I've spent six years on the buying side of a small company that builds e-bikes and automated motion equipment. I sign about $450,000 worth of motor and drivetrain purchase orders every year—maybe closer to $420,000 when you exclude prototype orders. I've also documented every warranty return in our cost tracking system, so I get to see where those orders went wrong.
If you ask me, the phrase 'an ebike Bafang motor is starting to make noise' is usually misleading. The Bafang M630 motor rarely dies. What dies is a component that was selected by price and hidden inside a gearbox, a bearing housing, or a linear slide.
The Surface Problem: The Motor Gets All the Blame
Most buyers search for a bafang-motor kit, compare peak power and torque, and click buy. The purchase decision ends at the motor. But the performance and reliability decision starts after the motor arrives.
I remember taking apart one customer's cargo bike after they told us their Bafang M630 motor was knocking hard. The motor windings and controller were fine. The noise came from an auxiliary right-angle gearbox whose straight bevel gears had worn unevenly, and the gearbox shaft was supported by a ball bearing that had a few thousandths of play. The motor's torque did exactly what it was designed to do: it found the weakest link.
What's a Ball Bearing? A Question Worth Revisiting
What's a ball bearing? Most people would say it's a metal ring filled with little balls. That description gets the shape right but misses the part that matters: specification. Two parts can both look like a 608 bearing, but one is made to a known tolerance class and one is only shaped like the right part.
On a high-torque motor, the bearing does more than support a wheel. It holds a shaft in alignment under changing loads. If the internal clearance is too loose, you get vibration and noise. If it is too tight, you get heat and premature wear. The difference between those outcomes is not visible in the product photo.
The Real Cause: Torque Is an Auditor
Here is what took me years to understand. The Bafang M630 motor can push much more torque than a small part is expected to see. In a standard bicycle, a bearing may only see the rider's weight and road shock. In a high-power assisted bicycle or an industrial linear actuator, the bearing also sees torque reaction, side loads, and constant vibration. What I mean is that torque will eventually audit every component in the power path. If the audit finds a cheap part, that part fails.
The same lesson applies to linear bearing guide components. On a servo axis, a linear guide supports a moving carriage and keeps it straight. If the rail or truck doesn't match the load and duty cycle, the first sign is small play. Then the axis vibrates. Then the whole product starts producing scrap. A procurement person sees the rejection report later. By then, the motor is usually the component that kept working.
Straight Bevel Gears: The Loud Failure No One Wants
Straight bevel gears are common in right-angle drives where the motor axis is perpendicular to the output axis. They have teeth cut in a straight line across the cone's face, which makes them simpler and cheaper than spiral bevel gears. They also carry the load with a smaller contact area. If you buy straight bevel gears from a supplier who gives you no hardness specification, no quality standard, and no backlash number, you are gambling. I've seen that gamble win on low-power test benches. On an actual torque-rich bike, the gear wears, backlash grows, and the motor starts clacking. The motor didn't suddenly break. The contact patch became too small to carry the load.
What a Cheap Part Actually Costs
This is the section that matters to me. Let's say you have an order for 500 units. A premium bearing costs $2.10 each. An unbranded bearing that looks identical costs $0.58. Saving: $1.52 per unit, $760 total. You approve it to win the contract. The bearing is not terrible. It just isn't made to the same standard. Two months later, 23 units show up with noise complaints. Each failure costs around $115 by the time you pay labor, shipping, and some form of refund. That is around $2,600 of direct cost. It does not include the time spent arguing with the customer or the lost sale. The 'cheap' part cost more than three times what it saved.
The Damage Is to Your Brand, Not Just the Motor
To be fair, some cheap components survive. Not every application needs premium parts. A low-speed, hobby-use bike with straight bevel gears moving modest torque may run fine for years. I get why budget-tier products use budget parts—prices are real and margins are thin. But the perception problem appears when a part fails in the field. The customer does not say a $0.58 bearing failed. They say that the brand makes noisy e-bikes. The expensive Bafang motor doesn't get to testify.
In my opinion, quality perception is not a soft marketing topic. It has a procurement consequence. When we went back to traceable, higher-grade bearings and straight bevel gears, the motor was unchanged. Yet our after-sales noise complaints dropped by perhaps 20 to 30 percent. I can't prove a direct causal link with numbers that small, but the direction was consistent over two production batches.
The Short, Unsexy Solution
I don't have a clever fix. I have a routine.
- Get a specification for every bearing and gear: load rating, tolerance class (ABEC/ISO P-class), sealing, material. If a supplier can't provide it, treat the part as a gamble.
- For linear bearing guide components, ask for preload and rail straightness before you order.
- Test one sample at a load higher than you expect in the field. A no-load test bench does not uncover hidden quality differences.
- Keep two budgets: the purchase order budget and the warranty budget. They talk to each other more than you think.
If you are building around a bafang m630 motor, the motor price is the easy part. The bearings and gears will decide whether your product is remembered as smooth or as one that started making noise.
So, what's a ball bearing? After six years of purchase orders and warranty returns, I'd say it's a quality decision. Get it wrong and the motor gets the blame. Get it right and nobody has to think about the motor at all—and that is exactly how a brand should feel.
Discuss this article with engineering