机器人关节件的失效，十次里有六次不是断，是尺寸跑了。
件没裂、没变形，装上去间隙就不对，噪音上来了，精度掉了。拆下来一测，尺寸比图纸大了几十微米。
很多人这时候去查模具、查注塑参数，查了一圈，其实问题在材料吸水。
这篇只讲一件事：为什么机器人关节用尼龙，调湿要排在强度前面。
开篇先讲个现场
上个月处理了一起很有代表性的客诉。做谐波减速器配套的客户，输出端的尼龙衬圈装到整机上之后，间隙时大时小，批次间波动明显，装配线上按一个固定的补偿量调，三天两头要返工。客户第一反应是料不行，要求换料。
我们让客户把三个批次的件连同注塑记录寄过来，测试结果出来：三批料的牌号和性能完全一致，差异全在含水量——一批烘得透，两批没烘到位，而这两个批次的件是在雨季生产的。
这就是关节精密件的特殊性：普通结构件差个零点几毫米没人管，关节件的间隙按丝算，吸湿带来的尺寸漂移直接吃掉设计余量。选料在这个场景里反而退居其次，状态管理升到第一位。
这篇就围绕这个特殊性展开。先讲为什么关节件的第一敌人是尺寸而不是载荷，再把调湿这件事从原理到参数讲透——很多厂听说过调湿，但不知道它同时还是一道稳定工序。然后是谐波减速器周边的塑化边界、三类精密件的材料方向、尺寸链上的五个坑。
给机器人传动团队的一句话：读完这篇，把调湿写进你们的作业指导书，比换十次料都管用。
一、关节件的第一敌人，是尺寸，不是载荷
先说清一个反常识：关节件的载荷，通常不是最难的部分。
因为关节尺寸大、壁厚厚，做到扛得住并不难。真正难的是——在百万次运动里，让间隙一直保持在设计范围里。
而尼龙会吸水。吸水就涨，涨了就改变间隙。
PA6 饱和吸水率约 8-10%- PA66 约 8-9%- PA46 饱和吸水率更高，约 13% 量级- PA6T / PA9T 属于半芳香族，吸水率低，PA9T 大约只有 PA46 的十分之一
尺寸变化量级：吸湿饱和后，未做处理的尼龙件尺寸变化可以到 0.3-0.6% 量级。
一个 φ30mm 的件，0.4% 就是 0.12mm。这对普通结构件无所谓，对精密传动是事故。
选关节件，先问"这个件在什么湿度环境下工作"，再问"强度够不够"。这句话很多人是反着问的。
二、调湿：让件在出厂时就"吸够水"
调湿（也叫加湿处理）就是在件出厂前，主动让它吸到接近使用环境的含水量。
逻辑很简单：
刚注塑出来的件是"干态"，尺寸最小- 放到正常环境里会慢慢吸水，尺寸逐渐变大- 如果按干态尺寸交付，客户装好之后件自己还会继续涨——间隙自己就变了
所以精密件的正确流程是：注塑 → 退火 → 调湿 → 复测 → 交付。
调湿的几个关键点：
| 项 | 常规做法 | 说明 |
|---|
| 方式 | 热水浸泡 / 恒温恒湿箱 | 浸水快，箱控稳 |
| 温度 | 70-90℃ 水浴 | 温度高则快，但需防变形 |
| 时间 | 视壁厚，数小时至数十小时 | 壁厚决定扩散时间，不是看件大小 |
| 目标 | 接近使用环境的平衡含水率 | 南方与北方、室内与室外，目标不同 |
| 验收 | 调湿后复测尺寸 | 干态尺寸只是过程数据 |
这里有个内行细节：壁厚决定调湿时间，不是重量。
同一个件，壁厚从 2mm 加到 4mm，水分扩散到底部的时间大约变成 4 倍（扩散时间约与厚度的平方成正比）。所以调湿时间必须按最厚截面算，不是按整件平均。
很多"调湿了还是不行"的案例，都是按平均壁厚估时间，厚的地方根本没吸透。
怎么判断调湿到不到位：称重最实在
比"泡够时间"更可靠的判断方式，是称重。
按目标含水率算出一个重量区间，抽几件称，落在区间内就算达标。时间只是过程，含水量才是结果。
一个经验口径：尼龙件在 23℃、50%RH 环境下的平衡含水率，PA66 大约 2-2.5%，PA6 大约 2.5-3%。调湿的目标是把它做到接近使用环境的平衡含水率，而不是"泡到饱和"。
泡到饱和反而有风险——饱和状态的尺寸是最大的，如果使用环境偏干燥，件会自己往外放水，尺寸再缩回去。精密件的尺寸，要的是"和使用环境一致"，不是"最大"。
