Why We Can Fix a Motor We've Never Seen

Almost every ebike hub motor on the road is the same basic device underneath: a three-phase brushless DC motor. Three thick wires carry the phase current, and a set of Hall sensors inside the shell report which of six rotor positions the motor is in right now. A controller has to fire those three phases in the right order, at the right instant, and it has to know the motor's gear ratio and magnet count to calculate speed correctly. Get any of that wrong and a perfectly healthy motor looks "dead."

When you fill out our intake form, we do not know your motor's phase-to-Hall alignment, magnet count, or internal gear ratio — nobody could from a web form, and guessing is exactly how a good motor gets misdiagnosed and thrown away. So we don't guess. We rely on three layers instead:

• A self-learning controller. Of the 36 mathematically possible ways to wire three phase leads against three Hall signals, only a handful actually spin the motor cleanly. A self-learning controller shorts a dedicated pair of leads, spins the wheel through those combinations on its own, and locks in the one that runs smooth and quiet with the lowest no-load current — what used to be a bench technician's afternoon becomes one power-up.
• A parameter preset table. Pole-pair/gear-ratio number, current ceiling, speed-signal source, brake logic, and PAS magnet count are pre-filled per motor family before your kit ships, based on our bench records and documented factory specs — not guessed at checkout.
• A pre-install technician check. Before anything leaves our shelf, someone reviews your photos and your bike's known electrical profile against our case history and confirms the kit has an actual track record on that bike or family.

What this method fixes — and doesn't — matters just as much. Self-learning corrects a wrong wiring order. It cannot correct a Hall sensor physically mounted at the wrong angle, a burned winding, a blown power transistor, or a locked proprietary firmware. A motor that only turns while the learning leads are connected, and stops the moment you unplug them, isn't a wiring problem — it's a deeper electrical fault that no re-learning fixes. We say so up front, not after a return.

Source: bench notes (research on self-learning controllers and phase/Hall pairing) · vendor documentation.

Step 0 — Hard Blockers

Before we talk parts, we check whether your bike can even take a generic controller-and-display kit. Five conditions stop us at the door, plus a sixth, newer category that is spreading fast across 2023-and-later models. If your bike matches any of these, an electronics-pack repair is not the right path — but it isn't a dead end either.

Source: engine rules (our diagnostic rule set blockers) · bench notes on expanding pairing-lock platforms.

Step 1 — The Bike's Mechanical Seven

Before anything electrical gets discussed, we ask for seven physical facts about your bike. None of them require tools you don't already have — a tape measure and a look at the tire sidewall cover all seven.

Source: engine rules (our diagnostic rule set match fields) · bench notes on wheel and drivetrain fitment.

Step 2 — The Symptom Tree (S1–S7)

Every intake starts with a free self-check before any part gets recommended. The steps below are the same seven branches our diagnostic engine walks — each one starts with something you can check for nothing, moves to a simple multimeter measurement if the free check doesn't resolve it, and only then points at a part.

S1 — Bike won't power on at all

1. Free check: battery seated, its own switch/key on, charged to a full green light on the charger.
2. Inspect the battery-to-controller plug and fuse for looseness, scorch marks, or corrosion at the mount terminals.
3. Measure: discharge-port voltage should be close to nominal — about 54.6V on a full 48V pack, about 42V on a full 36V pack. Near 0V or far below nominal points at the battery or BMS.
4. Try a known-good spare battery; if the bike still won't wake, try swapping the display before the controller — the power button lives on the display.

Most likely order: battery/BMS → fuse → display → controller.

S2 — Display stays dark (battery confirmed good)

1. Free check: reseat the display plug (the headtube is the most water- and tug-exposed spot on the bike); hold power 3–5 seconds in case it's just asleep.
2. Measure: confirm the controller is actually sending battery voltage down the display's power line — this line carries full pack voltage (36–52V), so a wrong pin here can destroy a display instantly. No voltage points at the controller's supply stage, not the display.
3. A replacement display that powers up but won't read speed or take an assist change is a protocol mismatch, not a dead display — see Step 4.

