机器人要不要用塑料件，这个问题在两年前还有争议。今年再看，答案已经变了——不是"能不能用"，是"哪个部位先用、用哪一种"。
行业里有个被反复引用的经验数：整机自重每减掉 1 公斤，末端能多带走接近 0.8 公斤的负载。
金属件密度大、加工贵、还得持续润滑。改性尼龙密度只有铝合金的一半左右，纤维增强之后比强度能追上部分金属，还自带自润滑。
但机器人身上不是每个件都能换。这篇把机器人用尼龙拆成五个部位，逐个讲清楚。
开篇先讲个现场
上个月行业展会上，我在一家整机厂的展台前站了半小时。展台上的样机来回做动作，围观的观众看的是灵巧的手指，我看的是它小臂上那块白色的壳体——边缘有一道细小的浇口痕。散场后找到他们的结构工程师聊，对方说得很直白：这台样机身上已经有一百多个塑料件，但齿轮箱和关节轴承还是金属的，谁也不敢先动。
这段对话浓缩了当下机器人用尼龙的真实状态：外壳结构件已经大面积塑化，传动件半开半合，核心运动副还在观望。三道防线，推进速度完全不一样。而我们这一年接到的机器人询盘，问法也在变——去年问的是塑料能不能做，今年问的是这个部位用哪种尼龙、耐磨和刚性哪个优先。
这篇就把五个部位逐一拆开。先建立三种力学状态的分类框架，这是理解机器人选型的钥匙；然后按部位给材料地图：关节和齿轮为什么耐磨排在强度前面、骨架件为什么 GF30 是起点不是终点、外壳走线为什么反而不要刚性。
再讲五个坑和一条边界——哪些件现在还不适合上尼龙。做机器人结构设计的读者，建议对照着自己手上的 BOM 读，逐个部位过一遍。
一、先分清：机器人身上的件，分三种力学状态
同样是"机器人用尼龙"，选料逻辑可能完全相反。因为件分三种状态：
第一种 · 高频往复运动件。 关节齿轮、轴承保持架、丝杠螺母、灵巧手的小齿轮。它们的核心矛盾不是"能不能扛住"，是"磨不磨得起"。
第二种 · 长期承力结构件。 大腿连杆、小腿骨架、机身承重框。核心矛盾是刚性和疲劳，还要反复经受冲击。
第三种 · 外覆、走线、缓冲件。 外壳、线束护套、足底缓冲、仿生皮肤基底。核心矛盾是韧性、耐候和成型。
这三种件，材料方向根本不是一套。 拿承力件的思路去选齿轮，会得到一个"很刚但很快磨平"的件。
先定件的力学状态，再谈树脂。这是机器人选料的第一句话。
二、五个部位的材料地图
把机器人从上到下拆开，能换尼龙的位置大致是这五处：
| 部位 | 常见基材方向 | 改性体系 | 关键指标 |
|---|
| 关节齿轮 / 保持架 | PA66、PA46 | 耐磨自润滑 | 摩擦系数、PV 值、疲劳寿命 |
| 承力连杆 / 骨架框 | PA66、PA6、PA6T | 玻纤 / 碳纤增强 | 弯曲模量、比强度、抗蠕变 |
| 减速器外围壳体 | PA66、PA46、PA6T | GF30 + 热氧稳定 | 尺寸稳定、耐温 120℃+ |
| 灵巧手结构件 | PA66、PA12 | 增强 + 增韧复合 | 小尺寸精度、韧性 |
| 线束护套 / 足底缓冲 | 增韧 PA6、PA12 | 增韧、耐候 | 低温不脆、耐弯折 |
| 外壳 / 仿生皮肤基底 | 增韧 PA6、PA/ABS | 增韧、易着色 | 抗冲击、可喷涂、低气味 |
这张表里最值得看的是第一行和第二行的差别。
同样是 PA66，做齿轮要往耐磨体系走，做连杆要往高玻纤走。两个方向加的东西不一样，甚至互相冲突——耐磨改性往往牺牲一点刚性，高玻纤又会放大齿轮的对偶磨损。
一句话结论：机器人用尼龙，先按"磨"还是按"扛"分家，再谈牌号。
三、关节和齿轮：耐磨要排在强度前面
这是机器人最容易被选错的地方。
很多人第一步就问"拉伸强度多少"。问错了。 关节齿轮的失效很少是"被拉断"，绝大多数是磨损失效和疲劳点蚀——材料没断，精度先没了。
看三个数，而不是一个数：
① 摩擦系数。 加 PTFE 或二硫化钼的耐磨尼龙，摩擦系数可以做到 0.1 上下，接近无油润滑。PV 值决定它能承受多高的转速与压力组合。
② 疲劳强度。 关节是百万次量级的往复，看的是 S-N 曲线在 10⁶ 到 10⁷ 次循环下的剩余强度，不是单次冲击值。
③ 吸水后的尺寸漂移。 这一点最容易被忽略，也最要命。
PA66 和 PA46 吸湿后尺寸会涨，涨的量级在精密传动里就足以改变侧隙。齿轮件必须看调湿态数据，不能只看干态。
| 位置 | 树脂方向 | 为什么 |
|---|
| 低速大扭矩关节 | PA46 + 耐磨体系 | 结晶度高、熔点高、流动性好，适合高齿数精密齿 |
| 通用关节齿轮 | PA66 + 耐磨体系 | 成本与性能平衡，供应最稳 |
| 轻载小齿轮（灵巧手） | PA66 / PA12 + 耐磨 | 小模数尺寸精度优先 |
| 纯承力连杆 | PA66-GF50 / PA6T-GF | 走高刚性，不走耐磨路线 |
（均为方向性建议，具体以牌号 TDS 与实测为准）
为什么 PA46 在齿轮里常被提起？ 它结晶度可以做到 70% 左右，熔点接近 295℃，流动性好，能把小模数齿形打满。代价是吸水率不低，必须做调湿和尺寸补偿。
四、结构件和骨架：GF30 是起点，不是终点
承力件的逻辑反过来——这里强度和刚性真的重要。
