The battery is the part of the system you will replace
Panels routinely outlive their warranties by years and fail slowly. Inverters are electronics and either work or need service. The battery is the consumable: it has a finite number of cycles, it ages whether or not you use it, and how you treat it in the first year sets how long you keep it. Nearly every account of an off-grid system that disappointed its owner is, underneath, a battery story.
That makes the battery decision the one worth slowing down for. It is also the one where the market has changed most: lithium iron phosphate has moved from exotic to ordinary, and the arithmetic that made lead-acid the default for decades no longer holds in most new builds.
Chemistry, in the terms that actually matter
Lead-acid batteries, whether flooded, AGM, or gel, are cheap per unit of nameplate capacity and expensive per unit of usable energy, because you should not routinely take them far below half charge without shortening their life sharply. Flooded cells need watering, ventilation, and equalization, which is real recurring maintenance done on a schedule rather than when convenient.
Lithium iron phosphate costs more up front, delivers far more of its nameplate capacity usefully, tolerates deeper and more frequent cycling, needs no watering, and generally lasts more cycles. Its specific weakness is cold: charging a lithium battery below freezing damages it, which is why cells for cold climates should be heated, kept in conditioned space, or protected by a battery management system that simply refuses to charge until the cells are warm enough.
Compare on cost per usable kilowatt-hour over expected cycles rather than on sticker price, and include the maintenance you will actually do rather than the maintenance the manual assumes. The battery calculator on this site makes that comparison with your own numbers, and the battery spec table lists the published figures for common models so you are comparing like with like.
Sizing: days of autonomy, honestly
Autonomy is how long the house runs on stored energy alone with no meaningful production. Two to three days is a common design target, but the right number is a function of your climate and your tolerance. A run of dark, still, snowy days is normal in some regions and rare in others, and a site with a generator can carry less battery than a site without one.
Do not size on nameplate capacity. Size on usable capacity, which is nameplate multiplied by the depth of discharge that chemistry tolerates routinely, then reduced again for the fact that a cold battery delivers less than a warm one. A bank that looks generous on paper at room temperature can be noticeably smaller in an unheated shed in January.
Resist the urge to solve every problem with more battery. Battery stores energy; it does not make it. If the array cannot refill the bank in your worst month, a larger battery just gives you a longer, slower descent into a dead bank, and the fix is panels, a generator, or a smaller load list.
Choosing an inverter without buying twice
The inverter converts battery direct current into the alternating current your house uses, and it is sized by power rather than by energy. It needs to carry your largest realistic simultaneous load, and it needs a surge rating that can start motors: well pumps, compressors, and power tools all draw several times their running figure for a moment.
Insist on a true sine wave output. Modified sine wave is cheaper and is hard on motors, some electronics, and many medical devices. This is not a place to economize in a house you live in.
Idle consumption matters more than the datasheet makes it look, because the inverter is on all the time. A unit that idles at a few tens of watts consumes a meaningful share of a small system's daily energy doing nothing at all. Look for a low no-load draw or a search mode that sleeps until something asks for power.
Decide early whether you want a hybrid or all-in-one unit that combines inverter, charge controller, and generator input in one box, or separate components. Integrated units are simpler to wire and cheaper; separate components are easier to replace one at a time and let you mix the best of each. Both are legitimate. What is not legitimate is discovering after installation that your inverter cannot accept a generator, or cannot be paired with a second unit later.
Charge control and the settings nobody changes
An MPPT charge controller converts the array's higher voltage down to battery voltage while extracting more of the available power than a simple switching controller, and the advantage is largest in cold weather and in winter, when panel voltage is high and you need every watt. For anything but a very small system it is the default choice.
The controller must be configured for your battery. Absorption and float voltages, temperature compensation, and charge current limits are all chemistry-specific, and factory defaults are usually set for lead-acid. A lithium bank run on lead-acid defaults is a common and quietly damaging mistake. Set them from the battery manufacturer's own documentation and write the settings in your systems binder.
Check them again after any firmware update or component swap. Controllers sometimes return to defaults, and nothing announces it.
System voltage and the wiring that follows from it
Higher system voltage means lower current for the same power, and lower current means thinner, cheaper, less lossy wire and smaller fuses. Small systems at twelve volts are simple and use common parts. Larger houses generally settle at forty-eight volts, and the wiring savings alone often justify it.
Choose the voltage before you buy anything, because it constrains every other component. Changing it later means replacing the inverter, the controller, and often the battery. This is one of the few decisions in an off-grid build that is genuinely hard to reverse.
Habits that decide how long the bank lasts
Keep the battery in a temperature range it likes, which usually means insulated, ventilated where the chemistry requires it, and not in an unheated outbuilding in a cold climate. Avoid leaving lead-acid partially discharged for long periods. Avoid charging lithium below freezing. Let the controller finish a full charge regularly rather than leaving the bank permanently part-full.
Watch state of charge rather than voltage alone, since voltage under load tells you very little. A shunt-based monitor is inexpensive next to the battery it protects and turns guesswork into a number.
Log a reading once a week for the first year. It takes a minute, and it means that when something changes you will know when it started, which is most of a diagnosis.
Expanding later without stranding what you own
Most systems grow. The way to make growth cheap is to decide the system voltage once, choose components rated with headroom, and avoid designs that lock you into a single supplier's ecosystem for parts that may not be available in ten years.
Mixing old and new batteries is the classic mistake. Cells of different ages and states of health placed in the same bank tend to be dragged toward the behavior of the weakest, which means an expensive new addition can be spoiled by what it was added to. Plan capacity for where you expect to be in a few years, or plan to replace the bank as a unit.
Leave physical room, spare breaker positions, and conduit capacity for the second array or the second inverter. Space and a length of conduit are cheap during the build and expensive afterwards.
