and Interconnection System Equipment.
• Enphase IQ8 Series Installer Manual.
• SolarEdge Home Wave Series Installer Manual.