How to Service High-Torque Mid-Drive E-Bike Reduction Gears: Grease Guide (2026)

Root Cause Engineering & Mechanical Principles

The Bafang Ultra M620 (factory designation G510) is a 1,000 W nominal, 160 Nu00b7m mid-drive built around a two-stage reduction train driven by an inrunner BLDC rotor spinning at roughly 4,200–4,600 RPM under load. That rotor speed must be converted down to a crank cadence of 60–90 RPM, which means the gearbox carries a total reduction ratio near 1:21.9. The first stage is a steel helical pinion cut directly into the rotor shaft, meshing with a large nylon-composite (PA66-GF30) primary gear. The second stage is a steel-on-steel spur mesh that carries torque into the pawl clutch and out to the bottom bracket spindle. Understanding this layout matters because the two stages have completely different lubrication requirements and completely different failure signatures. For detailed technical setup and system integration, reference our analysis on test and replace brushless hub motor hall sensors on.

Strictly speaking, the M620 does not use a true epicyclic planetary set the way a geared hub motor or the smaller BBSHD-style units are often described. Riders search for “planetary gears” because the visual impression of the stacked reduction cluster suggests one, and because Bafang’s own parts catalogs list “reduction gears” generically. The practical consequence is identical, though: you are servicing a multi-stage reduction cluster where tooth-root stress, grease film thickness, and bearing preload all interact. Calling it the reduction stack is more accurate, and the procedure below applies regardless of nomenclature.

The dominant root cause of M620 noise and premature wear is simple and well documented across owner teardowns: the factory applies a token quantity of thin lithium soap grease — frequently measured at 12–18 g total across a gearcase with an internal free volume near 180 cmu00b3. That grease is typically an NLGI #2 general-purpose lithium with no extreme-pressure (EP) additive package and a base oil viscosity around 100–150 cSt at 40 °C. Under the Hertzian contact stress at the steel-to-nylon first-stage mesh, which can exceed 400 MPa at peak torque, a non-EP grease of that consistency does not maintain an elastohydrodynamic film. The grease is also too stiff to re-flow into the mesh at cold-start temperatures, so the gear teeth run effectively dry for the first several minutes of every ride.

The secondary failure mechanism is thermal. The M620’s sealed aluminum case has poor internal convection, and sustained climbing at 30 A battery current pushes internal gearcase temperatures to 75–95 °C. At that temperature, a thin lithium grease bleeds its base oil, the thickener migrates to the case walls, and what remains at the tooth flank is a dry soap residue. This is why a motor that is acceptably quiet for the first 300 miles develops a characteristic rising whine between 800 and 2,000 miles. The nylon primary gear then begins to suffer adhesive wear, shedding fine white polymer dust that you will find caked on the magnet ring and inside the case seams during teardown.

A third contributor, less common but far more destructive, is pawl clutch degradation. The one-way clutch between the secondary gear and the output spindle uses three spring-loaded pawls engaging a ratchet ring. When grease starvation reaches the clutch, the pawls fail to retract cleanly and begin skipping under load, producing a sharp mechanical clack and, eventually, shearing the ratchet teeth. Clutch skip is a hard-stop condition: continue riding and you will destroy the secondary gear as well. The U.S. Consumer Product Safety Commission (CPSC) maintains recall and incident reporting resources covering e-bike drivetrain and battery hazards, and a drivetrain that can disengage without warning under climbing load is precisely the kind of sudden-loss-of-control condition worth taking seriously rather than riding out.

Grease selection is where most rebuilds succeed or fail. The target is a tacky, high-base-oil-viscosity, EP-fortified product in the NLGI #0 to #1 range. Cornhead grease — the semi-fluid lithium-complex product sold for agricultural combine header gearboxes, John Deere Corn Head Grease being the archetype — has become the de facto community standard for exactly the right reasons: NLGI #0 consistency so it re-flows into the mesh at 0 °C, base oil viscosity in the 400–600 cSt range at 40 °C for a thick EHL film, and a tackifier that keeps it clinging to vertical gear faces instead of slumping to the case floor. Lithium-complex EP #0 or #00 gearbox greases from Lucas, Mobil, or Red Line perform equivalently. What you must avoid is moly-fortified paste (the solids abrade nylon), any NLGI #2 chassis grease (too stiff to circulate), and anything calcium-sulfonate marketed for marine use (incompatible thickener chemistry if residual factory lithium remains).