PA66-GF30 是主力区间，绝大多数中载连杆和支架够用- PA66-GF50 用于高刚性、高抗蠕变位置，比如长悬臂结构，代价是冲击韧性和外观明显变差- 碳纤维增强（PA6-CF / PA66-CF） 用于既想轻又要极刚的地方，公开资料里这类方案的拉伸强度可以做到 150-220MPa 区间，弯曲模量突破 10GPa- 有机身拓扑优化空间时，公开报道有整机做过约 40% 的框架减重，但那是"结构设计 + 材料"一起做的结果，不是换料单独带来的
这里有个必须说清的边界：塑料件能替金属的前提，往往是结构先重做。
拿一个原本按铝合金设计的件直接换尼龙，十有八九会因为刚性不足而变形。以塑代钢，先改结构，再谈材料。 只换料不改设计，那就不是轻量化，是拆东墙补西墙。
碳纤件要多看一句
碳纤维增强的强度漂亮，但三条代价经常被忽略：
成本高，且碳纤长度在注塑后保留率低，性能比样条数据要打折- 碳纤件是导电的，靠近电气件要考虑绝缘隔离- 各向异性明显，流动方向和垂直方向的收缩差得更多，翘曲要提前算
五、外壳、走线和缓冲：这里不需要刚性
第三类件最容易做，也最容易做得不体面。
外壳、线束护套、足底缓冲、仿生皮肤基底——这些件要的不是刚性，是韧、耐弯折、低温不脆、可喷涂。
外壳：增韧 PA6 或 PA/ABS，抗冲击、低气味、表面可处理，做外观件比 PA66 好看- 线束护套：增韧尼龙或长碳链尼龙，反复弯折不裂- 足底缓冲：这里更接近弹性体的活，PA 弹性体（TPAE）或 TPU 体系更合适- 皮肤基底：公开资料显示这一块目前硅胶、TPE、改性 PA66 多条路线并行，还没收敛到唯一方案。改性好、成本可控是尼龙的相对优势
一句判断：越靠近"人"的部位，越不该用高刚性材料。刚性和亲和感是反的。
照抄整机厂料单，是新人最常见的坑
机器人行业里流传最广的选料捷径，是照着头部整机厂的料单抄牌号。我们接触的客户里，至少三成起步就是这么干的。这个捷径有两个暗坑。第一个：整机厂的料单是跟着它的供应链谈的，那个牌号对你来说可能起订量不够、没有本地库存，硬凑的结果是拿“性能接近”的替代牌号顶上，而替代牌号的验证又没做全，等于两头都松。
第二个：整机厂的料单是按它的工况定的，你的结构如果改过壁厚、改过散热，工况就变了，牌号不变的照抄会把隐患原样复制过来。正确的打开方式是抄逻辑不抄牌号：看它为什么在那个部位用那个档位，耐磨的还是刚性的、阻燃的还是不阻燃的，把逻辑拿回来，按自己的工况重新对一遍，再定牌号。料单是别人的答案，逻辑才是可迁移的方法。
六、机器人用尼龙的五个坑
坑 1：照抄整机厂的料单。 大厂公开的件是配套它的结构、它的工艺、它的供应商体系做的。抄牌号不改结构，等于把别人的答案写在自己的卷子上。
坑 2：只看干态强度。 关节件的尺寸问题，八成出在吸湿后的尺寸变化，不是强度不够。
坑 3：耐磨件用高玻纤顶。 玻纤会加剧对偶件磨损。齿轮的"耐磨"是材料与对偶的匹配问题，不是单纯堆硬度。
坑 4：忽略调湿和退火。 精密件打出来不是最终尺寸，调湿和退火之后的尺寸才是。很多"精度不够"其实是后处理没做。
坑 5：用普通注塑件当承力件。 熔接线位置、玻纤取向、浇口设计都会影响实际强度。同样的料，浇口换一个位置，件的破坏位置就变了。
七、边界：这些件现在还不适合上尼龙
| 部位 | 结论 | 原因 |
|---|
| 谐波减速器柔轮本体 | 暂时不适合 | 超高周疲劳 + 微米级形变控制，超出热塑性尼龙稳定区间 |
| 高精度行星滚柱丝杠 | 不适合 | 接触应力极高，需金属 |
| 主承力连接螺栓位 | 不适合 | 预紧力长期保持要求超出塑料蠕变能力 |
| 高温电机附近（>150℃ 长期） | 需谨慎 | 换 PA46、PA6T 体系仍未必要评估 |
| 长期高负载 + 长悬臂 | 需谨慎 | 蠕变累积，必须做长期验证 |
看这张表的正确姿势：不是"哪些不能用"，而是"哪些地方现在还不该用"。
能上的件先上，不能上的件守住。 一条链上盲目替换，会把整机的可靠性拖下来。
行业里的一条实感
机器人件询盘里，我们遇到最多的一种问法，是客户直接把整机厂公开的料单拍过来，问"你们有没有这个牌号"。
这个问题本身，就把方向问偏了。
因为整机厂选那个牌号，是基于它自己的结构设计、自己的注塑设备、自己的后处理工艺。 你把牌号抄过来，结构没改、工艺没调、调湿没做，出来大概率不是"性能差一点"，是"根本装不上"。
我们现在接这类询盘，会先反问三句：这个件现在是金属还是塑料？装在哪个位置、承受什么运动？你的公差要求是多少？
三句问完，很多客户自己就发现问题了——他连"这个件是磨还是扛"都没分清，就先去找牌号了。
先问件，再问料。这句话在机器人赛道比在任何行业都更值钱，因为这里的件，一半是新设计的。
读者追问两则
追问一：机器人迭代快，材料验证等不起怎么办？ 这是行业现状带来的真问题，样机三个月一改版，材料验证一做就是季度级。务实的做法是分层验证：把部件分成安全件、功能件、结构件三档，安全件走全流程一步不让，功能件和结构件用历史数据外推加关键项复测，验证周期压掉一半以上。
再建立一个经验库，同工况的历史验证结论直接调用，新项目只验证增量变量。快是设计出来的，不是省出来的。
追问二：耐磨和刚性打架时听谁的？ 给一个实用的判据：看摩擦副对磨件是什么。对磨件是金属轴的，耐磨优先，刚性过剩的料硬度高反而磨轴；对磨件是塑料对塑料的，刚性优先，两件都软就会粘连爬行。
