去年梅雨季，一个做电动工具配件的厂给我发照片。
照片里是一整箱刚打出来的外壳，表面上一道一道的银白色条纹，从浇口往远端散开——这是银纹，注塑里最典型的一种缺陷。
他的判断是："这批料有问题，肯定是掺了东西。"
我到现场没先看料，先问了三句话：
其一：这料干燥了吗？
烘了。
其二：烘了多久、多少度？
三个小时，80 度。
其三：用的什么机器烘的？
车间角落那台老的热风循环烘箱。
问题到这儿基本就有答案了——不是料掺了东西，是水没烘出来。
后面送检的结果印证了：熔融指数、灰分、相对黏数都在正常区间。唯独含水率，远超注塑要求的范围。
这件事给我留下最深的印象不是检测结果，是客户那句判断："肯定是掺了东西。"
他其实想知道的是"这批还能不能用"，但他问出口的是"是不是假料"。这两件事差得很远——前者要的是处置方案，后者要的是一个可以归咎的对象。
为什么验料这件事，总是排到最后
尼龙原料这行，检验手段都是现成的，标准号也是公开的。缺的不是方法，是把方法变成习惯。
原因说穿了很简单：检测的成本是显性的，收益却是隐性的。
一笔几百块的检测费，在采购单上是一行看得见的支出；而它换来的东西叫"这批没出事"——没出事是没有故事的，也没人会为此写一份报告。
所以真正拉开差距的往往不是技术，是有没有把它排进流程。下面八条按从快到慢排，前四条现场几分钟能做完，后四条要送检，但都不复杂。
第 1 条 · 要原厂 COA，并核对批号
这是性价比最高的一条。
怎么查：要原厂质检单（COA），看上面的批号与包装喷码、送货单是否一致。有疑问就向原厂或授权渠道核实。
能查出什么：这一条就能筛掉大部分问题货。 因为它要求对方拿出一个可被第三方验证的凭证——而伪造凭证的成本，远高于如实说明。
这里有个很实际的细节："事后补来的 COA"，价值要打很大的折扣。
原因是 COA 本来就该跟着批次走，被催过之后再补的那一份，你没法确认它补的是不是这一批的数据。所以当对方说"COA 下次补"，不是不能继续谈，但从这一刻起，你要把其他七条全部做完。
第 2 条 · 看包装的三个细节
怎么查：①封口工艺是否规整；②印刷与喷码的清晰度、字体是否一致；③内衬袋是不是原厂那种。
能查出什么：分装、二次包装的痕迹。
这一条只做初筛，不做结论——外观是能被仿的，而且仿外观的成本远低于仿指标。看到可疑不必下判断，看到"完美"也不要就此放心。
整托到货时多看一眼托盘：原厂整托的缠绕膜走向、打包带间距、托底新旧程度，都是一套固定动作留下的痕迹。 倒下来重新装托的，很难还原到那个程度。
第 3 条 · 看颗粒
怎么查：抓一把摊在白纸上，光线要好。全新料颗粒大小均匀、有光泽、切口整齐；回料通常大小不一、发暗、带杂质点或黑点。
能查出什么：回料的掺入迹象，以及造粒质量。
为什么回料的颗粒会大小不一？ 因为它的来源通常是水口、浇口、废件破碎后再过一次挤出——破碎本身就会产生粒径分布，再叠加一次造粒的波动，均匀性天然比连续聚合出来的新料差。
切口的形状也有讲究：拉条切出来的颗粒是规整的圆柱状，水下切出来的是圆片或扁圆状。同一批次里混着两种明显不同的形状，说明它可能不是同一条线出来的。
这一条不用工具，但有个前提很重要：你得见过好的料长什么样。 建议留一小罐确认过的标准样放在验收台上，人的眼睛对"差异"比对"标准"敏感得多。
第 4 条 · 烧一下闻
怎么查：取少量样点火。尼龙的特征是蓝底黄焰、起泡滴落、有烧焦毛发的气味，离火缓慢自熄。
能查出什么：材质是不是尼龙（初步）、有没有混别的东西。
两个提醒：
一是这只做定性初筛，不能替代仪器。 气味和火焰能骗过鼻子和眼睛，但骗不过红外和 DSC。
二是阻燃料会误导你。 加了阻燃剂的配方本来就改变了自熄行为，别因为"它烧得不像尼龙"就直接判定是假料——先看清楚你买的是不是阻燃牌号。
第 5 条 · 跨批次熔指，不是单批
怎么查：测熔融指数（MFR），但要跨批次测。
能查出什么：批次之间的稳定性。
这一条要特别说清楚：单批熔指正常，什么都说明不了。副牌料的核心特征不是"单点值低"，而是"批次之间不一样"。 所以只测一批，等于没测。
