上个月，一家做光模块结构件的客户发来一句很短的话。
只有九个字加一个问号："你们有没有低介电的尼龙？"
我回过去先问了三句：这个件的频率段落在哪里、插入损耗的预算给到多少？外壳做不做金属化？长期湿热条件是多少？
他给出的答案是：28 GHz 附近的一段、外壳要做导电化处理、长期 85℃ 与 85% 相对湿度。
三句答完，这件事的轮廓就变了——他真正要解决的不是"介电低不低"，是"在 85℃ 85% 湿度里放五年，尺寸、镀层和损耗还能不能同时守住"。
这篇把高频连接器材料这件事讲开：先看这个件被什么约束，再看几条基材各自的位置，然后是判据表，最后说清哪一段频率之上，塑料不该接这个活。
一、低介电是个相对词，先说清"低"是跟谁比
材料圈里说"低介电"，是个相对说法，得先找参照。
普通尼龙不是一个低损耗的体系。
PA6 和 PA66 的分子链上有密集的酰胺基团，这是极性基团。极性基团在交变电场里会跟着翻转，翻不动就发热——这就是介质损耗的物理来源。
所以 PA66 在 1 GHz 下的介电常数大致在 3.5 到 4.0 之间，介质损耗在 0.02 到 0.04 量级。
对比一下：LCP 和改性 PPE 的介质损耗在 0.002 到 0.007 之间。
差了一个数量级。这个差距不是靠配方抹平的，是分子结构决定的。
再说第二件事：吸水。
水的介电常数是 80 量级，比任何工程塑料都高一大截。
尼龙吸进去的每一份水，都在替材料抬高介电常数、抬高介质损耗。PA66 的平衡吸水率是 8% 到 9%，PA9T 明显更低——这就是"低吸水基材"在射频件上被高看一眼的原因，不只是尺寸，还有电性能。
第三件事：玻纤。
E 玻璃的介电常数在 6.6 量级，比树脂高。
也就是说，你为了刚性加进去的每一份玻纤，都在把介电常数往上抬。刚性要、介电要低，这两个方向在玻纤身上是对着走的。
所以"低介电尼龙"这四个字的真实含义是：在一个本来不低的体系里，尽量把这三样东西压住——少极性基团、少吸水、少高介电填料。
一句话：所谓低介电尼龙，做的不是把尼龙变成本征低损耗材料，是把损耗的增幅控在预算以内。
二、工况六维：高频件被什么夹住
频率与损耗预算。这是最先要问清的。
插损预算通常按整条通道分配，连接器或外壳这一段能分到的份额有限。频率越高，同样的介质损耗折算出来的衰减越大。
频率从 10 GHz 走到 28 GHz，同一份介质损耗，衰减量级会翻上两倍以上。所以同一个料在 100G 的件上够用，到 400G 就不够了。
温度。高频件分两段：过炉那几十秒的峰值 240 到 260℃，以及长期工作温度。
光模块外壳的长期温度常按 70 到 85℃ 算，堆叠密集的机柜里会更高。要盯的还是长期值。
湿度。这是高频件最容易被低估的一维。
85℃ 与 85% 相对湿度是行业里常见的加速试验条件。在这个条件下，尼龙的吸湿会走向接近饱和。
吸水之后，尺寸涨、介电常数抬、介质损耗抬，三件事同时发生。
载荷。低频件谈插拔力，高频件谈的是插拔时的定位精度。
光模块的插入公差常见在 0.05 毫米量级，件一涨，插拔力就从"刚好"变成"卡"。
外观与工艺。外壳常常要做金属化或导电喷涂，镀层的附着力是硬项。
表面一旦有迁移层，涂层就抓不住——这一条跟胶芯那篇说的表面析出是同一件事。
合规。阻燃 V-0、无卤、RoHS，以及部分客户对挥发与析出的额外要求。
这六维里，频率和湿度是高频件独有的，其余四维和普通连接器共用一个逻辑。
三、几条路线，各自站得住的地方
把能做这个件的基材摆开，大致是五条路线。
| 路线 | 介电常数（1 GHz 量级） | 介质损耗量级 | 在这个件上的位置 |
|---|
| PA6T / PA9T 增强 | 约 3.4–3.8 | 约 0.01–0.02 | 耐回流焊与尺寸稳定兼顾；高玻纤时介电会上去 |
| PPA（半芳香族共聚） | 约 3.3–3.8 | 约 0.008–0.015 | 耐温与低吸水平衡，加工窗口宽一些 |
