A pack's usable capacity is set by its weakest cell, and the number that decides how much that costs is the count of cells in series. A 51.2 V module holds sixteen of them. A 460.8 V stack holds a hundred and forty-four. One cell delivering one percent less charge than its neighbours takes one percent off the whole pack, whatever figure the label prints.
The number worth knowing is the count of cells in series
In a series string the same current passes through every cell, and discharge ends when the first cell reaches its cut-off. The string's deliverable charge is therefore the weakest cell's charge, and the pack's usable energy follows it. Cells arranged in parallel behave differently: parallel strings add their charge, so a weak cell in one string is diluted by the number of strings beside it.
That is the whole mechanism. Consistency matters because of how the cells are wired, not because of how they are advertised.
Where that number comes from
Two published figures are enough to count the cells. Every 51.2 V module publishes that nominal voltage, and the cell chemistry is published as LiFePO4, whose nominal cell voltage is 3.2 V. The division gives sixteen cells in series per module. The publishes 153.6 V for three modules and 460.8 V for nine, which is forty-eight and one hundred and forty-four cells in series.
The published module energy confirms the arithmetic rather than contradicting it. A 5.12 kWh module at 51.2 V is 100 Ah, and 100 Ah is the charge the string carries per cell. A 10.24 kWh module at the same voltage is 200 Ah, which is either a larger cell or a second parallel group, and the published specification does not say which.
What a one percent spread costs
Worked example. Apply a one percent deficit in delivered charge to the published energies. The assumption is illustrative: no matching window is published for these models, and the figure below is arithmetic, not a product claim.
Cell counts divide the published nominal voltages by 3.2 V; energies and the one percent column are arithmetic on the module capacities published on the Ruibit Energy product pages, verified as of 2026-09. Voltages are in volts, energies in kWh.
Two things follow. The percentage penalty lands on the whole pack, so the same tolerance costs nine times as much energy on a nine-module stack as on a single module. And the low-voltage stack shows the other direction: three strings in parallel dilute one weak string to a third of the penalty, which is why parallel capacity is forgiving where series capacity is not.
What the number does not tell you
Three things, and they are the three a buyer usually assumes.
It does not tell you the spread inside a delivery. The published pages examined here state capacity, chemistry, nominal voltage, current limits, cycle life, efficiency, temperature range, communication, protection rating, dimensions, weight and warranty. A matching window is not among them, so the distribution behind a nominal figure is unknown until it is requested.
It does not tell you what the end-of-line capacity test measured on the pack you receive, because that is a production record rather than a datasheet field. And it says nothing about internal resistance, which is the property that decides how hard the management system has to work to hold a long string together. Grade A is a label, not a distribution.
When the number changes
The cell count changes with the format before it changes with anything else. Moving from a single 51.2 V module to a nine-module 460.8 V stack multiplies the series count by nine in one step, and the is also the decision about how many cells stand in one line.
Spread changes underneath it. Cells that start a fraction apart diverge as they age, and the published cycle life is quoted at 80 percent depth of discharge, so the depth the household actually uses decides how quickly the divergence becomes visible. A pack that reports imbalance is usually reporting a consistency problem before it reports a capacity problem.
What to ask for instead
Ask for the capacity distribution of the delivery and the internal-resistance distribution beside it, the sample size behind both, and the end-of-line capacity result for the serial number being shipped. Put the matching window in writing, because a window applied to a different measurement condition is a different window. A supplier who sends a distribution and the conditions it was measured under has answered the question; a supplier who sends one nominal value has restated the label. The sets out which documents to request, and the holds the rest of this cluster.
If the quotation gives a single nominal capacity, it has described a design rather than a delivery. If it gives a usable capacity, ask at which depth of discharge that figure was measured.
Two consequences are fixed at assembly. The series count cannot be changed once the stack is installed, and the cells cannot be exchanged for better-matched ones without rebuilding the string. The shows the short string; the shows the scale at which the handover record, as the baseline that later imbalance is compared against, matters most.
FAQs
1. Why does one weak cell reduce the whole pack's usable capacity?
Because a series string shares one current. Discharge stops when the first cell reaches its cut-off, so the string can only deliver what its weakest cell can deliver. Every cell behind that one still holds charge when the pack stops, and that residual charge is the capacity the owner paid for and cannot use.
2. How many cells are in a home battery pack?
It follows from the published voltage. A 51.2 V module contains sixteen cells in series at 3.2 V each. A 153.6 V stack contains forty-eight, and a 460.8 V stack contains a hundred and forty-four. The count rises with the format, not with the kilowatt-hours.
3. Does cell consistency matter more in a longer string?
The percentage penalty lands on the whole pack either way, so the same tolerance costs more energy where more energy is stored. A longer string also offers more positions for an outlier cell to occupy. Parallel strings work the other way and dilute a single weak string.
4. Does the site publish a cell matching window?
No. The product pages publish capacity, chemistry, nominal voltage, current limits, cycle life, efficiency, temperature range, communication, protection rating, dimensions, weight and warranty. A matching window and an internal-resistance distribution are not among the published fields, so both have to be requested.
5. What should I ask for instead of a Grade A label?
The capacity distribution of the delivery, the internal-resistance distribution beside it, the sample size behind both, the measurement conditions, and the end-of-line capacity result for the serial number being shipped. A distribution with its conditions answers the question; a single nominal value restates the label.