Symptom-to-Failure-Mode Diagnostic Matrix

Step-by-Step Diagnostic & Calibration Procedure

Work on a clean bench with the bike either removed from the frame or supported in a repair stand with the drive side accessible and level. Budget two hours for a first attempt. Photograph every stage before you separate it — gear stack orientation is not obvious on reassembly, and the nylon primary gear is handed. For detailed technical setup and system integration, reference our analysis on regrease mid-drive nylon reduction gears (2026).

Phase 1 — Electrical Isolation and Baseline Measurement

1. Disconnect and verify. Remove the battery from the frame. Wait 90 seconds for controller capacitor discharge, then confirm with a multimeter across the main power connector: must read below 2.0 V DC before you proceed. The M620’s 48 V/52 V bus holds a lethal-to-components charge in the controller caps.
2. Record baseline noise. Before disassembly, log the speed at which whine onset occurs and the perceived frequency. A whine that tracks wheel speed indicates an output-side fault; one that tracks motor RPM independent of gear selection points at the first-stage mesh. This distinction saves you an unnecessary clutch teardown.
3. Baseline thermal. Ride a known 1-mile climb at consistent power and take an IR thermometer reading on the case center within 30 seconds of stopping. Note the value. You will repeat this after the rebuild — a correctly repacked M620 typically runs 8–15 °C cooler on the same climb.
4. Measure Hall and phase integrity. With the motor connector split, check phase-to-phase resistance on the three fat leads. All three pairs must agree within 0.1 Ω. Then verify each Hall signal line toggles between roughly 0.2 V and 4.8 V DC as you slowly rotate the cranks with a 5 V reference applied. Any phase imbalance or dead Hall means your problem is electrical, and no amount of grease will fix it.

Phase 2 — Gearcase Access

5. Remove the drive-side crank arm. The M620 uses an 8 mm hex crank bolt. Back it out counterclockwise, then use a crank puller to draw the arm off the splined spindle. Do not hammer — spindle deflection destroys output bearing alignment.
6. Remove the chainring spider and lockring. Note the spacer stack order and orientation. Photograph it. Spacer sequence determines chainline, and getting it wrong introduces a 2–3 mm chainline error that will cost you shifting quality.
7. Remove the gearcase cover. Five M4 socket-head screws retain the right-side cover on most M620 revisions; a small number of later production units use six. Back them out in a star pattern, quarter-turn at a time, to avoid cocking the cover. Lift straight off — do not pry at the parting line with a screwdriver, which gouges the sealing surface and guarantees future weepage.
8. Inspect and document before cleaning. With the cover off you are looking at the reduction stack. Record: total grease mass present (scrape into a tared container — you will almost certainly find 12–20 g), grease color and texture (gray and crumbly means it has bled out and is spent), and the presence of white nylon dust (indicating primary gear wear already underway).