一个工况对照：同一个齿轮件，在干燥空调房里放置和使用，与在南方无空调车间里使用，平衡含水率可以差一个百分点以上。对应到 φ30mm 的件，就是 0.03-0.05mm 的尺寸差——正好是精密件的公差量级。
三、谐波减速器周边：哪些件能塑化
谐波减速器是机器人关节的核心，结构上是三大件：柔轮、刚轮、波发生器。
先把结论放前面：
| 部件 | 能否塑化 | 说明 |
|---|
| 柔轮本体 | 不适合 | 靠弹性变形传扭，超千万次疲劳，微米级形变控制，超出热塑性尼龙稳定区间 |
| 刚轮（内齿圈） | 谨慎 | 有研究方案，但对齿面精度与长期尺寸稳定要求极高 |
| 波发生器轴承位 | 不适合 | 接触应力高 |
| 减速器外壳 / 端盖 | 适合 | 承力但非精密接触，GF30 方向 |
| 关节模组壳体 | 适合 | 主力塑化位置 |
| 关节内的连接支架 | 适合 | 高刚性方向 |
| 编码器座 / 传感器支架 | 适合 | 尺寸稳定优先 |
| 电机端盖 / 定子包胶件 | 适合 | 耐温 + 绝缘 |
| 丝杠螺母 / 行星齿轮（低速级） | 看工况 | 低速大扭矩可尝试，高速级不建议 |
正确理解这张表：能塑化的是"外围件"，不是"核心传动件"。
一次性把整套关节都换成塑料，是把可靠性押在材料上；只换外壳、支架、端盖，是把成本省在安全的地方。这两个动作，风险差一个量级。
一个必须说清的锐度判断
关节里的塑料件，做的应该都是"结构任务"，不是"传动任务"。
传动任务（柔轮、高精度丝杠、高速齿轮）需要金属，短时间内没有替代方案。谁跟你说关节可以整体塑化，都在替你承担他没有资格承担的风险。
四、三类精密件的材料方向
| 件类 | 树脂方向 | 改性体系 | 为什么 |
|---|
| 关节模组壳体 | PA66 / PA6T | GF30 + 热氧稳定 | 刚性 + 耐温 + 尺寸稳 |
| 精密小齿轮（低速级） | PA46 / PA66 | 耐磨 + 低吸水体系 | 流动性好能打满齿形，耐磨 |
| 高精度支架 / 编码器座 | PA6T / PA9T | GF15-30 | 低吸水、尺寸稳定优先 |
| 电机端盖 / 绝缘件 | PA66 / PA6T | 阻燃 + 耐温 | 绝缘 + 耐温 120-150℃ |
| 关节密封 / 缓冲 | PA12 / 增韧 PA66 | 增韧 | 韧性、低吸湿 |
为什么精密件偏爱 PA6T / PA9T？
半芳香族结构让吸水率显著低于 PA66 和 PA46。吸水率低 = 湿度不敏感 = 尺寸稳定 = 少做调湿也能稳住。
代价是熔点高（PA6T 熔点在 320℃ 以上）、加工温度高、成本高。所以它的使用逻辑是：越精密、越怕涨的件，越值得用贵料。
一句话：能用 PA66 解决的，别上 PA9T；但一旦公差做到 ±0.05mm 以内，贵料的钱是省不下来的。
调湿工艺的三个参数
调湿这件事实操起来就三个参数：温度、时间、判定。温度上，沸水直煮最快但内应力大的件容易煮出问题，稳妥做法是九十度上下的热水分段调；时间上，不要凭经验说煮几小时，按壁厚算——每毫米壁厚给足对应的小时数，厚壁件宁可过不可欠；
判定上，调没调够不看时间看重量，调湿前后称重，增重率到了规格书给的平台区才算完成。三个参数之外还有一个纪律：调湿后的件要密封存放、尽快装配，在车间敞开放三天的调湿件，等于白调。
把这三参数一纪律写进工艺文件，调湿就从老师傅的手艺变成了任何班次都能执行的标准动作。关节件的间隙稳定，就从这一页文件开始。
五、尺寸链上的五个坑
坑 1：用干态尺寸交付。 前文已述。这是精密件最常见的错，且客户往往在装配后才发现。
坑 2：按平均壁厚算调湿时间。 厚截面没吸透，等于没调。
坑 3：忽略模具收缩率与玻纤取向的关系。 玻纤件纵横收缩率不同，长条件的两个方向尺寸要分别给收缩率。一个收缩率打天下，长件必翘。
坑 4：只测一个批次。 调湿批次之间的含水率差异会直接反映到尺寸公差上。只测一批料的供应商，和只测一批料的采购，风险是一样的。
坑 5：把退火和调湿当一回事。 退火是消除内应力、稳定结晶，调湿是补水分。两件事，顺序不能换——先退火后调湿，反过来做了等于白做。
一个工况案例