实在两难的，上低摩擦改性加玻纤的平衡牌号，再辅以间隙设计补偿。判断顺序对了，多数“打架”其实是假两难。
BOM 过一遍的五栏表
给机器人结构团队一张五栏表，把手上 BOM 逐行过。第一栏记力学状态：静态受力、动态传动、还是包覆缓冲，一栏一个词。第二栏记工况：有没有热源贴身、有没有摩擦副、有没有介质接触。
第三栏记现行材料：金属的就标金属，塑料的标牌号。第四栏记塑化优先级：外壳和包覆类给高，传动类给中，核心运动副给低，优先级是资源分配的依据，不是所有件一起动。第五栏记验证等级：安全件全验证、功能件关键项、结构件外推加复测。
五行填完，你会得到一张塑化路线图：先做哪些、缓做哪些、不碰哪些，一目了然。我们给几家机器人客户做过同样的梳理，BOM 里两三成的件当年就能动，多数团队之前不是不知道尼龙，是没有这张表分不清从哪下手。
结语
机器人用尼龙的机会是真的，但有两条边界要同时记住：
一条是位置的边界——高频运动件看耐磨，承力件看刚性，外覆件看韧性。三类件三种逻辑，混着选必错。
一条是设计的边界——以塑代钢不是换料，是结构、材料、工艺一起改。只换料，换出来的是问题，不是轻量化。
Whether robots should use plastic parts was still a controversial issue two years ago. Looking at it this year, the answer has changed — it's no longer 'whether they can be used,' but 'which part should use them first, and which type to use'.
There is an often-cited rule of thumb in the industry: for every 1 kilogram reduction in the weight of the machine, the end can carry an additional load of nearly 0.8 kilograms.
Metal parts have high density, are expensive to process, and require continuous lubrication. Modified nylon has only about half the density of aluminum alloy, and after being reinforced with fibers, its specific strength can match some metals, while also being self-lubricating.
But not every part of a robot can be replaced. This article breaks the robot into five parts using nylon and explains each one individually.
Let's start with a scene.
At last month's industry exhibition, I stood in front of a complete machine manufacturer's booth for half an hour. The prototype on display moved back and forth, and the spectators were watching the nimble fingers, while I was looking at the white casing on its forearm—a tiny gate mark on the edge. After the event, I found their structural engineer to chat. He spoke very frankly: this prototype already has over a hundred plastic parts, but the gearbox and joint bearings are still metal; no one dares to touch them first.
This conversation condenses the current reality of robots using nylon: the outer shell components have largely been plasticized, the transmission parts are half engaged, and the core motion pairs are still on standby. The three lines of defense are advancing at completely different speeds. And the robot inquiries we received this year are also changing—the question last year was whether parts could be made of plastic, and this year it's about which type of nylon to use for a particular component, and whether wear resistance or rigidity should be prioritized.