Phase 3 — Inspection and Tolerance Verification

9. Measure backlash. Mount a dial indicator against a primary gear tooth flank, lock the pinion, and rock the gear. Acceptable backlash for this stack is 0.05–0.15 mm. Above 0.25 mm indicates significant tooth wear and the gear should be replaced regardless of how it looks.
10. Measure tooth thickness. Using a dial caliper or gear tooth vernier, compare chordal thickness at the pitch line on a loaded tooth against a lightly loaded one. A differential exceeding 0.15 mm means the nylon primary is past service life. Steel secondary gears tolerate more — replace at 0.25 mm differential or at any sign of pitting or spalling on the flank.
11. Test the pawl clutch. Rotate the output in the drive direction: engagement must be crisp and occur within 20° of arc. Rotate in the freewheel direction: pawls must retract cleanly with an even three-click cadence. Apply roughly 15 Nu00b7m by hand in the drive direction; any slip condemns the clutch and ratchet ring, which must be replaced as a matched set.
12. Check every bearing. Spin each accessible bearing — typically 6902, 6903, and 6802 series depending on position and revision. Any notching, roughness, or audible dryness is a fail. Radial play, checked with a dial indicator, must stay under 0.05 mm. Replace with sealed 2RS bearings from a reputable manufacturer; the cost difference against generic is trivial against a second teardown.
13. Inspect the nylon primary for heat glaze. A glossy, slightly translucent sheen on tooth flanks indicates the gear has exceeded its thermal limit and the surface layer has softened and re-hardened. Glazed gears fail abruptly. Replace them.

Phase 4 — Cleaning and Repacking

14. Remove all factory grease. This is non-negotiable. Mixing a lithium-complex EP grease with residual simple-lithium soap can cause thickener incompatibility, resulting in a dropped base oil and a case full of soap paste. Use isopropyl alcohol or a plastic-safe brake cleaner on a lint-free cloth. Keep solvent away from the stator windings, the Hall sensor assembly, and all wire insulation.
15. Degrease the parting surfaces. Both the case and cover sealing faces must be clean, dry, and free of old gasket residue. Inspect the gasket: if it shows compression set, flattening, or any tear, replace it. A new gasket costs a few dollars; a weeping gearcase costs a rebuild.
16. Pack the first stage. Work grease directly into the tooth roots of the steel pinion and the nylon primary mesh with a gloved finger or a small acid brush. The goal is complete tooth root fill with a visible fillet, not a surface smear. This stage carries the highest contact stress and sees the highest sliding velocity.
17. Pack the second stage and clutch. Apply the same tooth-root technique to the steel-on-steel secondary mesh. At the pawl clutch, use a lighter hand — a thin, even film on the pawl pivots and ratchet faces, roughly 3–5 g total. Over-greasing the clutch causes hydraulic pawl lag and intermittent engagement delay, which feels exactly like a slipping clutch.
18. Hit the target fill mass. Total grease in the case should land between 60 g and 80 g. Weigh it. The internal free volume is near 180 cmu00b3, and you are targeting 40–45% void fill. Below 50 g you have not solved the starvation problem. Above 90 g, the grease has nowhere to expand when it reaches 90 °C, internal pressure rises, and it will force past the seals and contaminate the stator.
19. Clear the vent path. Many M620 cases have a small breather or pressure-equalization path. Confirm it is not packed with grease. A blocked vent converts normal thermal expansion into seal-blowing pressure.

Phase 5 — Reassembly and Torque Specification

20. Seat the gear stack in recorded order. Reference your photographs. The nylon primary gear is handed — installing it reversed produces immediate severe noise and rapid tooth failure.
21. Fit the cover and torque the M4 screws to 4 Nu00b7m in a star pattern, in two passes (2 Nu00b7m, then 4 Nu00b7m). Use a quality 1/4-inch torque driver. These screws thread into aluminum; exceeding 5 Nu00b7m strips the boss, and a stripped cover boss is a case-replacement event.
22. Reinstall the chainring spider and lockring to 40 Nu00b7m, verifying your spacer stack matches the original photograph.
23. Torque the crank bolt to 40 Nu00b7m. Apply a thin film of anti-seize to the spindle splines first. Re-check this bolt after your first 25 miles — initial spline seating commonly loosens it.
24. Reconnect and bench-test. Reinstall the battery. With the rear wheel off the ground, apply assist at the lowest PAS level and listen. The motor should sound noticeably duller and lower in pitch than before. New grease produces a slight initial drag that disappears within 10–15 miles as it distributes.
25. Thermal verification. Repeat the same 1-mile climb from step 3 and take an IR reading. A correctly executed repack typically returns a case temperature 8–15 °C lower than baseline. If temperature is unchanged or higher, you have over-filled the case or a bearing is dragging — open it back up before you put serious miles on it.