一个机器人关节支架件，材料 PA66-GF30，图纸公差 ±0.08mm。
供应商按干态尺寸交付，客户装配时发现间隙偏小、需要额外选配。换成"退火 + 调湿 + 复测交付"之后，装配一次通过，不需要选配。
件没换料，图纸没改，只是把后处理流程补齐了。
六、边界：这些件别动
| 件 | 结论 | 原因 |
|---|
| 谐波柔轮 | 不适合 | 超高周疲劳 + 微米级弹性变形 |
| 高精度丝杠 / 滚柱 | 不适合 | 接触应力、预紧保持 |
| 高速级齿轮（>3000rpm） | 需谨慎 | 温升 + 动平衡 |
| 长期 >150℃ 的关节件 | 需谨慎 | 换 PA46 / PA6T 仍需评估 |
| 主承力连接位 | 需谨慎 | 蠕变长期保持 |
记住一句话：关节里能塑化的位置，都是不参与"力的精密传递"的位置。
行业里的一条实感
这个坑我们自己踩过。
早些年给客户做一批精密齿轮件，下线检测全部合格，按干态数据出的报告，交期也赶上了。结果客户装了两周，反馈噪音上来了，拆下来复测，尺寸超差。
我们当时第一反应是查模具、查注塑参数，查了三天没找到问题。
后来才想明白：件下线是干的，客户那边在南方梅雨天里放了半个月，件自己吸湿涨了。
补做调湿之后复测，尺寸正好落在公差带中间——件从头到尾都是好的，只是"报告给了错的尺寸"。
从那之后我们内部定了一条规矩：精密件交付，一律给调湿后尺寸。干态数据只做过程记录，不上报告。
这条规矩后来救过好几次交期——因为很多"客户投诉尺寸"，其实是"我们自己报告给错了状态"。
读者追问三则
追问一：调湿能不能在整机上做，不拆件？ 不建议。整机调湿受限于装配状态，水汽只能从缝隙渗，件与件之间的调湿程度不均匀，而且整机里的金属件和电子件经不起水热。调湿要在装配前、按件做，这是行业共识，别为了省工序在整机上赌。
追问二：有没有不需要调湿的尼龙？ 有方向：低吸湿改性和长碳链体系把平衡含水率压低一个量级，尺寸漂移随之变小。但要注意，低吸湿牌号的代价通常在刚性和价格上，关节件要不要换，先算尺寸链的账：如果设计余量够，普通增强尼龙加调湿纪律就够；余量本来就被压得很紧，那低吸湿体系的溢价才有意义。
追问三：批量生产时调湿节拍跟不上产能怎么办？ 两个思路：一是把调湿工序前移到注塑厂出厂前，采购合同时就约定调湿态交货并附增重率记录；二是改设计，把对间隙敏感的配合面留给金属件或低吸湿件，尼龙件只承担非精密配合。节拍问题的本质是工艺和设计的接口问题，两端各让一步就顺了。
调湿记录三栏台账
调湿要落地，配一张三栏台账：第一栏记批次参数，烘箱温度、时长、件数、壁厚范围；第二栏记称重结果，调湿前重、调湿后重、增重率，增重率没到平台区的批次标记续调；第三栏记流转去向，调湿完成的件进了哪个订单、密封包装有没有到位、装配日期。
台账每周复盘一次，两个信号要盯：增重率长期到不了平台的，查水温或件内应力；增重率波动大的，查烘箱装载量和翻转纪律。台账坚持三个月，你会拿到一样意外收获——自己产品的真实平衡含水率数据，这个数比供应商给的通用值准得多，设计尺寸链的时候直接用它。调湿从手艺变成数据资产的路径，就是这么简单。
最后说说和整机厂的沟通话术。关节件供应商最常见的委屈是：我们料没问题，是整机厂设计余量太紧。这句话对，但没用，说出去只会把关系谈僵。更有效的表达是拿数据说话：把三个批次的尺寸漂移实测和吸湿曲线摆出来，告诉整机厂这个漂移是材料物理特性，任何供应商都一样，与其在图纸上留紧公差，不如双方一起把调湿态交货和装配环境的湿度写进协议。
把物理问题转化成流程问题，双方从对立面变成同一条战壕，这才是精密件供应链的成熟相处方式。
结语
机器人关节用尼龙，说到底是两个问题：
一个是"能不能扛"，靠树脂和纤维解决；一个是"稳不稳"，靠吸水率和后处理解决。
前者决定件能不能用，后者决定件能不能一直用。 精密件上，后者比前者更容易翻车。
The failure of robot joint components: six out of ten times it's not a break, but the dimensions are off.