只测一批料的供应商，和只测一批料的采购，风险是一样的。
采购上的一条判断：采购最常省掉的一步是跨批次对比熔指。单批熔指正常，其实什么都说明不了。 让供应商出一份当批检测报告很容易，跨三批、五批的数据才见真章——要的是趋势，不是快照。
真正管用的做法是把数据排成一列看走势：一批 +5%、一批 -4%、一批 +6%，这叫"围绕中心跳"；连续三批往同一个方向走，那通常不是波动，是工艺在漂。
第 6 条 · 相对黏数：识别掺混回料的硬指标
怎么查：测相对黏数（乌氏粘度计，GB/T 1632.1 聚酰胺黏数测定）。
能查出什么：聚合度。 这一项直接反映分子链的长短，而再生料的黏数会明显下降。
原因是不可逆的：尼龙每过一次高温加工，分子链都要断一截。 链短了，拖动起来阻力小，黏数就低。所以它是判断"里面有没有掺回料"最硬的一项。
为什么这一项比拉伸强度更值得先看？
因为力学、耐热、耐老化这些性能，本质上都挂在同一条分子链上。 你把一条长链拉断了一半，短时间内测拉伸可能还在合格线上，但它的耐热老化和耐久性已经提前透支了——这一项掉了，别的迟早跟着掉。
第 7 条 · 灰分：反算实际玻纤含量
怎么查：马弗炉灼烧称重（GB/T 9345.1），算出无机残留的比例。
能查出什么：对增强料来说，灰分能直接反算出实际的玻纤含量。
操作很简单：坩埚称重，称 5 克左右样品，600℃ 灼烧 2 小时，冷却称重，残渣比例约等于无机物含量。PA66-GF30 烧完应该剩 30% 左右，剩 22% 就是差不多少加 8 个点。
这条专门用来对付一种情况：标称 GF30，实测只有十几。
玻纤含量不足，不只是刚性不够——收缩率也会跟着变。 粗略地说，玻纤含量每低 5 个点，收缩率大致往回走 0.1-0.2 个点；一个 120mm 的件，0.2% 就是 0.24mm，足以把你的装配尺寸从"刚好"推到"超差"。
而模具是按原来的收缩率开的补偿，它没有能力知道你换了料。
第 8 条 · 含水率：最容易藏问题的一项
怎么查：干燥前测含水率（卡尔·费休法，GB/T 12006.2），并且要测烘完之后、直接从料斗里取的那批料。
能查出什么：注塑风险的源头。
尼龙天生吸水。这里有一笔账值得算清楚：
PA6 在 23℃、50% 相对湿度的环境里放久了，平衡吸水率大致能到 2.5-3%。而注塑前的通常要求是控制在 0.2% 以下。
换成看得见的单位：一吨料要从 3% 烘到 0.2%，等于要从里面抽走差不多 28 公斤水。
28 公斤，差不多是一桶桶装水的量——它被分散在一千公斤颗粒的内部，不是表面那层潮气。
这就是为什么尼龙用普通热风干燥机，约等于没干燥。
热风烘箱能带走的是表面附着水；而尼龙吸进去的水是在分子链之间的，只有低露点的除湿干燥才能把它拉出来。行业里的做法通常是除湿干燥机走 -20℃ 到 -40℃ 露点的空气，配上足够的保温时间（常见 4-6 小时，具体看料号和初始含水率）。
还有一件事特别容易漏：烘好的料如果不密封，放在潮湿车间里会重新吸回去。 南方梅雨季尤甚——上面那个厂之所以会出现整批银纹，一半是干燥方式不对，另一半是烘完之后在车间敞着放了一夜才开始用。
含水率偏高的后果从来不是"外观差一点"，是注塑时水解降解：银纹、气泡、力学性能下降。而这件事最讨厌的地方在于——它在原料袋上根本看不出来，只在成品件上显形，而且显形得很晚。
这里还有一个不用仪器的判断办法：看注塑机射嘴挤出的那一段熔体。
烘透了的料，挤出来表面光滑、连续、没有气泡；没烘透的会有细密气泡、断面发白，靠近了还能听到轻微的噼啪声。
这个办法粗，但有个优点很关键：它查的是"你即将用那一刻"的状态，而不是烘箱上的设定值。 很多问题恰恰出在这中间——设定是全对的，但从烘箱转移到机台料斗这一段路上，料又吸回去了。
先把标准定下来：合同怎么写才有用
上面八条都是技术动作，但它们有个共同前提——你得先把标准定下来。
所以最终防线只有两步：
其一，把指标写进合同。 玻纤含量、阻燃等级、拉伸强度、含水率、相对黏数范围，全部落到书面。合同里写不下的指标，等于不存在。