Tools, Calibration Equipment & Component Replacement

The tool list splits into three tiers. The first tier covers a grease-only service and is inexpensive. The second tier covers gear and bearing replacement. The third tier covers measurement and verification, and this is the tier most home mechanics skip — which is precisely why so many rebuilds end in a second teardown.

Tier 1 — Grease Service Essentials

• Hex key set, 2–8 mm — quality steel only. The M4 cover screws round out instantly under a worn or loose-fitting key.
• Crank puller — standard square-taper/ISIS-style puller suits the M620 spindle.
• 1/4-inch torque driver, 1–6 Nu00b7m range — mandatory for the 4 Nu00b7m cover screws. A 3/8-inch automotive click wrench cannot resolve accurately at this torque.
• 3/8-inch torque wrench, 20–60 Nu00b7m — for the 40 Nu00b7m crank bolt and lockring.
• NLGI #0–#1 EP gear grease, 1 lb minimum — cornhead-type lithium complex, 400–600 cSt base oil, tackifier present, no solid moly. One pound covers roughly six services.
• Digital kitchen scale, 1 g resolution — the single most undervalued tool here. Fill mass is a specification, not a feeling.
• Isopropyl alcohol and lint-free cloths — for complete removal of the incompatible factory grease.
• Replacement gearcase gasket — order it before you start, not after you tear the old one.

Tier 2 — Component Replacement

Several vendors now supply hardened steel replacements for the nylon primary gear. These are genuinely stronger in tooth-root bending and are the right answer for riders running 1,500 W-plus controllers or heavy cargo loads. The trade-off is real and worth stating plainly: the nylon gear functions as a mechanical fuse and as the primary noise-damping element. Convert to all-steel and you gain durability but accept a measurably louder motor and the removal of the sacrificial element that protects the clutch and spindle during a shock load. For a stock-power M620 on recreational use, a fresh OEM-spec nylon primary plus correct grease is the better engineering choice.

• Primary reduction gear (nylon-composite or steel upgrade) — specify your motor revision; early and late M620 production differ.
• Secondary steel reduction gear — replace only on confirmed pitting or >0.25 mm tooth thickness loss.
• Pawl clutch and ratchet ring set — always replace as a matched pair. Mixing a new pawl set into a worn ratchet ring produces point-contact loading and rapid re-failure.
• Bearing set (6902-2RS, 6903-2RS, 6802-2RS as applicable) — buy a known brand. Generic bearings in a 160 Nu00b7m gearbox are a false economy.
• Bearing driver set or arbor press — never drive a bearing by its inner race when seating on the outer, or you will brinell the raceway before the first revolution.
• Gear puller, small two-jaw — for drawing the primary gear off its shaft without lever damage.
• Speed sensor and magnet — cheap, and worth replacing preemptively while the case is open.

Tier 3 — Measurement and Verification

• Digital multimeter, 0.1 Ω resolution — phase balance and Hall signal verification.
• Dial indicator with magnetic base — backlash and bearing radial play.
• Digital caliper, 0.01 mm resolution — tooth thickness comparison.
• Feeler gauge set — parting line verification.
• Infrared thermometer — before-and-after thermal validation, the cleanest objective proof the rebuild worked.
• Nitrile gloves and parts tray — grease contamination of the stator is the most common self-inflicted rebuild failure.

Preventative Maintenance & Long-Term Reliability Protocols

The M620 responds extremely well to scheduled service, and the schedule is short. Treat it the way you would a small industrial gearbox rather than a bicycle component.

