AC-Coupled vs DC-Coupled Batteries: Where Storage Joins

A battery can join a photovoltaic system at two places, and the place it joins determines almost everything else. On the DC side it shares the array’s direct-current bus with the solar charge controller and the inverter. On the AC side it sits behind its own inverter, connected to the same alternating-current wiring as the solar inverter and the building. Neither arrangement is newer or better in general. Each satisfies a different set of constraints, and the arithmetic of conversion steps, existing equipment and backup requirements decides which one a given roof can use.

The two topologies

A DC-coupled system puts the array, the battery and the inverter on one DC bus. The array’s charge controller regulates current into the battery, and a single inverter converts the combined DC to AC for the building. When the sun is up and the battery is charging, the energy goes through one conversion. When the battery discharges in the evening, it goes through one conversion. The inverter is sized to carry both the array and the battery, and it is the one piece of equipment that must be replaced if either grows.

An AC-coupled system leaves the solar inverter exactly where it is. The battery arrives with its own inverter, often called a hybrid or storage inverter, and connects to the AC wiring. Solar production flows through the existing inverter to the building, and any surplus can be routed through the battery inverter in reverse to charge the battery. That routing costs an extra conversion in each direction: DC to AC at the solar inverter, then AC back to DC at the battery inverter. Round-trip efficiency for the stored energy is therefore the product of two inverter efficiencies and the battery’s own round-trip figure, not one inverter’s figure.

Conversion steps and what they cost

Silicon cells produce roughly half a volt each, and a string of sixty or seventy cells produces the several hundred volts a grid-tie inverter expects. Conversion efficiency at each stage is typically in the mid-to-high nineties for modern equipment, and the losses multiply rather than add. A DC-coupled charge and discharge path might see one inverter conversion and the battery’s internal losses. An AC-coupled path sees two inverter conversions plus the battery’s losses. The difference is a few percentage points of the energy that passes through storage, which matters most in a system that cycles daily and least in one that is held for outages.

The same arithmetic applies to charging from clipped DC. When an array is oversized relative to its inverter, the inverter limits output at its rated AC power and the array’s excess DC is lost. A DC-coupled battery on the same bus can absorb that excess before the inverter’s limit is reached, because the charge controller and the battery sit upstream of the AC conversion. An AC-coupled battery cannot recover clipped energy, because by the time the energy reaches the AC side the inverter has already discarded it. Whether clipping exists at all depends on the array-to-inverter ratio, which is a design choice made when the solar system was built.

Retrofitting storage to an existing array

Adding a battery to a working grid-tied array is the case where AC coupling usually wins, and for a structural reason rather than a preference. The existing solar inverter is listed, warranted and interconnected with the utility. Replacing it with a DC-coupled hybrid inverter means removing working equipment, revisiting the interconnection agreement and often re-permitting the system. An AC-coupled battery leaves the solar inverter in place and adds a second inverter alongside it. The existing array keeps producing through its own equipment, and the new equipment carries its own listing and its own interconnection application.

This is also where the local rules matter most. Interconnection rules for adding storage are set by the state and by the utility, and they change. Some jurisdictions treat a storage addition as a modification to an existing interconnection and require a new application; others have a separate fast-track process for storage below a certain size; some require visible disconnect hardware or limit how much stored energy may be exported. None of this is general. The reader has to check the utility’s interconnection tariff, the state’s net-metering or export-compensation rules, and the local permitting authority’s requirements for the specific address before choosing a topology. A design that is straightforward in one service territory can require a new application in the next.

Backup behaviour and frequency-shift control

During an outage, a grid-tie inverter must stop exporting, because exporting onto a de-energised utility line endangers line workers. The standard mechanism is frequency-shift control: the battery inverter forms a local grid at a nominal frequency, and when the battery is full and solar production exceeds the building’s load, the battery inverter raises the frequency slightly. The solar inverter detects the rise and reduces its output, then shuts down if the frequency continues up. This is how an AC-coupled system keeps a grid-tie inverter producing during an outage without a second transfer switch.

