Why high-precision EV gears quote at 12–16 weeks, the specification items that actually control noise, and how to book grinding capacity before the shortage does it for you.
Photo: Foxcorner / CC BY-SA 4.0 / Wikimedia Commons
Electric drive has simplified the gearbox and complicated the gear. A single-speed reducer replaces a six- or eight-speed transmission, which removes synchronisers, clutch packs and a large part of the part count — and simultaneously removes every acoustic mask that used to hide gear quality. What remains is two or three meshes running at motor speed, in a vehicle that has no engine to drown them out.
That combination has produced a supply situation that purchasing teams still find surprising. Demand for EV reducers is growing at a double-digit rate while the capacity to grind gears to the required accuracy grows much more slowly, because grinding capacity is bought in years and demand arrives in quarters. The result is a market where the quoted lead time says more about the supplier than the quoted price does.
A 12 to 16 week quotation is not a negotiation tactic; it is the sum of stages that cannot be compressed much further. The breakdown below is typical for a new part number at a Chinese precision gear supplier, assuming the drawing is frozen and the DFM review is complete.
| Stage | Weeks | Why it cannot be shortened |
|---|---|---|
| Forging die design and build | 3–5 | die sinking, trial, correction loops |
| Forging, normalising | 1–2 | furnace cycle, structure homogenisation |
| Soft machining: turn, hob | 1–2 | fixturing and first-piece approval |
| Carburising and quenching | 1–1.5 | diffusion time at temperature |
| Fixtured grinding | 2–3 | booked machine hours, often the bottleneck |
| Inspection, PPAP documents | 1 | CMM queue, noise bench, reports |
Read the table as a queue rather than a chain. Soft machining capacity is elastic: a supplier can add a hobbing machine in a quarter. Carburising furnaces and gear grinding machines are not — grinding hours are booked months ahead, and a new grinding cell takes a year or more to specify, install, qualify and staff. This asymmetry, not the price of steel, is what makes precision gears structurally tight.
Most EV gear drawings specify an accuracy grade and stop there. The grade is necessary but not sufficient: two gears both meeting GB/T 6 can differ audibly, because the grade bounds individual deviations but says nothing about how they combine into transmission error, and nothing about the modification that shapes the mesh.
| Characteristic | Typical target | What it controls |
|---|---|---|
| Profile and lead grade | GB/T 5–6, DIN 5–6 | basic excitation level |
| Single pitch deviation | ≤ 6–8 µm at m2–3 | high-frequency whine |
| Transmission error, peak | ≤ 1.0–1.5 µm under load | the tone the driver hears |
| Lead crown / end relief | designed, not inherited | load distribution, edge contact |
| Flank roughness Ra | 0.2–0.4 µm after grinding | film formation, scuffing |
| Effective case depth | 0.15–0.20 × module | pitting and bending capacity |
| Core hardness | 30–40 HRC | root toughness |
| Retained austenite | ≤ 25% near surface | dimensional stability, noise drift |
| Mesh noise, first three parts | limit at mesh frequency ± sidebands | process drift detection |
Two of those rows deserve emphasis. Transmission error under load is the number that correlates with what the customer hears, and it depends on tooth flank modification as much as on accuracy grade — which means it has to be designed into the drawing, not inspected afterwards. Retained austenite is the quiet killer of EV gears: a surface layer that transforms slowly in service changes dimensions by microns over the first few thousand kilometres, and a gear that was silent at end of line starts to whine in the customer's car.
It is worth calculating what a programme actually requires, because the answer explains the quotations. Take a reducer with two ground gears, one at 24 teeth and one at 72. Generating grinding of a helical gear of this size typically runs 6 to 12 minutes per part including dressing and load, so a set is roughly 15 minutes of grinding time. At 200,000 vehicles a year, that is 50,000 grinding hours.
A generating grinding machine running three shifts, 300 days a year at 85 percent availability delivers about 6,100 hours. So the programme needs eight dedicated grinding machines — and that is before scrap, before the second gear set some platforms carry, and before the supplier's other customers. This is why capacity, not price, is the term to negotiate three years ahead.
Three contractual moves change the position materially. First, replace the spot order with a rolling twelve-month forecast plus a booked capacity band: a supplier who has reserved hours for you will protect them, and a supplier who has not will serve whoever calls first.
Second, qualify a second source now rather than during a shortage. Tooling alone — forging dies and grinding fixtures — takes six to ten weeks, so a qualification started when the shortage hits completes after the line has already stopped. Dual sourcing costs a few percent more per piece and buys a great deal of schedule insurance.
Third, agree the document set before the first shipment, not after the first complaint: case depth and core hardness measured at fixed positions, single-flank or mesh-noise results on the first three parts, and lot traceability from steel heat number to grind lot. These three documents are also the fastest way to detect the slow process drift that produces a noise complaint six months later.
The risks that actually hurt EV gear programmes are rarely metallurgical. They are: a drawing frozen too late, which pushes die and fixture work into the critical path; a single grinding source with no reserved capacity; a specification that names accuracy grade but not transmission error, so nobody can arbitrate a noise dispute; and a change of heat-treat recipe between PPAP and series production, which is invisible until the noise audit.
A buying team that fixes those four issues will still face the 12 to 16 week reality, but it will face it with a schedule that holds. That is a better outcome than a three-week saving on piece price that arrives with a whine at 80 km/h.