这里有一个经常被人漏掉的细节：只写指标不写方法，等于没写。
比如"玻纤含量 30%"，用什么标准测？灰分还是别的？"拉伸强度 80MPa"，是干态还是调湿后？样条厚度多少？
建议每一条指标后面跟一个标准号（像 GB/T 9345.1、GB/T 1632.1 这类），并且约定清楚：争议时以哪一方的第三方检测为准、异议期多长。 这三件事写清楚，很多质量纠纷在发生之前就消失了。
其二，到货抽检。 不是首批检一次，是按频率持续抽检。
问"为什么便宜"得到的是解释，问"便宜在哪一项"得到的是判断。
抽检频率怎么定：别把它交给心情
"抽检"这两个字好说，落到实处最难的是频率。这里给一套可以照抄的排法：
| 情况 | 建议做什么 | 大概频次 |
|---|
| 新供应商、新料号首批 | 八条全做 | 每批 |
| 稳定供应中 | 全套一次 + 快速项（颗粒·灼烧·含水率） | 全套约每 5 批 1 次，快速项每批 |
| 换季（梅雨季入夏、秋冬转干） | 加一次含水率与熔指 | 季节切换各 1 次 |
| 换批号、换产地 | 按新供应商对待 | 每批 |
| 客户端出过问题之后 | 全套 + 加严自检 | 连续 3 批 |
三条原则：
一是"换批号等于换供应商"， 即便同一家、同一个牌号，批号一变就要重来一轮，这是最基本的一条。
二是含水率要按季节排， 它在梅雨季和干燥季完全是两种难度，用同一个频次是不合理的。
三是留样跟着抽检走， 每抽一次就把那一批封一袋标好的样——检测决定你能不能查出来，留样决定你有没有得查。
结语
尼龙原料采购这件事，本质上是用检测成本换不确定性。
一次相对黏数加灰分加含水率，几百块到一千出头。而一批料出问题，代价是几吨料、几次调机、一批返工件，以及一个客户的信任。
哪笔账更划算，做采购的都算得清。真正难的不是算账，是在"这批看着没问题"的时候，依然按流程走完那十分钟。
回到开篇那个客户。后来他换了一台除湿干燥机，把干燥后的料加了个密闭周转箱，同一批料重新上线——银纹一件都没再出现。
料没换，供应商没换，价格没换。换的是流程里的两个动作。
如果你手上有批料要判断能不能用，把三样发过来：牌号与标称指标、实测的黏数或熔指、现在用在什么件上——我们给一份判断。
有些生意我们不做。
不问用途就报价的，不做。
把副牌料说成正牌卖的，不做。
Last year during the rainy season, a factory that makes electric tool accessories sent me a photo.
The photo shows a whole box of freshly molded shells, with silver-white streaks on the surface that spread from the gate toward the far end—these are silver streaks, one of the most typical defects in injection molding.
His judgment was: "There is something wrong with this batch of material; it definitely has something mixed in."
When I arrived at the site, I didn't check the materials first; I asked three questions first:
First: Is this material dry?
Baked.
Secondly: How long was it baked, and at what temperature?
Three hours, 80 degrees.
Third: What machine was used to roast it?
The old hot air circulating oven in the corner of the workshop.