The part didn't crack or deform, but after installation the clearance was wrong, noise increased, and accuracy dropped. After taking it apart and measuring, the dimensions were tens of microns larger than the drawing.
At this time, many people check the molds and injection molding parameters, and after checking around, the problem is actually with the material absorbing moisture.
This article only talks about one thing: why robot joints use nylon, and why moisture regulation is prioritized over strength.
Let's start with a scene.
Last month I handled a very typical customer complaint. The customer who uses harmonic reducers as a complete set found that after installing the nylon bushing on the output end onto the whole machine, the clearance was sometimes large and sometimes small, with significant fluctuations between batches. On the assembly line, adjusting with a fixed compensation amount meant rework was needed every couple of days. The customer's first reaction was that the material was no good and they requested a material replacement.
We asked the customer to send over the parts from three batches along with the injection molding records. The test results came out: the grade and performance of the three batches of material were completely consistent, and the differences were entirely in the moisture content—one batch was thoroughly dried, while the other two batches were not properly dried, and these two batches were produced during the rainy season.
This is the particularity of precision joint components: for ordinary structural parts, a difference of a few tenths of a millimeter doesn't matter, but the clearance in joint parts is measured in threads, and dimensional changes caused by moisture absorption directly eat into the design margin. In this context, material selection actually takes a back seat, while state management becomes the first priority.
This article revolves around this specificity. First, it explains why the primary enemy of joint components is size rather than load, and then thoroughly covers humidity conditioning from its principles to its parameters—many manufacturers have heard of humidity conditioning but don’t know that it is also a stabilizing process. Then it addresses the plasticization boundaries around harmonic reducers, the material orientation of three types of precision components, and the five pitfalls in the dimension chain.
A word to the robot transmission team: After reading this, put humidity control into your work instructions; it's more effective than changing the material ten times.
1. The first enemy of joint components is size, not load.
Let's first clarify a counterintuitive fact: the load on the joint is usually not the most difficult part.
Because the joint is large and the wall is thick, making it strong enough is not difficult. What is truly difficult is keeping the clearance within the design range over millions of cycles.
Nylon absorbs water. When it absorbs water, it expands, and when it expands, it changes the clearance.
PA6 has a saturated water absorption of about 8-10% - PA66 about 8-9% - PA46 has a higher saturated water absorption, about 13% level - PA6T / PA9T are semi-aromatic, with low water absorption, and PA9T is only about one-tenth of PA46
Magnitude of dimensional change: After moisture absorption saturation, the dimensional change of untreated nylon parts can be on the order of 0.3-0.6%.
A φ30mm part, 0.4% is 0.12mm. This doesn't matter for ordinary structural parts, but for precision transmission, it's an accident.
When selecting joint parts, first ask 'In what humidity environment will this part operate?' and then ask 'Is the strength sufficient?' Many people ask these questions in reverse order.
2. Moisture adjustment: Allow the item to "absorb enough water" before leaving the factory
Moisture conditioning (also called humidification treatment) is the process of actively allowing the item to absorb a moisture content close to that of the usage environment before it leaves the factory.
The logic is very simple:
The part just molded is in a 'dry state', with the minimum dimensions. When placed in a normal environment, it will gradually absorb moisture and the dimensions will increase. If delivered according to the dry-state dimensions, after the customer assembles it, the part will continue to expand on its own — and the clearance will change by itself.
