Precision reducer: Why are these three major thresholds so difficult to overcome?

Writer：tilireducer Time：2026-10-29 10:10:26 Browse：1℃

As we discussed before, after using a robot for a period of time, reaching out to pick up a cup on the table may cause a deviation in the contact between the fingers and the cup.

The reason for this situation may be related to the reducer inside, which is responsible for converting the power of the motor into force that can be used for specific tasks.

The key may lie in the process of conversion, where the accuracy of the conversion directly affects whether there will be any deviation when the finger touches the cup.

However, what are the thresholds that need to be overcome in order to produce a good reducer like this?

Next, let's talk about this topic.

Materials

First look at the materials, after all, with good materials, it is possible to create good things.

The flexible wheel (matched with the rigid wheel) in the harmonic reducer has a wall thickness of only about 0.3 to 0.5 millimeters. It is stretched open from the inside by an elliptical cam during operation, with thousands of thrusts per minute and continuous operation for tens of thousands of hours without cracking or permanent deformation. This requires that the materials used to make it not only have good elasticity and can bounce back smoothly after repeated bending, but also have sufficiently hard tooth surfaces that are not easily worn.

The truly difficult parts are still hidden in those tiny details. If there are micrometer sized impurities in the steel, such as alumina particles, they will become the beginning of fatigue cracks, which will be gradually amplified during repeated operation. The top grades of Japanese steel have a high level of impurity control, but there is a gap in the purity of local materials in the past. This extremely small difference is almost indistinguishable in a single rotation, but after thousands of hours of continuous operation, a gap will appear. In the past, the lifespan of local gearbox test benches was mostly between 6000 and 8000 hours, while imported products could reach 10000 or even 20000 hours or more. However, this gap is narrowing. Industry data shows that under standard operating condition bench tests, the lifespan of mainstream local products has been extended to 8000 to 12000 hours, with some high-end models exceeding 12000 hours.

So, the difficulty in the material barrier is not enough. Poor purity, poor grain uniformity, may not be visible in a short period of time or during factory testing; But if used for a period of time, the performance gap will gradually emerge.

In addition, even if the samples can achieve excellent lifespan indicators, there is still the challenge of mass production, whether every flexible wheel produced in bulk can maintain the same excellent performance.

Processing

After discussing the materials, let's now talk about processing.

The cycloidal gear in the RV reducer has a very precise tooth profile curve. If there is a slight deviation in the tool path during machining, the tooth profile may not achieve ideal meshing, causing jamming, noise, or increased backlash. According to relevant research calculations, the contribution weight of cycloidal gear profile error in the overall transmission error can reach 35% to 45%.

If you want to reduce this tooth profile error, you need to rely on the precision of the grinder and the grinding wheel dressing process. A large number of Japanese leading enterprises provide self-made specialized grinding machines, which can stably achieve processing control of 2 to 3 microns; In the past, locally available general-purpose grinding machines could generally only achieve a size of 5 to 8 microns. The difference in processing capability is ultimately reflected in the accuracy of the reducer. The return clearance of Japanese benchmark products can be less than 1 arc minute, and the transmission error is controlled within the range of 30 to 50 arc seconds; Most local products are concentrated in a hysteresis of 1 to 3 arc minutes and a transmission error of 50 to 80 arc seconds. One arc minute is one sixtieth of a degree, which may not seem significant based on the number alone. However, after prolonged use, this deviation will continue to accumulate, ultimately leading to significant drift in the rhythm and processing yield of the entire production line.

In addition, there is another issue with the processing threshold, which is batch consistency. Creating a high-precision cycloidal gear in the laboratory and continuously producing a batch of cycloidal gears with similar performance in the workshop are two different concepts. During the grinding process, the grinding wheel will continue to wear, the processing temperature will fluctuate, and the internal stress caused by workpiece clamping will also change. If any variable is not properly controlled, it will cause the distribution of finished product accuracy to diverge.

Industry reports show that the average process capability index (CPK) of top local manufacturers RV reducers is approximately 1.03, while international first tier manufacturers usually require a mass production admission of 1.33 or above.

This roughly means that making a qualified sample is not the most difficult, the real difficulty is to ensure that the performance of each part produced in bulk on the assembly line is sufficiently close.

Testing

If the first two thresholds are about whether the product can be made, then testing is probably about what kind of product is considered qualified.

In the testing indicators of precision reducers, there are backlash and transmission errors. Measuring the return clearance requires a high-precision angle encoder and torque sensor with more than 17 bits, and the cost of a comprehensive measurement equipment can reach several hundred thousand yuan. In terms of lifespan testing, the rated lifespan of a reducer often reaches 6000 to 10000 hours, and even with an accelerated lifespan testing scheme, the testing cycle can last for thousands of hours; The test bench also needs to simulate torque fluctuations, temperature changes, and vibration environments under real working conditions. There are not many local manufacturers with the ability to conduct large-scale and long-term life verification. This is not a shortcoming that can be filled by simply purchasing a few instruments. It requires sufficient test samples, continuously operating test benches, and a long-term accumulated failure database.

In addition, there is also the issue of uniform standards. For a long time, there was a lack of unified and authoritative testing standards in the industry for key parameters such as transmission accuracy, backlash, and torsional stiffness. Testing conditions and methods varied among different companies, making it difficult to directly compare data from different manufacturers. This situation has undergone new changes in 2026, with new industry standards establishing a unified testing framework for parameters such as torsional stiffness, backlash, backlash, transmission error, and repeatability accuracy. It takes some time to digest and adapt standards from paper to workshop production lines, but at least it indicates that the industry is moving towards using the same standard for evaluation.

Write it at the end

So, the threshold of precision reducers cannot be solved by any single breakthrough. The material determines the upper limit of the lifespan, processing determines whether the accuracy can be stably achieved, and testing provides a ruler. The three interact with each other, and if the materials are not up to standard, it is impossible to achieve batch consistency in processing; The testing standards are not unified, the actual level of processing is difficult to objectively compare, and there is a lack of reliable data feedback for material improvement.