The problem basically has an answer here—it’s not that something was mixed into the batter, it’s that the water wasn’t baked out.
The results of the subsequent tests confirmed: the melt index, ash content, and relative viscosity were all within the normal range. Only the moisture content was far above the range required for injection molding.
What impressed me the most about this matter was not the test results, but the client's judgment: 'It must have been adulterated.'
What he actually wanted to know was 'Can this batch still be used?' but what he asked out loud was 'Is it fake material?' These two things are very different—the former is asking for a solution, while the latter is looking for someone to blame.
Why is the inspection of materials always scheduled last?
In the nylon raw materials industry, the testing methods are all readily available, and the standard numbers are also public. What is lacking is not the methods, but turning the methods into habits.
The reason, simply put, is very straightforward: the cost of testing is explicit, while the benefits are implicit.
A few hundred yuan testing fee appears as a visible line item on the purchase order; what it brings in return is called 'this batch is problem-free' — being problem-free means there is no story, and no one will write a report about it.
So what really creates the gap is often not technology, but whether it has been integrated into the process. The following eight items are ranked from fast to slow; the first four can be completed on-site in a few minutes, while the latter four need to be sent for testing, but none are complicated.
Article 1 · Must have the original factory COA and verify the batch number
This is the most cost-effective one.
How to check: You need the original factory quality inspection report (COA) and should check whether the batch number on it matches the packaging print code and the delivery note. If there are any doubts, verify with the original factory or authorized channels.
What can be checked: This point alone can filter out most problematic goods. Because it requires the other party to provide a certificate that can be verified by a third party—while the cost of forging a certificate is much higher than simply stating the truth.
Here is a very practical detail: a 'COA obtained afterwards' should be heavily discounted in value.
The reason is that the COA is supposed to follow the batch. The part that was added after being urged—you can't confirm whether the data it supplements belongs to this batch. So when the other party says 'COA will be supplemented next time,' it's not that you can't continue negotiating, but from this moment on, you need to complete all the other seven items.
Article 2 · Three details to look at on the packaging
How to check: ① Whether the sealing process is neat; ② The clarity of printing and coding, and whether the fonts are consistent; ③ Whether the inner liner bag is the original factory one.
What can be detected: signs of repackaging or secondary packaging.
This point is only for preliminary screening, not for drawing conclusions—the appearance can be imitated, and the cost of imitating the appearance is much lower than imitating the indicators. Seeing something suspicious does not require judgment, and seeing something 'perfect' should not make you feel at ease.
Take an extra look at the pallet when the full pallet arrives: the direction of the stretch film, the spacing of the straps, and the condition of the pallet base are all traces left by a consistent set of actions. If the pallet has been unloaded and then repacked, it's hard to restore it to that level.
Article 3 · Look at the granules
How to check: Grab a handful and spread it on a white sheet of paper, with good lighting. New material particles are uniform in size, shiny, and have neat cut edges; recycled material is usually uneven in size, dull, and may have impurities or black spots.
What can be detected: signs of recycled material mixing, and granulation quality.
Why are the particles of recycled material uneven in size? Because they usually come from sprues, gates, and crushed scraps that are extruded again—crushing itself creates a particle size distribution, and adding another granulation process introduces further variation, so the uniformity is naturally worse than newly continuously polymerized material.
The shape of the incision is also important: particles cut with a strip are regular cylindrical shapes, while those cut underwater are round or oblong slices. If two distinctly different shapes are mixed in the same batch, it indicates that they may not have come from the same line.
This method doesn't require any tools, but there's an important prerequisite: you must have seen what good material looks like. It is recommended to keep a small jar of confirmed standard sample on the inspection table, because the human eye is much more sensitive to 'differences' than to a 'standard'.
Article 4 · Burn a little and smell
How to check: Take a small sample and ignite it. The characteristics of nylon are a blue base with a yellow flame, bubbling and dripping, an odor of burnt hair, and it slowly self-extinguishes when removed from the flame.
What can be checked: whether the material is nylon (preliminary), and whether anything else is mixed in.
Two reminders:
First, this is only for qualitative preliminary screening and cannot replace instruments. Smell and flame can deceive the nose and eyes, but they cannot fool infrared and DSC.