So the correct process for precision parts is: injection molding → annealing → moisture conditioning → re-inspection → delivery.
Several key points about humidity control:
| item | Conventional practice | Explanation |
|---|
| Method | Hot water immersion / Constant temperature and humidity chamber | Water absorption is fast, and box control is stable |
| Temperature | 70-90℃ water bath | Higher temperature speeds it up, but deformation must be prevented |
| Time | Depending on the wall thickness, from several hours to several tens of hours | The wall thickness determines the diffusion time, not the size of the part. |
| Target | Moisture content balanced close to the usage environment | South and North, indoors and outdoors, different goals |
| Acceptance | Re-measure dimensions after humidity adjustment | The dry-state dimensions are just process data |
Here's an insider detail: the wall thickness determines the humidity adjustment time, not the weight.
For the same part, when the wall thickness increases from 2mm to 4mm, the time for moisture to diffuse to the bottom becomes about 4 times longer (diffusion time is approximately proportional to the square of the thickness). Therefore, the conditioning time must be calculated based on the thickest section, not the average of the whole part.
Many cases of 'moisture adjustment still doesn’t work' are all estimated based on the average wall thickness, and the thick parts are not absorbed at all.
How to determine if the humidity adjustment is done right: weighing is the most reliable
A more reliable way to judge than 'soaking long enough' is to weigh it.
Calculate a weight range based on the target moisture content, take a few samples and weigh them, if they fall within the range, it's considered qualified. Time is just the process, moisture content is the result.
An empirical guideline: The equilibrium moisture content of nylon parts at 23°C and 50% RH is approximately 2-2.5% for PA66 and about 2.5-3% for PA6. The goal of conditioning is to bring it close to the equilibrium moisture content of the usage environment, rather than "soaking it to saturation."
Soaking to saturation actually carries risks—the dimensions are largest when saturated, and if the usage environment is relatively dry, the part will release water on its own, causing the dimensions to shrink back. For precision parts, what matters is having dimensions that 'match the usage environment,' not being 'maximized.'
A working condition comparison: For the same gear component, placing and using it in a dry air-conditioned room versus using it in a workshop in the south without air conditioning, the equilibrium moisture content can differ by more than one percentage point. For a φ30mm part, this corresponds to a size difference of 0.03-0.05mm — just exactly the tolerance range of precision parts.
3. Peripheral parts of the harmonic reducer: Which components can be plastified
The harmonic reducer is the core of the robot joint, structurally consisting of three main components: the flexible wheel, the rigid wheel, and the wave generator.
Put the conclusion first:
| Component | Can it be plasticized? | Explanation |
|---|
| Flexible wheel body | Not suitable | Transmits torque through elastic deformation, over ten million cycles of fatigue, micron-level deformation control, exceeds the stable range of thermoplastic nylon |
| Sun gear (internal ring gear) | Cautious | There is a research plan, but it requires extremely high tooth surface accuracy and long-term dimensional stability. |
| Wave generator bearing position | Not suitable | High contact stress |
| Gearbox Housing / End Cover | Suitable | Bearing force but not precision contact, GF30 direction |
| Joint module housing | Suitable | Main plasticizing position |
| Intra-articular support bracket | Suitable | High rigidity direction |
| Encoder Mount / Sensor Bracket | Suitable | Dimensional stability priority |
| Motor End Cover / Stator Encapsulation Part | Suitable | Temperature resistant Insulation |
| Lead Screw Nut / Planetary Gear (Low-Speed Stage) | Check operating conditions | Low speed with high torque can be tried, high-speed levels are not recommended |
Correctly understand this table: the parts that can be plasticized are the 'peripheral parts,' not the 'core transmission parts.'
Replacing the entire set of joints with plastic at once bets the reliability on the material; only replacing the outer shell, bracket, and end cap saves costs in terms of safety. The risk between these two actions differs by an order of magnitude.
A sharpness judgment that must be clarified
The plastic parts in the joints should all be designed for 'structural tasks', not 'transmission tasks'.
Transmission tasks (flexible wheels, high-precision screws, high-speed gears) require metal, and there is no alternative in the short term. Whoever told you that joints can be entirely plasticized is taking on the risks they are not qualified to bear on your behalf.
