Skip to content
WattCurious

String Inverters vs Microinverters vs Power Optimizers

Updated 21 September 20267 min readSolar Energy

All three architectures do the same two jobs: find the operating point that extracts the most power from the array, and turn direct current into alternating current. What separates them is where those jobs happen. Put one tracker in charge of twenty modules and they rise and fall together; give each module its own and they become independent. Every difference that matters — shading behaviour, layout freedom, monitoring detail, what fails and how you reach it — falls out of that single choice.

Key takeaways

  • Every architecture performs the same two functions; they differ only in where tracking happens and where conversion happens.
  • A string inverter tracks a whole string at once, so modules in that string are only as productive as their common operating point allows.
  • Microinverters convert on the roof, which removes high-voltage DC from the system entirely and makes every module independent.
  • Optimizers keep central conversion but move tracking to the module, which is the shading benefit without the roof-mounted conversion.
  • The real trade is not performance but where the electronics live: one accessible box, or many units on a roof for the life of the system.
On this page
  1. The two jobs, and where they happen
  2. One tracker for twenty modules
  3. What moving the electronics costs
  4. Where the differences actually show up
  5. Choosing by the roof, not the technology

The two jobs, and where they happen

Strip away the product categories and there are exactly two functions to place. Maximum power point tracking chooses the voltage at which the array delivers the most power, continuously, as light and temperature change. Inversion turns the resulting direct current into alternating current at grid frequency.

Three architectures, and where each puts the two jobsThree single-line diagrams. In the string architecture, modules are wired in series and one ground-mounted inverter performs both maximum power point tracking and inversion for the whole string. In the microinverter architecture, each module has its own small inverter beneath it performing both functions, and the modules are joined in parallel on the alternating-current side. In the optimizer architecture, each module has a small direct-current unit performing tracking only, the modules remain in series, and one central inverter performs the inversion.DCACString inverterinvertertracking + inversionto the gridone tracker for the whole string · high-voltage DC comes down from the roofMicroinvertersµinvµinvµinvto the gridtracking and inversion at every module · what leaves the roof is ACPower optimizersoptoptoptinverterinversion onlyto the gridtracking per module, inversion once at the bottom · modules stay in seriesTwo jobs:find the operating pointthat extracts the mostpower, and turn DCinto AC.
The same two jobs, placed differently. Tracking per string or per module; conversion once at the bottom or once per module on the roof.

A string inverter does both, once, for a whole string. A microinverter does both, per module, on the roof. A power optimizer splits them: tracking per module, conversion once at the bottom. Everything that follows is a consequence of those placements.

One tracker for twenty modules

The string inverter's limitation is not its efficiency, which is excellent. It is that a single operating point has to suit every module on its input.

One operating point for every module, or one eachSix modules on one tracker, five identical and one weakened by shade, soiling or age. The tracker must choose a single voltage for all of them: set to suit the five, the weak module holds the string current down; set to suit the weak one, the five deliver well below their capability. Alongside, the same six modules each with their own tracker, each sitting at its own best point, so the weak one is reduced and the other five are unaffected.One tracker, six modulesSix trackers, six modulesthe weak onewhat each deliversOne operating point has to suit all six.Chosen for the five, the weak module holds thestring current down. Chosen for the weak one,the five deliver far below their capability.its own tracker eachEach sits at its own best point.The weak module is reduced, and only the weakmodule. The other five never notice it.On a clean array facing one direction, the modules are near enough identical and one point serves them all.Mismatch is what creates the benefit: shade, several roof planes, different tilts, different ages, or one module ageing faster.
One operating point has to serve every module on the input. Give each module its own and the weak one stops being everyone's problem.

On a clean, unshaded array facing one direction, the modules are near enough identical and one operating point serves them all. Nothing is lost, and this is why string inverters remain entirely sensible for straightforward roofs.

The picture changes as soon as modules stop matching. Partial shade is the obvious cause and the one examined in how partial shade affects panels, but it is not the only one: different roof planes, different tilts, modules of different ages, or one module that has degraded faster than its neighbours all produce the same mismatch. Many string inverters hedge by providing two or more independent trackers, which handles two roof planes cleanly and a complicated roof not at all.