A DC-coupled system does not need frequency-shift control for its own array, because the array is on the same DC bus as the battery and the charge controller can simply stop charging. The inverter forms the local grid and the array’s output is managed on the DC side. Both arrangements can back up a building; the difference is which piece of equipment enforces the limit and whether the existing solar inverter is capable of responding to frequency-shift commands. Older grid-tie inverters may not support it, and that capability has to be confirmed from the inverter’s documentation before an AC-coupled battery is specified for backup.

DC-coupled and AC-coupled batteries side by side

What people get wrong

The common misdiagnosis is to treat AC coupling as the retrofit option and DC coupling as the new-build option, as though the choice followed from the age of the array. It is a natural mistake because the two cases usually line up that way in practice, and because retrofit projects are where the existing inverter’s listing and interconnection status dominate the decision. But the topology follows from the constraint, not the calendar. A new build with a backup requirement and a solar inverter that does not support frequency-shift control may need DC coupling regardless of age. A retrofit where the existing inverter is being replaced for other reasons may end up DC-coupled.

The second misdiagnosis is to compare round-trip efficiencies from two datasheets and conclude that the higher number wins. The datasheet figure is measured at the battery terminals or at the inverter’s DC input, not at the point where the building’s loads are served. In an AC-coupled system the stored energy passes through two inverters, so the system-level round trip is lower than either component’s figure. Comparing a battery’s round-trip efficiency to a hybrid inverter’s efficiency is comparing different boundaries.

The third is to assume that a DC-coupled battery will always capture clipped energy. It will only do so if the charge controller and battery are sized to absorb the excess and if the battery has room to take it. A full battery captures nothing, and a charge controller rated below the array’s maximum current will limit the array before the inverter does. The margin that designs get wrong is the cold-morning case: array voltage rises as cell temperature falls, and a string sized for a warm afternoon can exceed the charge controller’s maximum input voltage on a clear winter morning. The temperature coefficient of open-circuit voltage, usually quoted per degree Celsius, is the figure that governs this, and it has to be applied to the coldest expected cell temperature at the site, not to standard test conditions.

Checking the design against the specification

The specification the decision has to satisfy is not “AC or DC” but a list: the existing inverter’s model and its frequency-shift capability, the array-to-inverter ratio and whether clipping occurs, the backup loads and their daily energy, the coldest expected cell temperature at the site, the charge controller’s maximum input voltage and current, and the utility’s interconnection process for storage. Each of these is a number or a document, and each can be checked before equipment is ordered.

The commands below read the documents a design review depends on. They are run on a laptop against downloaded files, not on any energised equipment.

pdftotext inverter-manual.pdf - | grep -i "frequency"

This extracts the text of the inverter manual and searches for frequency references. Output lines mentioning frequency-watt or frequency-shift response indicate the inverter can be controlled by an AC-coupled battery inverter during an outage. No output means the capability is not documented and must be confirmed with the manufacturer.

pdftotext inverter-datasheet.pdf - | grep -iE "max.*(input|voltage|current)"

This lists the inverter’s or charge controller’s maximum input voltage and current. The voltage figure is the one to compare against the cold-morning string voltage, calculated from the site’s record low temperature rather than the 25 degrees Celsius of standard test conditions. The utility’s storage and interconnection requirements are read from its own tariff and interconnection documents; whether adding a battery counts as a modification or a new application is the utility’s call and differs between utilities.

Where the work stops

A photovoltaic array produces dangerous DC voltage whenever there is light on it and cannot be switched off at the panel. Nothing in this page is a procedure for opening a combiner, probing a live string, disconnecting an energised connector or working on the DC side. Those tasks belong to licensed electricians, and roof work, mains connection and meter changes are licensed work in most jurisdictions. The design decision described here is made from documents and arithmetic before any of that work begins.