How we judge and rank

An order that nobody can argue with is an order nobody should trust. This page sets out what ours is built from, so that it can be argued with.

The one rule the rest hang from

Every family here is assembled the same way: read what the maker publishes, set it beside the document that defines the figure, listen to what owners consistently report after a year, and attach a source to every number that appears.

That last part carries most of the weight. A sourced number invites correction; an unsourced one asks for belief. Where a manual contradicts a product listing, the manual is what gets printed. Where nothing credible exists, the field stays empty and the page says it is empty — because a plausible substitute is indistinguishable from a real figure once it is typeset.

Six published things, asked in a fixed order

Three worries drive every purchase on this shelf: will it work with what is already installed, will it do the job asked of it, and will it still be serviceable in five years. Six published facts answer them, and they are read in this sequence:

• The figure that decides compatibility before anything else. The voltage window and current limit a solar input accepts, the connector a panel lead ends in, the chemistry profiles a charger can be set to, the cross-section a terminal will close on, the battery models an expansion unit is declared to fit. It comes first because it is the commonest reason one of these products is sent back, and because no photograph settles it.
• The headline figure, carrying the condition that gives it meaning. Capacity marked nominal or usable and never blurred between the two; a surge rating printed with the duration it holds; a cycle count written with the depth tested and the capacity still expected at the end; peak panel watts kept separate from what a roof delivers; a runtime quoted with the load it was measured at. Where a maker prints the headline and drops the condition, this site records the omission instead of passing it on.
• The detail that decides whether it is still in service in five years. Whether the connectors and terminals follow something a third party also makes, whether the monitoring works without an account, whether a firmware update can be refused, whether cells, fuses and films are sold as parts. Each family has one such detail, and that is the one written out.
• What the box omits, and what the omission costs. The fuse and the cable a converter needs before it can be switched on, the shunt that makes a bank legible, the bracket a panel needs on a railing, the adapter between one connector family and another. This is weighted deliberately heavily: a product that needs a second purchase to function was never the price on the label.
• What it costs to keep alive. Whether a consumable follows an open specification or a proprietary one, whether the maker sells spares at all, whether a battery can be replaced or only the whole unit, whether the owner is tied to one supplier. Products that lose on the headline figure regularly finish above stronger ones on this line alone.
• Whether it can be bought where the reader lives. Every reference is checked store by store, and each page names the markets where an offer was confirmed. An excellent product nobody stocks in your country is not a recommendation.

Three sources, weighted unequally and always named

Manufacturer documentation comes first for anything structural — dimensions, connectors, voltage windows, the batteries an accessory declares compatibility with — and last for anything printed as a headline, because a headline is written to sell. Published ratings and markings come second: a rated current, a direct-current voltage rating, a breaking capacity, an ingress code, a cycle figure at a stated depth. They are treated as figures to confirm against the current listing rather than as permanent properties, because a maker can change a cell supplier without changing a product name.

Third, and irreplaceable for one question: owners a year or two in. No specification sheet answers what a pack is like after its first cold winter, what a firmware update did to a charging profile, or whether the support address still replies. Where a judgement rests mainly on one of these three, the page says which. Where none of them answers, the page says that too, rather than reasoning its way to a number.

No marks out of ten, only a position with a reason

There is no score anywhere on this site. A single number would have to weigh a large home battery against a bag of connectors, and there is no honest exchange rate between them. What gets published instead is a position in a family, with the sentence that put it there written beside it.

Each hub runs from the modest to the substantial rather than from the most expensive downwards. Nearly everyone arrives with a figure in mind and a job to do; a page that opens with the largest product on the shelf is answering a question almost nobody asked.

What pulls a site like this off course

Commission income pulls towards the expensive, the newly launched and the easy to describe. What pushes back here is structural rather than a promise: not one family has been reordered because of what a reference earns, a number of pages finish by recommending thicker cable or a proper fuse instead of a larger machine, and no position has ever been traded for anything at all.

One distortion belongs specifically to this subject, and naming it works better than resisting it quietly. Large numbers sell energy products. A page organised around the biggest figure ends up recommending whichever station holds the most watt-hours or whichever inverter claims the highest peak, without mentioning that those watt-hours are nominal or that the peak is held for a moment. No page here is built that way. When the honest conclusion is that an upgrade changes nothing — a larger bank behind a cable that is already the bottleneck, a bigger inverter for a load the current one carries, another panel for a roof shaded at midday — the page reaches that conclusion.

One bias will not go away: we cover what is actually purchasable through the Amazon stores of the United States, the United Kingdom, Germany, France, Italy and Spain. Equipment sold only direct from its maker, or through specialist installers, is therefore under-represented here. Where that shapes a recommendation, the page says so.

Figures are reported from the maker, never recomputed here

This shelf is sold on numbers, and numbers are the easiest thing in a listing to repeat without thinking. One rule governs every page that carries them: every capacity in watt-hours, continuous and surge output, cycle count, depth of discharge, peak panel wattage, open-circuit voltage, charge current, rated current, breaking capacity, cable cross-section, ingress rating, efficiency and temperature threshold on this site is taken from the manufacturer’s own documentation, in the unit it was published in, with its source named — and none of it is recalculated, converted or inferred here. Where a maker publishes nothing, the page records that nothing is published rather than supplying a stand-in.

Capacity is where this earns its keep. A battery takes its name from the watt-hours its cells hold; the watt-hours that reach a socket are a smaller figure, printed further down the sheet or nowhere at all. This site writes which of the two a maker published, keeps them apart, and never lets the larger stand in for the smaller. Nor is any product here called faster, tougher or more capable than another on the strength of a specification sheet by itself.

What a printed number actually promises

One sentence sits beneath all of this, and it is why these pages read more flatly than the ones beside them in a search result: a published figure describes equipment; it does not promise an outcome. A station quoting a very high peak is describing what its inverter tolerates for an instant, not the load it will carry through an evening — which depends on the bank, the temperature and the cable. A panel quoting peak watts is reporting a laboratory measurement under defined conditions that do not occur outdoors. A cycle count is a test result carrying two assumptions. Treating any of them as a promise sells more equipment and leaves people disappointed by a product that did precisely what its sheet said it would.

Two things this site will not write

Two limits are fixed, and both read better stated than left to be inferred. No fuse, breaker, isolator, earthing arrangement or setting is described here as making an installation safe. A page reports what the maker specifies for its own product — a rated current, a direct-current voltage rating, a breaking capacity, an ingress code — and hands the rest to the maker’s instructions and the rules in force where you live, because whether an installation is safe depends on the cable, the enclosure and the person doing the work, and no listing knows any of those. And nominal capacity and usable capacity are two figures that are never merged here: one is what the cells contain, the other is what arrives at the socket, and an order built on a mixture of the two would be comparing products that are not comparable.

Where a page hands over

None of this substitutes for the manufacturer’s instructions. For lithium cells and their charging limits, for petrol generators and their exhaust, for anything wired into a building or a vehicle, and for anything that meets the mains, the page points to those instructions — and to the local regulations — instead of paraphrasing them, even though a page that stops there earns nothing at all.

One limit is repeated wherever it applies rather than assumed once: compatibility is decided by a number or a list the maker published, and never by appearance. The models a maker names decide which expansion battery attaches; the connector type decides which panel plugs in; the voltage marked on a controller decides which bank it was built for; the current rating on a socket decides what it will carry. Where a maker publishes none of that, the page says the answer is not published rather than deducing it from a photograph or a product name.