What moving the electronics costs

Module-level electronics fix mismatch by removing the shared operating point. The question is what comes with them.

Where the electronics sit, and what leaves the roofThree sections through the same building. With a string inverter, one unit sits at ground level within reach, and high-voltage direct-current cabling runs down from the roof. With microinverters, one unit sits under each module on the roof and alternating-current cabling runs down. With optimizers, one small unit sits under each module plus one inverter at ground level, and the direct-current cabling drops to a safe voltage when the inverter stops. Each is annotated with how many units are on the roof and what kind of current leaves it.StringMicroinvertersOptimizersinverterNothing on the roofOne box at eye level, reachable.High-voltage DC comes down the wall.One unit per moduleConversion happens on the roof,so AC comes down — no high-voltage DC.inverterUnits per module, plus a boxThe hardest electrical work stays atground level; DC drops when it stops.This is the structural difference, and the one that lasts.A single accessible box, or a unit under every module in the hottest and least reachable place on the building, for the life ofthe system. Heat is what ages electronics, and a roof is where the heat is.
The structural difference, and the one that lasts. A single box at eye level, or a unit per module on a roof, for the life of the system.

Microinverters put a conversion stage under every module. That removes high-voltage DC from the installation entirely — what leaves the roof is alternating current — and it makes each module fully independent. It also places electronics in the hottest, least accessible place on the building, in numbers, and heat is what ages electronics.

Optimizers take the middle path: a small DC unit per module that tracks and conditions, with a single inverter doing the conversion somewhere reachable. There is still a unit per module on the roof, but the component that does the hardest electrical work is not up there.

Two failure patterns across the life of a systemTwo timelines compared. The string architecture shows a long stretch of uninterrupted operation broken by a single large event: output falls to zero, the whole system stops, and one accessible unit is replaced at ground level. The module-level architecture shows several small events spread across the years: each costs one module's output while the rest of the array continues, and each requires roof access and lifting a module to reach the failed unit.One large failure, or several small onesString inverterfull outputeverything stopsand resumes, once replacednoticed the same day · one unit · replaced from the groundModule-levelfull outputone unitanotheranotherone replacedthe array carries on · a few percent each · roof access to reach any of themNeither pattern is better in the abstract.A string failure is loud and total, so it is noticed the same day and fixed from the ground. A module-level failure is quiet andpartial — nothing announces it unless someone watches the per-module data the architecture exists to provide.
One large, visible, reachable failure against several small ones that are none of those things. Which is preferable depends on who will still be maintaining the system in year twelve.

Neither pattern is better in the abstract. A string failure is loud and total, which means it is noticed the same day and fixed from the ground. A module-level failure is quiet and partial — the array carries on, the shortfall is a few percent, and nothing announces itself unless someone is watching the per-module data that the architecture exists to provide.

Where the differences actually show up

Set side by side, the trade is clearer than any general claim about efficiency.

How the three architectures differ, and why
String inverterMicroinverterOptimizer
TrackingOne per stringPer modulePer module
ConversionOnce, at the inverterPer module, on the roofOnce, at the inverter
Mismatched modulesAll share one operating pointIndependentIndependent
Mixed orientationsLimited by tracker countUnrestrictedUnrestricted
Cable from the roofHigh-voltage DCAC at ordinary voltageDC, dropped when the inverter stops
Monitoring detailPer stringPer modulePer module
Single failureWhole system stopsOne module stopsOne module, or the whole system
Reaching a failureOne accessible boxLift a moduleLift a module, or the box
Units on the roofNoneOne per moduleOne per module

Structural characteristics, not a recommendation, and not a comparison of particular products. Any of the three can be the right answer; which one depends on the roof, not on the technology.

Two rows carry most of the decision. Monitoring detail is what makes diagnosis fast: per-string data narrows a fault to a group, per-module data names it, and the difference is a glance against an afternoon of elimination — the sequence described in why a system produces less than expected.