This article breaks down the five parts one by one. First, it establishes a classification framework for three mechanical states, which is the key to understanding robot selection; then it provides a material map by parts: why joints and gears prioritize wear resistance over strength, why GF30 is the starting point rather than the endpoint for frame components, and why the routing of the outer shell actually does not require rigidity.
Let's talk about five more pitfalls and one boundary—parts that are still not suitable for nylon. Readers engaged in robot structural design are advised to read this while comparing it with the BOM they have on hand, going through each part one by one.
1. First, distinguish: the parts on a robot can be divided into three mechanical states
Even though it is 'robots using nylon,' the logic for selecting materials could be completely opposite. This is because the parts are divided into three states:
Type One · High-frequency reciprocating parts. Joint gears, bearing cages, lead screw nuts, and the small gears of dexterous hands. Their core issue is not 'whether they can withstand it,' but 'whether they can resist wear.'
The second type · Long-term load-bearing structural components. Thigh link, lower leg framework, fuselage load-bearing frame. The core contradiction is rigidity and fatigue, and they must also repeatedly withstand impact.
The third type · Outer covering, wiring, cushioning components. Shell, wire harness sheath, foot cushioning, bionic skin substrate. The core contradictions are toughness, weather resistance, and molding.
These three types of parts are fundamentally made from different materials. Using the approach for load-bearing parts to select gears will result in a component that is 'very stiff but wears out quickly.'
First determine the mechanical state of the part, then talk about the resin. This is the first sentence in robot material selection.