4. Material Directions for Three Types of Precision Components
| item type | Resin direction | Modified system | Why |
|---|
| Joint module housing | PA66 / PA6T | GF30 Thermo-oxidative stability | Rigid Temperature-resistant Dimensionally stable |
| Precision Small Gear (Low-Speed Stage) | PA46 / PA66 | Wear-resistant, low water absorption system | Good liquidity, can fill the full tooth shape, wear-resistant |
| High-precision bracket / encoder mount | PA6T / PA9T | GF15-30 | Low water absorption, dimensional stability prioritized |
| Motor End Cover / Insulating Parts | PA66 / PA6T | Flame retardant Temperature resistant | Insulation Temperature Resistant 120-150℃ |
| Joint Sealing / Cushioning | PA12 / Toughened PA66 | Toughening | Toughness, low hygroscopicity |
Why do precision parts prefer PA6T / PA9T?
The semi-aromatic structure makes the water absorption significantly lower than PA66 and PA46. Low water absorption = not sensitive to humidity = dimensionally stable = can remain stable with less moisture conditioning.
The price is high melting point (PA6T melting point is above 320℃), high processing temperature, and high cost. Therefore, the logic of its use is: the more precise and the more sensitive to expansion the part is, the more worth it is to use expensive material.
In a nutshell: if PA66 can do the job, don’t use PA9T; but once the tolerance is within ±0.05mm, there’s no way to save money on expensive materials.
Three parameters of the humidity control process
There are only three parameters when it comes to moisture conditioning: temperature, time, and judgment. In terms of temperature, boiling water directly is the fastest but parts with high internal stress are prone to problems, so a safer approach is to adjust in stages with hot water around ninety degrees Celsius; regarding time, do not rely on experience and say 'boil for a few hours,' calculate based on wall thickness—allocate the corresponding number of hours for each millimeter of wall thickness. For thick-walled parts, it is better to overdo it than underdo it.
In terms of judgment, whether it has been conditioned enough is determined by weight, not time. Weigh before and after conditioning, and only when the weight gain reaches the range specified in the specification is it considered complete. Besides these three parameters, there is another rule: conditioned items must be stored sealed and assembled as soon as possible. If conditioned items are left exposed in the workshop for three days, it’s as if the conditioning was done for nothing.
Incorporate these three parameters into the process document, and humidity adjustment changes from a master craftsman's technique to a standard action that any shift can perform. The stability of the joint clearance starts from this page of the document.
5. The Five Pitfalls in the Dimensional Chain
Pitfall 1: Deliver using dry-state dimensions. As mentioned earlier, this is the most common mistake with precision parts, and customers often only discover it after assembly.
Pitfall 2: Calculating moisture adjustment time based on average wall thickness. Thick sections don't absorb fully, which is equivalent to not adjusting at all.
Pitfall 3: Ignoring the relationship between mold shrinkage and fiberglass orientation. Fiberglass parts have different shrinkage rates in the longitudinal and transverse directions, so the dimensions in the two directions of a long part must be given separate shrinkage rates. Using a single shrinkage rate for everything will cause the long part to warp.
Pitfall 4: Testing only one batch. The difference in moisture content between batches with adjusted humidity will directly affect dimensional tolerances. Suppliers who test only one batch and purchasers who test only one batch face the same risk.
Pitfall 5: Treating annealing and humidity adjustment as the same thing. Annealing eliminates internal stress and stabilizes crystallization, while humidity adjustment adds moisture. These are two different things, and the order cannot be swapped — anneal first, then adjust humidity; doing it the other way around is equivalent to doing nothing.
A working condition case
A robotic joint bracket, material PA66-GF30, drawing tolerance ±0.08mm.
The supplier delivered according to the dry dimensions. When the customer assembled it, they found the gap was too small and required additional optional parts. After switching to 'annealing, moisture conditioning, and re-measurement before delivery,' the assembly passed in one go without needing optional parts.
The parts weren't changed, the drawings weren't modified, only the post-processing procedure was completed.
6. Borders: Do not move these parts
| piece | Conclusion | Reason |
|---|
| Harmonic Drive | Not suitable | Ultrahigh cycle fatigue Micron-level elastic deformation |
| High-precision lead screw / roller | Not suitable | Contact stress, preload retention |
| High-speed gears (>3000 rpm) | Need to be cautious | Temperature Rise Dynamic Balancing |
| Joints exposed to over 150℃ for long periods | Need to be cautious | Replacing PA46 / PA6T still needs evaluation |
| Main load-bearing connection point | Need to be cautious | Creep long-term maintenance |
Remember one thing: any position in a joint that can be molded is a position that does not participate in the 'precise transmission of force.'