Reaching a failure is what decides how a system ages. A string inverter fails as a single, visible, replaceable event. Module-level electronics fail one at a time, each costing little, each requiring roof access — a slow trickle of small jobs instead of one large one.

Choosing by the roof, not the technology

The honest summary is that the array decides, not the architecture.

What the roof demands, rather than what the technology offersThree roof situations. A single unobstructed plane with uniform modules, where one tracker serves everything and a string inverter is sufficient. A roof with a chimney and a neighbouring tree casting moving shade, where module-level tracking recovers output a single tracker cannot. A building with modules on three planes at different angles, where independent tracking is necessary for the array to behave at all. Each is annotated with what the site demands rather than with a product.The array decides, not the architectureOne plane, uniform, unshadedNothing differs between modules,so module-level tracking hasnothing to recover.A string inverter is sufficient —the simplicity is an advantage,not a compromise.A chimney, and moving shadeModules stop matching each otherfor part of every day, and theshadow moves between them.Module-level tracking recoversoutput a single shared operatingpoint cannot.Three planes, three anglesEach plane peaks at a differenthour, so no one operating pointsuits them all.Independent tracking is not arefinement here — it is whatmakes the array work.The question is whether the modules differ from one anotherWhere they do not, module-level tracking has nothing to recover. Where they do, it is the whole point.The middle cases — one chimney, or two planes at similar angles — are usually settled by how many independent trackers thestring inverter has, which is why orientation should be decided before the inverter is chosen rather than after.
The question is whether the modules differ from one another. Where they do not, module-level tracking has nothing to recover; where they do, it is the whole point.

One clear plane, uniform modules, no shade: a string inverter does the job, and the simplicity is a genuine advantage rather than a compromise. Moving shade, several planes, or an array that will be extended later with different modules: module-level tracking is not a refinement but the thing that makes the array work.

The middle cases are the awkward ones — a roof with a single chimney, or two planes at similar angles — and they are usually settled by the tracker count on the string inverter rather than by changing architecture. Two independent trackers handle two planes cleanly, which is why orientation and tilt should be settled before the inverter is chosen rather than after.

And between them sits the question nobody asks until year twelve — how the system will be maintained, by whom, and whether anyone will want to be on that roof to do it. That is the same calculation that decides whether a system connects to the grid at all: not which technology is best, but which one suits a building that has to live with it.

Frequently asked questions

Which one produces the most energy?

On an unshaded array with one orientation, the difference is small and the architectures perform much alike. Module-level tracking earns its place where modules differ from one another — partial shade, several roof planes, mismatched ages or a module that has degraded faster than its neighbours. Where nothing differs, there is little for it to recover.

Do microinverters really remove high-voltage DC?

Yes, and it is their clearest structural advantage. Conversion happens at each module, so what runs from the roof is alternating current at ordinary voltage rather than a string of modules in series at several hundred volts. Optimizers reduce the DC risk differently, by dropping the string to a safe voltage when the inverter stops.

What happens when one unit fails?

With a string inverter, one failure stops the whole system, and it is a single accessible replacement. With module-level electronics, one failure costs one module's output and the rest carries on — but reaching it means going onto the roof and lifting a module. The first is a bigger outage that is easy to fix; the second is a smaller one that is harder to reach.

Is module-level monitoring worth it?

It is the difference between knowing the system is down and knowing which module is down. For diagnosis it is genuinely valuable: a per-module view turns a week of elimination into a glance. Whether that justifies the extra hardware depends on how complicated the array is and how likely anyone is to look.

Can I mix orientations on one string inverter?

Only within limits, because one tracker has to find a single operating point for everything on that input. Many string inverters have two or more independent trackers precisely so that different roof planes can be kept separate. Beyond that, module-level electronics are what make genuinely mixed layouts behave.

Sources

Named organisations whose published material underpins this article. Where no link is given, the source is named rather than linked.

Editorial Team

Research, drafting and review

Articles are drafted from primary engineering and physics references with AI-assisted tools, then reviewed and fact-checked line by line by a human editor before publication. We publish explanations, not recommendations: no products, no pricing, no country-specific rules, and no invented author personas.

Last reviewed 21 September 2026. How we research and review