| 改性 PPE / PPO | 约 2.5–2.9 | 约 0.002–0.007 | 射频表现好；耐温与耐溶剂要单独评 |
| LCP | 约 3.0–3.3 | 约 0.002–0.005 | 极薄壁与高频段优势明显；韧性与流动方向性强 |
| PPS / PEEK | 约 3.0–3.2 | 约 0.002–0.005 | 耐温与化学稳定好；价格与成型工艺另有取舍 |
不做"谁更好"的结论，只谈差在哪。
PA6T 与 PA9T 的账是：耐温够过炉、吸水低、尺寸稳，是高频件里能被客户接受的折中；代价是它的本征损耗不算低，频率往上走之后要看具体件能不能过。
改性 PPE 的账是：射频这一项确实好，这是它的本征优势；但它的耐温、耐溶剂和尺寸稳定性不如尼龙家族，落到要过炉的件上要重新评。
LCP 的账是：极薄壁和低损耗两头都占；代价是各向异性很强，装配和跌落工况要单独验证。
还有一条路线值得单独提：把玻纤换掉一部分。
用矿物、玻璃微珠这类低介电填料替掉高玻纤，介电常数能压下来一截，代价是刚性往下走。
这条路在"介电差一点点、刚性有余量"的件上很划算，但它不是普适解——刚性和介电的取舍点要靠件来定。
一句提醒：换基材体系，等于换一整套界面方案和验证方案。别拿 A 体系的损耗数据去推 B 体系的件。
四、选型判据表（这一页值得收藏）
把上面的约束落成可核对的指标。下表的门限值是方向性建议，不是验收标准——实际数值必须由具体项目、具体工况和实测确定。
| 指标 | 方向性门限 | 验证方法 / 标准 | 常见失效 | 通行解法 | 对应助剂体系 |
|---|
| 介电常数 | 按通道阻抗预算定，通常在 3.0–3.8 区间取 | 谐振腔法或传输线法，注明频率点 | 阻抗偏差、反射变大 | 低吸水基材 + 低介电填料 | 偶联剂 |
| 介质损耗 | 按插损预算定，高频段常要压到 0.01 以下 | 同上，且必须注明测试频率 | 插损超预算、局部发热 | 减极性基团密度 + 降吸水 | 偶联剂 |
| 吸湿后电气漂移 | 调湿态与干态的差值按件核 | 85℃/85%RH 处理后再测 | 装上去才超标 | 要求供应方给调湿态数据 | 不靠助剂 |
| 尺寸稳定（插拔公差） | 常见 0.05 mm 量级 | 调湿前后实测 / ISO 294 | 插拔力变大、卡滞 | 低吸水基材 + 调湿交付 | 不靠助剂 |
| 镀层附着 | 百格或拉脱法合格 | 按客户涂层体系做件级试验 | 起泡、脱落、局部失光 | 控制表面迁移 + 确认涂层相容 | 润滑剂 |
| 长期热氧保留率 | 85℃×1000 h 后按件定 | ISO 527 | 件发白、脆化 | 稳定化体系 | 抗氧剂 |
| 阻燃 | V-0（按件的最小壁厚报） | UL94 / IEC 60695 | 离火不自熄 | 无卤阻燃体系 | — |
怎么用这张表：不要逐行打分。先看前两行。
介电与损耗过不去，后面尺寸和镀层做得再好也没意义，因为信号根本走不通。
一个提醒：表里"介电常数"和"介质损耗"这两行，测试频率点必须写进协议。同一个材料在 1 GHz 和 28 GHz 下是两个不同的数字，不写频率点的数据，等于没有数据。
五、四条常见失效，和它们的真实根因
失效一：样品阶段损耗达标，量产之后插损往上走。
这个现象的根因常常不是配方，是吸湿。
样品在实验室里是干态或半干态，量出来的损耗是一个很乐观的状态。件出了门、在南方仓里放了两周、再装到机柜里跑，吸水之后的数值就变了。
通行解法：把调湿态的损耗数据写进技术协议，干态数据只做过程记录。这一条比换料有效得多。
这类事的时间线，通常是这个形状：
起点：样品在实验室里测，干态，损耗刚好卡在预算线以内，方案通过了。
潜伏：量产件出了门，在南方仓库里放了两周，吸水一点点往里走，没人量过。