A real feeling in the industry
We have fallen into this trap ourselves.
A few years ago, we made a batch of precision gear parts for a customer. All passed the offline inspection, and the report was issued based on dry data. The delivery time was also met. However, after the customer installed them for two weeks, they reported increased noise. When we disassembled and re-tested, the dimensions exceeded the tolerance.
Our first reaction at the time was to check the molds and the injection molding parameters, and we checked for three days without finding the problem.
Later I realized: the item was done offline, and the client kept it in the southern rainy season for half a month, so the item absorbed moisture and swollen on its own.
After redoing the humidity adjustment and retesting, the dimensions fell right in the middle of the tolerance range — the part was good from start to finish, it's just that the 'wrong dimensions were reported'.
Since then, we set an internal rule: for the delivery of precision parts, they must all be given at the dimension after humidity adjustment. Dry-state data is only recorded for process purposes and is not included in reports.
This rule later saved several delivery deadlines—because many 'customer complaints about dimensions' were actually 'us reporting the wrong status ourselves'.
Three Reader Inquiries
Follow-up question 1: Can humidity control be done on the whole machine without disassembling parts? Not recommended. Humidity regulation of the whole machine is limited by assembly conditions; moisture can only seep through gaps, causing uneven humidity control between parts, and the metal and electronic parts inside the machine cannot withstand water heat. Humidity regulation should be done before assembly and by piece; this is an industry consensus. Don't gamble on the whole machine just to save steps.
Follow-up question 2: Is there nylon that doesn't require humidity control? There is a direction: low moisture absorption modification and long carbon chain systems can lower the balanced moisture content by an order of magnitude, reducing dimensional drift. But note that the cost of low moisture absorption grades usually lies in rigidity and price. Whether to replace joint parts depends on the size chain: if the design allowance is sufficient, ordinary reinforced nylon with humidity regulation discipline is sufficient; If the margin is already tightly squeezed, then the premium of a low moisture absorption system is meaningful.
Follow-up Question 3: What if the humidity control cycle can't keep up with production capacity during mass production? Two approaches: one is to move the humidity control process forward to before the injection molding plant leaves the factory, and the procurement contract should specify moisture control delivery with added weight rate records; the other is to modify the design by leaving gap-sensitive mating surfaces for metal or low moisture absorption parts, while nylon parts only handle non-precision fits. The essence of the cycle timing issue is the interface between process and design; just make concessions at both ends and it works smoothly.
Three-column Humidity Control Record Ledger
Humidity Control Must Be Implemented, Prepare a three-column ledger: the first column records batch parameters, oven temperature, duration, number of pieces, and wall thickness range; The second column records weighing results, including pre-humidity adjustment weight, post-adjustment weight, weight gain rate, and batch markings that have not reached the platform area for further adjustment; The third column records flow destination, which order received the completed moisture adjustment items, whether sealing and packaging are in place, and assembly date.
Review the ledger once a week, focusing on two signals: if the gain rate does not reach the platform for a long time, check water temperature or internal stress of the pieces; If the weight gain rate fluctuates greatly, check the oven loading volume and flipping discipline. If you stick to the ledger for three months, you'll gain an unexpected benefit—your product's real balance moisture content data, which is much more accurate than the general value provided by suppliers, and you can use it directly when designing the dimensional chain. The path from crafting to data asset for humidity control is that simple.
Finally, let's talk about communication language with OEMs. The most common grievance from joint suppliers is: our materials are fine, but the manufacturer's design allowance is too tight. This is true, but it's useless—saying it out loud will only make the relationship stiff. A more effective way is to use data to speak: lay out the dimensional drift measurements and moisture absorption curves for three batches, telling the OEM that this drift is a physical property of the material, and that this drift is the same for any supplier. Rather than leaving tight tolerances on the drawings, it's better for both parties to write the humidity adjustment delivery and assembly environment humidity into the agreement.
Turning physical problems into process issues, turning both sides from adversaries into a single trench—this is the mature way to coexist in the precision parts supply chain.
Conclusion
Robots use nylon for joints, which ultimately boils down to two issues:
One is "can it withstand it," which relies on resin and fibers; The other is "stability," which relies on water absorption and post-processing.
The former determines whether the parts can be used, while the latter determines whether the parts can be used indefinitely. In precision parts, the latter is more likely to fail than the former