2. Material Map of Five Parts
Disassemble the robot from top to bottom, and the positions where the nylon can be replaced are roughly these five places:
| Body part | Common Substrate Directions | Modified system | Key indicators |
|---|
| Joint Gear / Cage | PA66, PA46 | Wear-resistant and self-lubricating | Coefficient of friction, PV value, fatigue life |
| Load-bearing connecting rod / skeleton frame | PA66, PA6, PA6T | Glass fiber / carbon fiber reinforced | Bending modulus, specific strength, creep resistance |
| Reducer outer casing | PA66, PA46, PA6T | GF30 Thermo-oxidative stability | Dimensionally stable, resistant to 120°C |
| Dexterous hand components | PA66, PA12 | Reinforced Toughened Composite | Small size precision, toughness |
| Wiring Harness Sleeve / Foot Cushion | Toughened PA6, PA12 | Toughening, weather resistance | Not brittle at low temperatures, resistant to bending |
| Shell / Biomimetic Skin Substrate | Toughened PA6, PA/ABS | Toughened, easy to color | Impact-resistant, paintable, low odor |
The most notable thing in this table is the difference between the first and second rows.
It's the same PA66, but for making gears, you need to go towards a wear-resistant system, while for making connecting rods, you need to go towards high glass fiber content. The things added in the two directions are different and can even conflict with each other — wear-resistant modifications often sacrifice a bit of rigidity, while high glass fiber can exacerbate the wear on gears.
One-sentence conclusion: For robots using nylon, first separate by 'grind' or 'toughness', then talk about the grade.
3. Joints and gears: wear resistance should be prioritized over strength
This is the place where robots are most easily chosen incorrectly.
Many people ask 'what is the tensile strength' as the first step. That's the wrong question. The failure of joint gears is rarely 'due to breaking in tension'; the vast majority is wear failure and fatigue pitting — the material doesn't break, but the precision is lost first.
Look at three numbers, not just one number:
① Coefficient of friction. Wear-resistant nylon with added PTFE or molybdenum disulfide can achieve a coefficient of friction around 0.1, close to oil-free lubrication. The PV value determines the combination of rotational speed and pressure it can withstand.
② Fatigue strength. Joints undergo millions of cycles of reciprocating motion, and what is considered is the residual strength on the S-N curve at 10⁶ to 10⁷ cycles, not the single impact value.
③ Dimensional changes after absorbing water. This point is the easiest to overlook and also the most critical.
PA66 and PA46 will increase in size after absorbing moisture, and the magnitude of this swelling is enough to change the backlash in precision transmissions. Gear components must be checked based on data in a moisture-conditioned state, not just in the dry state.
| Position | Resin direction | Why |
|---|
| Low-speed high-torque joint | PA46 Wear-Resistant System | High crystallinity, high melting point, good flowability, suitable for high-tooth-count precision gears |
| Universal joint gear | PA66 Wear-Resistant System | Balanced cost and performance, most stable supply |
| Light-load small gear (Dexterous Hand) | PA66 / PA12 Wear-resistant | Priority is given to the dimensional accuracy of small modules |
| Pure load-bearing connecting rod | PA66-GF50 / PA6T-GF | Go for high rigidity, not wear resistance |
(All are directional recommendations; specific details should be based on the grade TDS and actual measurements)
Why is PA46 often mentioned in gears? Its crystallinity can reach about 70%, the melting point is close to 295°C, it has good fluidity, and can completely fill small module teeth. The cost is a high water absorption rate, so it is necessary to perform moisture conditioning and dimensional compensation.
4. Structural Components and Framework: GF30 is the starting point, not the end point
The logic of the load-bearing parts is reversed — here, strength and rigidity really matter.
PA66-GF30 is the main range, sufficient for the vast majority of medium-load connecting rods and brackets. PA66-GF50 is used in locations requiring high rigidity and high creep resistance, such as long cantilever structures, at the cost of significantly reduced impact toughness and appearance. Carbon fiber reinforcement (PA6-CF / PA66-CF) is used in places where both lightness and extreme rigidity are desired; according to public data, the tensile strength of such solutions can reach the 150-220MPa range, and the flexural modulus can exceed 10GPa. When there is room for structural topology optimization, public reports show that the whole machine can achieve about 40% frame weight reduction, but that is the result of "structural design + material" done together, not brought by changing the material alone.