Most likely order: display → controller's internal supply stage → matched display+controller pair.

S3 — Display works, but no assist from throttle or pedaling

1. Free check: brake cutoff is the single most common cause — unplug the brake sensor and try again.
2. Confirm the assist level hasn't reset to zero (common after deep discharge) and the motor's main plug is fully seated.
3. Measure: test throttle and pedal-assist as two separate circuits. Both dead points to the controller; only one dead points to that sensor.
4. Throttle signal should sweep roughly 0.8–3.6V smoothly; a PAS sensor typically needs a 1–3mm magnet-disc gap.

Most likely order: brake sensor (cheapest) → throttle or PAS sensor (whichever circuit is dead) → controller → motor.

S4 — Throttle unresponsive, pedal assist fine

1. Free check: cycle the throttle by hand, check for pinched or chafed wire, reseat the connector.
2. Confirm the throttle hasn't been disabled or capped to a 6 km/h walk-mode setting.
3. Measure: supply should read about 4.3–5V; signal should sweep roughly 0.8–3.6V as you twist. A flat, non-moving signal means the throttle has failed.

Most likely order: throttle assembly (cheapest, replace to rule out) → controller's throttle input → display harness → display.

S5 — Motor jerks, clicks, or hesitates

1. Free check: look for a partly-seated or water-damaged motor plug and a loose axle nut. If the motor ever engages on its own with no command, that's urgent — pull the battery immediately and stop riding.
2. Measure Hall sensors (multimeter, 20V DC range): the supply pair should read about 5V; turning the wheel slowly, each of the three signal wires should swing cleanly between roughly 0V and 5V. A wire stuck at a fixed value means that Hall element failed.
3. Measure phase wires (power off): each pair of the three leads should show small, roughly-equal resistance to each other and no continuity to the motor case. Continuity to the case, or a coarse "clicking" feel hand-turning the wheel, points to a shorted winding.
4. If both tests pass, the fault is more likely a blown power transistor in the controller.

Most likely order: Hall element(s) or motor core → motor winding/phase wiring → controller.

S6 — Display shows an error code

1. Identify which electronics family the code belongs to first — see Step 4. The same digits mean completely different things on a KT display versus a Bafang-family one.
2. Universal first move: reseat the relevant connector and do a full hard reset (battery disconnected 10–15 minutes) before concluding anything is broken.
3. A short-circuit code means stop riding immediately — check wiring insulation before ever powering the bike again.
4. A communication-fault code follows the standard order: display first (cheapest), then controller, then a matched display+controller pair.

S7 — Power cuts out or the bike reboots mid-ride

1. Free check: cutouts on bumps point at loose or corroded battery-bay contacts — pull the battery and inspect the mount terminals.
2. Measure: watch pack voltage under full load — a steep sag (a 48V pack dropping below roughly 42V under hard acceleration) means the battery is aging and its internal resistance is triggering a low-voltage cutoff.
3. Recovers after resting a moment: a protective cutoff (voltage sag, heat, overcurrent) — battery or reduced load is the fix.
4. Won't recover, or cuts out repeatedly on flat ground: points to a harness or controller fault rather than the battery.

Source: engine rules (our diagnostic rule set symptoms S1–S7).

Step 3 — Motor Keep-or-Replace: Three Questions

Before anyone talks about a new motor, we ask three questions that, in our case history, correctly keep the original motor in the overwhelming majority of repairs. If you answer no to all three, the motor stays — the fix is almost always in the controller, display, or a sensor.

1. Spin the rear wheel by hand with the bike off the ground. Any grinding noise, roughness, or gritty feel?
2. Look at the motor's wire and connector. Any visible damage, scorch marks, or signs water got in?
3. Has the bike ever been crashed or submerged?

Three "no" answers → the motor defaults to keep. From there, an optional bench check confirms it with numbers instead of a guess:

• Hall sensor test (multimeter, 20V DC range): the +5V/ground pair should read close to 5V; slowly turning the wheel, each of the three signal wires should swing cleanly between roughly 0V and 5V. A wire that never moves means that Hall element has failed.
• Phase short test (power off, continuity/resistance range): each of the three thick phase leads should show a small, roughly-equal resistance to the others and no continuity at all to the motor's metal case. Continuity to the case, or a coarse "notchy" feel when hand-turning the wheel with power disconnected, indicates a shorted winding — that's a genuine motor fault, not a wiring or controller issue.