Secondly, flame retardants can mislead you. Formulations with added flame retardants inherently change self-extinguishing behavior, so don't immediately judge it as fake material just because 'it doesn't burn like nylon'—first, make sure whether what you bought is a flame-retardant grade.
Article 5 · Melt index across batches, not a single batch
How to check: measure the Melt Flow Rate (MFR), but it should be measured across batches.
What can be checked: the stability between batches.
This point needs to be made very clear: a normal melt flow index in a single batch tells nothing. The core characteristic of secondary brand material is not a 'low single point value,' but 'inconsistency between batches.' So testing only one batch is equivalent to not testing at all.
Testing only one batch of material from the supplier and testing only one batch of material from the procurement carries the same risk.
A judgment in purchasing: the step most often skipped in procurement is the comparison of melt index across batches. A normal melt index in a single batch actually tells you nothing. It's easy to get the supplier to provide a test report for the current batch, but the true insight comes from data across three or five batches—you need the trend, not a snapshot.
The truly effective approach is to arrange the data in a row to observe the trend: one batch at 5%, one batch at -4%, one batch at 6%; this is called 'jumping around the center.' If three consecutive batches move in the same direction, that usually isn't fluctuation, it's a drift in the process.
Article 6 · Relative Viscosity: A Hard Indicator for Identifying Mixed Recycled Material
How to check: Measure relative viscosity (Ubbelohde viscometer, GB/T 1632.1 Determination of polyamide viscosity).
What can be detected: degree of polymerization. This item directly reflects the length of molecular chains, and the viscosity of recycled material will decrease significantly.
The reason is irreversible: every time nylon undergoes high-temperature processing, its molecular chains break. When the chains are shorter, there is less resistance when dragged, so the viscosity is lower. Therefore, it is the strictest criterion for determining whether recycled material has been mixed in.
Why is this item more worth looking at first than tensile strength?
Because properties like mechanical strength, heat resistance, and aging resistance are essentially all dependent on the same molecular chain. If you break a long chain in half, the tensile test may still be within acceptable limits in the short term, but its heat aging and durability have already been prematurely compromised—once this property drops, the others will inevitably follow.
Article 7 · Ash Content: Back-calculate Actual Glass Fiber Content
How to check: Muffle furnace calcination weighing (GB/T 9345.1) to calculate the proportion of inorganic residues.
What can be found: For reinforcements, ash content can be directly calculated by reverse calculating the actual glass fiber content.
The operation is simple: weigh in a crucible, weigh about 5 grams of sample, burn at 600°C for 2 hours, then cool and weigh, with residue ratio roughly equal to the inorganic content. PA66-GF30 should have about 30% left after burning; 22% is about 8 points added.
This is specifically for a situation: nominal GF30, but actual measurements show only a dozen or so.
Insufficient glass fiber content means not only insufficient rigidity but also changes shrinkage rate. Roughly speaking, for every 5 points lower in glass fiber content, shrinkage rate roughly drops by 0.1-0.2 points; For a 120mm piece, 0.2% is 0.24mm, enough to push your assembly size from "just right" to "over-deviable."
The mold compensates according to the original shrinkage rate; it cannot detect when you changed the material.
Article 8 · Moisture Content: The Most Problematic Item
How to Check: Measure moisture content before drying (Karl Fishew method, GB/T 12006.2), and measure the batch of material taken directly from the hopper after drying.
What can you find: The source of injection molding risk.
Nylon is naturally absorbent. Here's a calculation worth calculating:
PA6 After being stored for a long time at 23°C and 50% relative humidity, the equilibrium water absorption rate can reach about 2.5-3%. The usual requirement before injection molding is to keep it below 0.2%.
In a visible unit: one ton of material needs to be dried from 3% to 0.2%, which means about 28 kilograms of water must be extracted from inside.
28 kilograms is about the amount of water in a bucket—it's dispersed inside a thousand kilograms of pellets, not the surface moisture layer.
That's why nylon is basically dried at all with a regular hot air dryer.
Hot air ovens can remove surface water; while the water drawn in by nylon is between molecular chains, and only low-dew dehumidification drying can pull it out. In the industry, the usual approach is to use dehumidifying dryers with air at a dew point of -20°C to -40°C, combined with sufficient holding time (commonly 4-6 hours, depending on the part number and initial moisture content).