There is a boundary that must be clarified here: the premise for plastic parts to replace metal is often that the structure is first redesigned.
If you take a part originally designed for aluminum alloy and directly replace it with nylon, nine times out of ten it will deform due to insufficient rigidity. When replacing steel with plastic, first modify the structure, then discuss the material. Simply changing the material without altering the design is not lightweighting; it's just robbing Peter to pay Paul.
Carbon fiber parts need one more look
The strength of carbon fiber reinforcement is impressive, but three costs are often overlooked:
High cost, and the carbon fiber length retention rate is low after injection molding, so performance is lower than the filament data - Carbon fiber parts are conductive, so insulation and isolation need to be considered near electrical components - Anisotropy is obvious, the shrinkage in the flow direction and the perpendicular direction differs more, and warpage needs to be calculated in advance
5. Housing, routing, and cushioning: rigidity is not needed here
The third type of work is the easiest to do, but also the easiest to do poorly.
Casing, wire harness sheath, foot cushioning, bionic skin substrate—these parts need not rigidity, but toughness, bend resistance, low-temperature non-brittleness, and paintability.
Housing: Toughened PA6 or PA/ABS, impact-resistant, low odor, surface-treatable, looks better than PA66 for exterior parts.
- Wiring harness sheath: Toughened nylon or long-chain carbon nylon, doesn't crack under repeated bending.
- Sole cushioning: Here it is closer to elastomer properties; PA elastomer (TPAE) or TPU system is more suitable.
- Skin base: Public information shows that currently multiple paths are running in parallel here, including silicone, TPE, and modified PA66, and no single solution has been finalized yet. The relative advantage of nylon is that it can be well-modified and cost-controlled.
A judgment: The closer to the 'human' parts, the less you should use high-rigidity materials. Rigidity and a sense of affinity are inversely related.
Blindly copying the entire machine manufacturer's parts list is the most common pitfall for newcomers.
The most widely circulated shortcut for selecting materials in the robotics industry is to copy the part numbers from the bills of materials of leading complete machine manufacturers. Among the clients we have interacted with, at least 30% start this way. This shortcut has two hidden pitfalls. The first: the bills of materials of complete machine manufacturers are negotiated based on their supply chains, and the part numbers might not meet your minimum order quantities or may not have local stock. As a result, you end up using 'performance-approximate' alternative part numbers, and the verification for these alternatives is often incomplete, which means both ends are loose.
Second: The bill of materials from the complete machine manufacturer is based on its operating conditions. If your structure has undergone changes in wall thickness or heat dissipation, the operating conditions have changed. Copying the same material grades without change will also replicate the hidden risks. The correct approach is to copy the logic, not the material grades: understand why they used a certain grade in that specific part, whether it is wear-resistant or rigid, flame-retardant or not. Take the logic, match it to your own operating conditions, and then decide on the material grades. The bill of materials is someone else's answer; the logic is the transferable method.
6. Five Pitfalls of Using Nylon for Robots
Pitfall 1: Copying the parts list of the whole machine manufacturer. The parts made public by large manufacturers are designed according to their own structures, processes, and supplier systems. Copying the part numbers without changing the structure is equivalent to writing someone else's answers on your own test paper.
Pitfall 2: Only look at the dry strength. The size problem of joint parts is mostly due to dimensional changes after moisture absorption, not insufficient strength.
Pitfall 3: Using high glass fiber content for wear-resistant parts. Glass fiber will accelerate wear on mating parts. The 'wear resistance' of gears is a matter of material compatibility with the mating parts, not simply increasing hardness.
Pitfall 4: Ignoring moisture adjustment and annealing. Precision parts are not at their final dimensions right after machining; the dimensions after moisture adjustment and annealing are the final ones. Many cases of 'insufficient precision' are actually due to the lack of post-processing.
Pitfall 5: Using ordinary injection-molded parts as load-bearing components. The location of weld lines, glass fiber orientation, and gate design all affect the actual strength. With the same material, changing the gate position will change the failure location of the part.