Source: SPEC electronics-first diagnostic flow · bench notes (Hall and phase test procedures).

Step 4 — Identifying Your Electronics Family

Two bikes can share the same connector shell and still speak completely different electrical languages. Before any part gets picked, we identify which of six electronics families your bike belongs to. You can usually tell from the boot screen, the label on the controller, the plug shape, or an error code — you don't need to open anything up.

• KT (Kunteng) — the white-label industry default (LCD3 / LCD5 / LCD8H). Menu is three layers deep (general settings, then P1–P5, then C1–C14, plus L-parameters on the larger LCD8H). Error codes appear as two digits with an "_info" suffix.
• No.2 generic sine-wave (S866 / S900 / S830 / S810 / SW900) — a separate UART family from KT despite sharing the same style of 5-pin display plug. Its menu is one flat P01–P20 table rather than KT's three layers. Error codes read E01 through roughly E13.
• Bafang UART — the aftermarket-kit side of the Bafang ecosystem. Error codes are two-digit numbers without the "_info" suffix (04, 08, 09, 30, etc.).
• Bafang CAN — the OEM, factory-integration side of the same brand family, and a different protocol value from Bafang UART even though both are "Bafang" — a UART display will not talk to a CAN controller, or vice versa.
• APT 5S — a distinct UART-style protocol used across several factory 48V platforms. Its error table (01 through 07, then reserved) doesn't overlap with KT's or Bafang's numbering.
• Closed proprietary CAN / paired platforms — factory one-piece and app-paired systems. These show branded error text (a startup-password or lockout condition) rather than a generic numeric code, and are the platforms Step 0 already screened out.

Source: engine rules (our diagnostic rule set error_codes) · bench notes (protocol family identification).

Step 5 — Picking the Parts

Once we know your mechanical seven and your electronics family, five factors decide exactly which controller, display, and wheel assembly go into your kit.

• Voltage. Controller, battery, and motor must share the same nominal voltage — this is the one spec with no tolerance for mismatch.
• Current. A higher-current controller can always be turned down in software, so kits are built around two overlapping current classes rather than one exact match to your stock amperage.
• Protocol family match. A display's protocol must equal its controller's protocol — a UART display paired against a CAN controller looks like it plugs in fine and simply will not communicate. The wiring harness itself is also sold in matching UART and CAN part numbers; mixing those produces the same silent failure.
• Connector shape. Handled separately in Step 6 — a shape match is necessary but never sufficient on its own.
• Clamp diameter and controller box length. Handlebar clamp sizing falls into three standard diameters — 22.2mm (side-mounted display on the grip section), 25.4mm, and 31.8mm (center-mounted, larger screens on the stem area) — so one display with the right shim pad covers all three; you don't need a different SKU per clamp size. Controller box length scales directly with current rating: roughly 98mm for a 20A-class box, 170–190mm for 35A, and 212–256mm for 45A. A higher-current controller frequently will not fit in a stock cavity built for a smaller factory unit, which is why some kits ship with an external, weatherproofed mounting option instead of a drop-in box.

Three physical pairings are locked together and can't be mixed later, regardless of protocol: wheel diameter is bound to the laced wheel assembly (a different diameter is a different SKU, not a later swap); freewheel-thread versus cassette-spline gearing is bound to the motor's hub body (the two mounting systems aren't interchangeable on the same hub); and disc-brake versus rim-brake wheels are built to the bike's original brake type at the factory.

Source: engine rules (our diagnostic rule set bafang_constraints) · bench notes on controller current/size scaling and clamp diameters.

Step 6 — Connectors and Adapters

An adapter cable solves one problem only: a physical connector shape that doesn't match. It does not, and cannot, fix a protocol mismatch — a data line either speaks the same language on both ends or it doesn't, and a perfectly-shaped adapter changes nothing about that. We say this plainly on every adapter listing because it's the single most common misunderstanding about what these cables do.