Another thing that easily leaks: if the dried material is not sealed, it will be sucked back in in a humid workshop. This is especially true during the southern rainy season—the reason the factory above developed a whole batch of silver markings was partly due to the wrong drying method, and the other half because it was left open overnight in the workshop after drying before starting to use. The consequence of
's high moisture content is never "poor appearance," but hydrolytic degradation during injection molding: silver grain, bubbles, and reduced mechanical properties. The most annoying thing about this is that it is not visible on the raw material bag, only visible on finished parts, and it shows up very late.
Here's another method that doesn't require instruments: look at the melt section extruded from the injection molding machine's nozzle.
Well-baked material produces smooth, continuous surfaces without bubbles; Unbaked material has fine bubbles and a pale cross-section, and you can even hear a faint crackling sound when you get close.
This method is rough, but has a key advantage: it checks the state of "the moment you are about to use it," not the set value on the oven. Many problems lie right here—the settings are all correct, but on the journey from the oven to the machine hopper, the material is drawn back in.
Set the standards first: how to write the contract for it to be useful
The above eight are all technical actions, but they all share a common premise—you have to set the standards first.
So the final line of defense is only two:
First, include the indicators in the contract. Glass fiber content, flame retardant rating, tensile strength, moisture content, and relative viscosity range—all in writing. If the contract can't include the indicators, it's as if they don't exist.
There's a detail that's often overlooked: only stating the indicators without the method is basically not stating it.
For example, "Fiberglass content 30%"—what standard is used to measure it? Is the ash content something else? "Tensile strength 80MPa"—is it dry or after humidity adjustment? What is the thickness of the spline?
suggests adding a standard number after each indicator (such as GB/T 9345.1, GB/T 1632.1, etc.), and clearly specify: in case of disputes, the third-party inspection by which party prevails, and how long the objection period is. If these three things are clearly stated, many quality disputes disappear before they even arise.
Second, spot check upon arrival. It's not the first batch inspection, but continuous sampling at frequency.
Asking "Why cheap" gets an explanation; asking "Which item is cheap" gets a judgment.
How to set the frequency of random inspections: Don't leave it to your mood
The word "random inspection" is easy to say, but the hardest to put into practice is frequency. Here is a copiable layout:
| Situation | What to Recommend | Approximate Frequency |
|---|
| New Supplier, New Material Number First Batch | All Eight Items Made | Each Batch |
| Stable Supply | Full Set Once + Quick Items (Granules · Burning · Moisture content) | complete set about once every 5 batches, fast item per batch |
| seasonal change (Meiyu season entering summer, autumn and winter drying out) | add moisture content and melting index | seasonal switching once each |
| batch number and origin change | treat as new supplier | per batch |
| After client issues | full set + strict self-inspection | three batches in a row |
Three principles:
First, "changing batch numbers equals changing suppliers"—even if it's the same company or grade, if the batch number changes, you have to redo it. This is the most basic rule.
Second, moisture content must be ranked by season. The rainy season and dry season are completely different in difficulty, so using the same frequency is unreasonable.
Third, keep samples and follow sampling inspections. Each time, the batch is sealed and labeled—testing determines whether you can detect it, and retaining samples determines whether you can check it.
Conclusion
Nylon raw material procurement essentially trades testing costs for uncertainty.
One batch of relative viscosity plus ash and moisture content costs several hundred yuan to just over a thousand. And when a batch has problems, the cost is several tons of material, several machine adjustments, a batch of reworked parts, and a customer's trust.
Which transaction is more cost-effective? Procurement professionals can calculate clearly. The real difficulty isn't the accounting, but still following the process for those ten minutes when "this batch looks fine."
Back to the customer at the beginning. Later, he switched to a dehumidifier dryer, added a sealed transfer box to the dried material, and relaunched the same batch — not a single silver pattern appeared.
Material didn't change, supplier didn't change, price didn't change. What changed were two steps in the process.
If you have batch materials and want to judge if they work, send me three items: grade and nominal label, measured viscosity or melting index, and what item is currently used on—we'll give you a judgment.
Some businesses we don't do here.
Quoting without asking about purpose, don't do it.
Claim that the secondary card material is sold as genuine brand, don't do it