7. Borders: These pieces are not yet suitable for nylon.
| Part | Conclusion | Reason |
|---|
| Harmonic drive flexspline body | Not suitable for now | Ultra-high cycle fatigue Micron-level deformation control, beyond the stable range of thermoplastic nylon |
| High-precision planetary roller screw | Not suitable | The contact stress is extremely high and requires metal. |
| Main load-bearing connection bolt position | Not suitable | The long-term requirement for preload exceeds the creep capacity of the plastic. |
| Near high-temperature motors (>150°C long-term) | Need to be cautious | Switching to PA46 or PA6T systems may still not require evaluation |
| Long-term high load Long cantilever | Need to be cautious | Creep accumulation must undergo long-term verification |
The correct way to look at this chart is not 'what cannot be used,' but 'which places should not be used right now.'
Install the parts that can be installed first, and hold on to the parts that cannot be installed. Blindly replacing parts on a chain will drag down the reliability of the entire machine.
A real feeling in the industry
In robot parts inquiries, the most common question we encounter is when customers directly take a photo of the bill of materials published by the complete machine manufacturer and ask, 'Do you have this grade?'
This question itself has already led the direction astray.
The reason the complete machine manufacturer chooses that grade is based on its own structural design, its own injection molding equipment, and its own post-processing technology. If you copy the grade, without changing the structure, adjusting the process, or performing moisture conditioning, the result will most likely not be 'slightly worse performance,' but 'simply won't fit at all'.
When we receive this type of inquiry now, we first ask three questions in return: Is this part made of metal or plastic? Where is it installed, and what kind of movement does it undergo? What are your tolerance requirements?
After asking three questions, many clients discovered the problem themselves—they hadn’t even figured out whether this piece was for grinding or for carrying, yet they went straight to looking for the grade.
Ask about the parts first, then about the materials. This saying is more valuable in the robotics field than in any other industry because half of the parts here are newly designed.
Two Reader Follow-up Questions
Follow-up Question 1: Robots iterate quickly, so what if material validation can't keep up? This is a real issue brought about by the current state of the industry. Prototypes are revised every three months, while material validation takes a whole quarter. A pragmatic approach is layered validation: divide components into three categories—safety parts, functional parts, and structural parts. Safety parts go through the full process without skipping any steps, while functional and structural parts can use historical data for extrapolation plus retesting of key items, reducing the validation cycle by more than half.
Establish another knowledge base, where historical validation conclusions under the same working conditions can be directly used, and new projects only need to verify incremental variables. Speed is designed, not saved.
Follow-up Question 2: When wear resistance and rigidity conflict, which one should be prioritized? Here's a practical criterion: look at what the friction pair is rubbing against. If the part being worn is a metal shaft, prioritize wear resistance; if the material is excessively hard for rigidity, it will wear down the shaft. If the part being worn is plastic against plastic, prioritize rigidity; if both parts are soft, they will stick and creep.
It's really a dilemma. Use a balanced grade with low-friction modified and glass fiber reinforced, and additionally compensate with gap design. If the judgment order is correct, most "conflicts" are actually false dilemmas.
BOM Review Five-Column Table
Give the robot structure team a five-column table and go through the BOM line by line. The first column records the mechanical state: static load, dynamic transmission, or covering cushioning—one word per column. The second column records the working condition: whether there is a heat source nearby, whether there are friction pairs, whether there is contact with medium.
The third column records current materials: metal is marked as metal, plastic is marked with the grade. The fourth column records plasticization priority: high for shells and coverings, medium for transmission, low for core motion pairs. Priority is for resource allocation, not all parts move at the same time. The fifth column records verification level: all safety parts fully verified, key points for functional parts, external extension plus re-testing for structural parts.
After filling in the five rows, you get a plasticization roadmap: which to do first, which to delay, which not to touch—clear at a glance. We have done the same sorting for several robot customers; 20-30% of the parts in the BOM could be activated that year. Most teams previously didn’t lack knowledge of nylon, but without this table, they couldn’t tell where to start.
Conclusion
The opportunity to use nylon in robots is real, but two boundaries must be remembered simultaneously:
One is the positional boundary—high-frequency moving parts look at wear resistance, load-bearing parts look at rigidity, and covering parts look at toughness. Three types of parts, three kinds of logic; mixing the selection will inevitably be wrong.
The other is the design boundary—replacing steel with plastic is not just changing the material, it requires changing structure, material, and process together. Changing only the material will create problems, not lightweight solutions.