The connector families you'll most often run into on a hub-motor lead:

• Higo Z910 (9-pin) — the de facto physical standard for hub motors, common across the Bafang ecosystem and widely cloned. Nine pins typically cover three phase leads, five Hall/sensor wires, and one speed-or-temperature line.
• Julet 9-pin (Z916 family) — physically similar in pin count to the Z910 shell, common on Rad Power and a large share of generic-brand bikes.
• Loose 8-wire (no housing) — the fallback for generic and self-modified bikes, terminated with crimp connectors instead of a molded shell.
• L915 and related screw-lock proprietary shells — a different mounting family seen on some brand-specific motors, sold as not interchangeable with standard Higo or Julet shells and difficult to source in the open market.

Two rules follow from this. A matching shell on both ends of an adapter never means "plug and play" by itself — confirm the pin assignment before powering up, using tables we publish once verified. And on any connector still marked unverified, we ask for close-up photos of both ends first, and build the adapter's wiring around what our bench actually measures on your plug rather than a public reference diagram.

Source: our connector library · bench notes on 9-pin pinout variation · SPEC electronics-first diagnostic flow.

The Three Confidence Tiers

Every repair we quote lands in one of three tiers. Which one depends on how much real evidence we have that your specific bike-and-kit combination actually works — not on how confident it sounds.

These tiers aren't a guess dressed up as a traffic light. Across our case-history review of real swap outcomes, success rates split sharply by kit type, not by brand: bike-specific kits built for one exact model succeed the large majority of the time; generic KT-family kits succeed on most attempts but nearly always need tuning or a wiring adjustment first; unpaired, randomly-sourced generic parts fail more often than they succeed. That gap is why the Yellow tier exists — the difference between a kit we've tuned in advance and one nobody has tried on your bike yet.

Source: SPEC electronics-first diagnostic flow · bench notes (case-history success rates by kit type).

What We Guarantee, and What We Can't

We guarantee the outcome of our process, not a guess about your specific bike sight unseen. Every kit goes through a pre-install technician check against your photos and your bike's known profile before it ships. If it turns out not to fit, or the display and controller won't communicate once installed, we upgrade you to the correct kit or refund the difference — that resolution is built into how we sell these kits, not a favor.

What we don't promise is that the very first kit we ship installs correctly on the first try, every time, for every bike. Electronics families that look identical on a spec sheet can differ in pinout between production batches of the same model, and a small number of fields in our own bike records stay marked unverified until our bench or a customer's install confirms them. We'd rather say that plainly than promise a number we can't back up.

Where we draw a hard line, we say so up front instead of discovering it after you've paid: the hard blockers in Step 0, a motor that fails its keep-or-replace check, and any bike with no documented successful third-party swap all get told to you before checkout, not after.

Source: SPEC electronics-first diagnostic flow.

The Five Photos We Ask For

Every intake — Green, Yellow, or Red — starts with the same five close-up photos. None of them require removing anything from the bike.

1. Display, powered on, front-facing. The boot screen and menu layout are often the fastest way to identify which electronics family a bike belongs to, before any error code even appears.
2. Display's plug. Pin count and shell shape narrow down the protocol family and rule out a same-name, different-wiring trap.
3. Controller label, with every plug attached visible. The label usually carries the current rating and sometimes the protocol; the plug count and layout tell us what has to be replicated on a replacement.
4. Motor lead plug, close-up. This is the connector we have the least public data on across the industry — a clear photo is often the only way to confirm pin count and shell family before recommending an adapter.
5. Brake levers and their wires. Brake-cutoff logic (normally-open versus normally-closed) usually can't be determined from the lever's outward brake type alone — hydraulic brakes, for example, still typically use an ordinary magnetic or mechanical cutoff switch at the lever, not a hydraulic pressure sensor — so a photo of the actual connector is what settles it.

Source: SPEC electronics-first diagnostic flow · bench notes (motor lead identification, brake cutoff logic).