Battery Health & Degradation

How Lithium-Ion Batteries Degrade

Battery degradation is an inevitable process where the battery's ability to store energy decreases over time. This happens through several mechanisms at the molecular level.

Zero Battery Specifications

Every Zero street and FX-family pack is 28 cells in series, from the 2012 ZF6 to the 2026 DSR/X, at roughly 102 V nominal. (The 2026 LS1 / XE / XB light line is the exception, at 74 V.) The manuals never name a cathode chemistry or a cell supplier, so neither do we.

For a health page only two capacity numbers matter, and confusing them is the single easiest way to compute a wrong state of health:

Every Zero pack, badge against usable

Which model carries which pack, and how a badge that covers two different physical constructions gives itself away in your log filenames, is on Battery Fundamentals; full detail on each pack is in the battery catalogue. One footnote: the figures above are the ones Zero prints from model year 2024 on. The MY2022-2023 manuals quoted the ZF15.6 as 13.7 kWh and the ZF17.3 as 15.2 — 0.1 kWh higher in each case — so an older manual will disagree by that much.

How we put a kWh number on your pack — two of them, never blended

There are two entirely different ways to say how much energy a used pack still holds, and this site computes both and shows them side by side. They share no inputs, so where they agree you can believe them, and where they disagree the gap is itself the finding.

Ampere-hours to kilowatt-hours: the constant that decides the answer

The register is in ampere-hours and every published Zero figure is in kilowatt-hours, so a pack voltage has to be chosen — and getting it wrong by 12% moves state of health by 12%. We use 115.6 V, which three independent measurements converge on:

• • Series count. Divide logged pack voltage by lowest cell voltage on every Gen3 BMS file and the answer is 28.02. The packs are 28S.
• • Top of charge. The highest cell voltage anywhere in the Gen3 corpus is 4128 mV, at 95% SOC. 28 × 4.128 V = 115.6 V — and that is a floor, because the pack was not yet full.
• • Nameplate ÷ register on the one bike whose nameplate every source agrees on and whose register has not moved: 14,400 Wh ÷ 124.5 Ah = 115.64 V.

The obvious wrong choice is the resting mid-SOC voltage, ~102 V, which is all most log files ever show — it understates capacity by about 12%. Because the constant matters that much, the API echoes it back in every response instead of hiding it, so the arithmetic can be redone by hand.

Why we call it capacity-against-badge and not remaining life

This is the part worth being honest about. Here is every bike in our corpus whose nameplate resolves cleanly, with its own register reading:

Three things are wrong with reading that column as remaining life:

• • Ranked by age, the order is exactly backwards. The oldest bike reads highest and the newest reads lowest — a rank correlation of −1 on three bikes. Real degradation would run the other way.
• • Two of the three sit at or above 100% of their badge. Nothing degrades upward. Those readings are telling us about nameplate rounding and cell-lot variance, not about wear.
• • Two identical packs disagree with each other by 2.33%. The 2020 and 2021 SR/F are the same ZF14.4 one model year apart, and their registers differ by more than either differs from its own badge. That difference is the floor of what this method can resolve.

So the figure is published as capacity against nameplate, ±3 percentage points — a band measured from that scatter, not chosen to look modest — and it is never labelled state of health remaining. A pack reading three points below its badge has told you nothing. A pack reading fifteen points below it has. The stability of the register is what makes the trend the useful part: the two bikes in our corpus with more than one log file read byte-identical across them, one pair a day apart and the other about three months apart, so movement means a relearn or a pack swap rather than measurement noise.

Capacity is one axis, and on its own it is a poor summary

“How healthy is my battery” sounds like one question with one number behind it. It is not. Usable capacity is the headline because it is the one you feel — it is your range — but a pack can hold plenty of energy and still be in trouble, and it can lose a little and be perfectly sound. These are the axes worth reading together.

The balance one is worth seeing rather than reading about. Both packs below hold almost the same energy; only one of them is fine. Switch between them:

This is why a single percentage is a poor summary, and why this site prints the capacity figure beside its provenance rather than as a score out of a hundred. If your logs carry per-cell voltages, the cell-imbalance trend on your battery page is reading exactly the spread shown above, over time.

Measuring it yourself, at the wall

You do not need us to do this. A cheap energy-metering plug between the socket and the charger will tell you what a charge cost, and if you note the state of charge at both ends you have measured your own pack. There is one trap in it, and it is worth more than the rest of this page.

A worked example, from a real 2016 FXS with a 6.5 kWh badge. The meter read 2.944 kWh for a charge that took the bike from 40% to 96% — 56 points of charge:

So that pack holds somewhere around 4.5 to 4.8 kWh usable today. Read against the 5.7 kWh it would have given when new, that is roughly 80% of its original usable energy after ten years, which is an ordinary result and not a fault.

Read against the 6.5 on the badge it would look like 71%, and that is the number that frightens people. It should not: the step from 6.5 to 5.7 was true on the day the bike left the factory and is not degradation at all. Only the step from 5.7 to 4.5 is.

Two honest limits on the method. The last few per cent of a charge run at constant voltage with the current tapering, so the energy per point of charge is not quite the same up there as it is in the middle — keep the window away from both ends where you can. And extrapolating from 56 points to a full 100 assumes the state of charge tracks energy linearly, which it does in the middle and does less well at the extremes. Both are errors of a few per cent, not of a third.

When we refuse to print a percentage at all

A percentage needs a denominator, and four sources can supply one: your own declared pack, the bike record, the VIN decode and the model spec table. If they disagree by more than 2% we do not pick a winner — we show the capacity in Ah and kWh and say why there is no percentage. Two of the five bikes in our corpus land there:

Both are fixed the same way: declare your actual pack in the component configurator. An owner-declared capacity is a human statement rather than a guess, so it wins outright over all four automatic sources and short-circuits the agreement check.

And when there simply are not enough measurements yet

A denominator is only half of it. The other half is how many times we have actually measured your pack — and below four charges the answer is noise. Across our whole fleet, of the sixteen bikes that once reported an impossible figure (over 100% health), thirteen had three measurements or fewer behind them. The same motorcycle, on that little evidence, has read anywhere from 51% to 113%.

So the page now shows a count instead of a percentage until four charges have counted — “2 of 4”, filling as they arrive. It is not a low score. It is a statement about our evidence, not about your battery, and an unmeasured pack is not a damaged one.

This matters most on Gen3 bikes, which log far fewer usable charges than Gen2 — a median of two against eight. It is why a 2023 SR/F can show a full page of telemetry and still say it is measuring.

Degradation Simulator

Use the sliders below to model how different factors affect battery degradation over time. This is a simplified educational model based on general Li-ion research.

Visualizing Cell Degradation

Watch how individual cells in a battery pack degrade over time. Notice how cell capacities diverge as the pack ages, leading to increased imbalance.

Signs of Degradation in Your Logs

For scale on the first row: across 10,555 real BMS discharge rows from four log files in our corpus, the median cell spread is 5 mV, 98.4% of rows sit at or under 40 mV, and only 0.2% exceed 120 mV. A pack in the 41-80 mV band is unremarkable, which is why nothing is raised there — the previous version of this table called 60 mV a warning while the product called the same reading normal.

How to Maximize Battery Life

Module-Level Diagnostics

These diagnostics live in thelog_entries.raw_dataJSON column — surfaced in your dashboard's battery-health views and accessible via the API. Look for them on Gen2 logs (FXS, S, SR, DS, DSR, FX with multi-brick packs); single-brick configs only emit one set of values per contactor event.