• Initial service at 200–300 miles. Open a new motor early. You are not fixing a problem — you are correcting the factory fill before it causes one. This single intervention is the highest-value action available to an M620 owner and routinely prevents the entire wear cascade.
• Inspection every 1,500 miles. Pull the cover, assess grease condition and distribution, and top up to the 60–80 g target. Expect to add 10–15 g per interval. Grease that has gone gray, crumbly, or dry means you should move to a 1,000-mile interval for your use case.
• Full repack every 3,000 miles. Complete removal and replacement, with bearing spin-check and clutch engagement test. Replace the gasket at every full repack.
• Shorten intervals by 40% for heavy-duty use. Cargo loads over 250 lb combined, sustained grades above 8%, ambient temperatures above 90 °F, or controller current above 30 A all accelerate base oil bleed. Adjust accordingly.
• Shift under reduced load. The M620’s torque sensor does not automatically cut power during a shift. Ease pedal pressure for roughly 200 ms at each shift. Shifting under full 160 Nu00b7m load is the single largest contributor to pawl clutch and ratchet wear.
• Observe the thermal ceiling. Sustained case temperatures above 90 °C accelerate grease degradation non-linearly. On long climbs, drop to a lower assist level and let cadence do the work — the motor is dramatically more efficient at 70–85 RPM crank cadence than at 45 RPM.
• Verify crank bolt torque quarterly. A loosening crank bolt allows spindle movement that misaligns the output bearing and damages the gear mesh. 40 Nu00b7m, checked with a wrench, not by feel.
• Keep a maintenance log. Record date, odometer, grease mass added, and case temperature after a standard climb. Three data points across two services will tell you more about your motor’s trajectory than any forum thread.

One operational note on break-in: after a full repack, expect 10–15 miles of slightly elevated drag and a marginal reduction in range while the fresh grease distributes and sheds its initial churning losses. This is normal. If drag persists beyond 25 miles, you have over-filled the case — pull the cover and remove grease down to the 80 g ceiling.

Frequently Asked Questions

Can I service the M620 gears without removing the cranks, as some owners claim?

Partially, and the distinction matters. On some M620 revisions you can remove a small access cover with three screws and reach the outer face of the reduction stack, which permits a crude grease top-up. That is not a service. You cannot reach the first-stage pinion mesh, you cannot inspect the pawl clutch, you cannot spin-check the bearings, and you cannot verify total fill mass because you have no idea what remains underneath. A partial top-up through the access port will quiet a whining motor for a few hundred miles while the underlying wear continues unobserved. For a genuine rebuild, pull the crank arm and the chainring spider and take the full cover off. The extra 20 minutes buys you the inspection data that determines whether you need a $12 tube of grease or a $90 clutch set.

Why does mixing grease types matter if they are both lithium-based?

Because “lithium” describes the thickener’s metal soap, not its full chemistry. Simple lithium soap and lithium-complex soap have different structures, and when they are blended the combined thickener matrix can destabilize — the soap loses its ability to retain base oil, releases it, and collapses into a soft paste with a sharply reduced dropping point. The practical result inside an M620 is a case full of separated oil pooling at the bottom and dry thickener residue sitting on the gear teeth, which is functionally worse than the under-greased factory condition you were trying to fix. The failure is not immediate; it typically emerges after the first few heat cycles above 80 °C, which is exactly when you have stopped suspecting the rebuild. Clean every trace of the old grease out with isopropyl alcohol. It takes 15 minutes and removes the variable entirely. For detailed technical setup and system integration, reference our analysis on calibrate bottom bracket torque sensors (2026).

Is upgrading to an all-steel primary gear worth it for a stock-power M620?

For a stock-power recreational M620, generally no. The nylon-composite primary gear serves two deliberate design functions beyond simply transmitting torque: it damps the gear mesh noise that a steel-on-steel first stage would broadcast through the aluminum case, and it acts as a sacrificial mechanical fuse during shock loads — a sudden shift under full torque, or a pedal strike. Replacing it with steel transfers that shock energy directly into the pawl clutch, the output spindle, and the rotor bearings, which are all considerably more expensive and more labor-intensive to replace than a $35 primary gear. The steel upgrade earns its place in a specific envelope: controllers pushing beyond 1,500 W, sustained heavy cargo operation, or riders who have already destroyed two or more nylon gears and have accepted the noise penalty as a deliberate trade. For everyone else, a fresh OEM-spec nylon gear running in 70 g of correct EP grease will outlast the rest of the drivetrain.