爆发：装到机柜里跑起来，某一段通道的插损超了预算，客户先怀疑的是 PCB。
追溯：把出问题的批号和留样拿出来，在湿热箱里过一轮，数据对上了吸湿。
结算：方案没换料，改的是交付状态与验收条件——调湿态的损耗进协议，干态只做记录。
这条线里最贵的一步是"追溯"，因为它要用留样去反推几个月前的状态。
失效二：热氧老化之后，同一批件损耗有差。
根因在助剂侧：稳定化体系的分散不均。
抗氧剂在混料阶段没混开，局部浓度高的地方老化慢，低的地方老化快，损耗自然就散开了。
先查混料工艺与母粒化，别急着换基材。
失效三：金属化之后，镀层局部起泡。
根因在助剂侧：外润滑用量偏高。
它往表面迁移，形成一层很薄的迁移层，和涂层的相容性不对，附着力就上不去。
这一条常被误判成"镀前清洗没做好"，其实先查的是润滑体系。
失效四：插拔力在新件上是好的，装到现场几个月后变大。
根因是吸湿膨胀。0.05 毫米的公差，摊在一件 30 毫米的外壳上，只需要千分之一点七的变化就吃掉了。
这里有一条要直说的：这个失效里最容易被省掉的一项，是量测条件。
同一个件，在干燥柜里量、在车间里量、在恒温恒湿间里量，得到的是三个尺寸。不把量测条件写进协议，争议永远说不清。
一句直说的：高频件的排查顺序是——先定频率点与量测状态，再看湿度史，最后才怀疑牌号。
顺序倒过来，会把一个"状态问题"当成"材料问题"处理掉。
六、加工与验证：几件必须提前定的事
干燥。PA6T、PA9T 这类基材的干燥窗口按实测含水率定，必须用除湿干燥机。
高频件对这一条的要求比普通件更硬，因为水分同时影响尺寸和损耗。
模温。模温影响结晶度，结晶度影响收缩，也影响吸湿速度。
模温开低了，表面结晶不充分，吸湿会比正常件更快——这一条在长周期的现场会慢慢显形。
玻纤取向。高频件多是长条形或薄壳，浇口位置决定纤维取向，取向决定两个方向的收缩差。
件的长方向如果不是流动方向，尺寸稳定性基本靠运气。
金属化的前处理。镀层附着力这件事，模具阶段就要考虑：脱模斜率、表面粗糙度、以及脱模剂的使用方式。
等件做出来再想镀层，能改的空间就很小了。
验证顺序。建议这样排，顺序不要换：
1. 干态电气：介电常数与介质损耗，标清频率点
2. 调湿态电气：85℃/85%RH 处理后再测同一组
3. 尺寸：调湿前后的关键配合尺寸
4. 件级装配：插拔力、定位精度、保持力
5. 涂层与老化：镀层附着力、老化后复测电气
前一项不通过就往下走，后面的数据没有解释意义。
这里有个内行细节：高频件的损耗数据，同一组试样至少要测三个湿度状态——干态、调湿中、接近饱和。
三个点连成一条线，才知道这个件对湿度有多敏感。单点数据只能说明那一刻。
七、边界：哪些高频件不该走改性尼龙
这一段可能比前面六段更值钱。
其一，介质损耗预算压到 0.005 以下的射频通道件。
这个量级已经进入 LCP、改性 PPE、氟塑料甚至陶瓷的地盘。尼龙家族的本征损耗决定了它在这一档上不是合适的选项，靠配方去挤，代价会远高于换体系。
其二，需要长期在 85℃ 与 85% 湿度下保持尺寸精度的配合件。
尼龙的吸湿是本性，配方能压低速度，压不掉终点。要长期尺寸绝对稳固的配合，要么改结构留余量，要么换低吸水体系。
其三，需要承担结构主承力的外壳。
薄壳件的刚性靠玻纤，玻纤又抬介电。这两个要求在同一件上是对着走的，如果这件还要承担主承力，取舍会变得很难看。
其四，需要做完整屏蔽的件。
真正决定屏蔽效果的通常是金属化层或金属结构件，塑料在这里提供的是形状与绝缘。把屏蔽要求押在塑料本身的配方上，方向就偏了。
把这四条写在前面，不是劝退，是省时间。
高频件的验证周期长、成本高，样品阶段很顺、卡在湿热老化复测上的项目，最后往往要整条通道重新评——回退的代价，比一开始就不做高得多。
还有一句要说清：高频连接器和光模块外壳不是一件事。连接器更看介电与插拔定位，外壳更看尺寸、镀层和散热。这两类件的料可能同源，判据的排序不同。
换料风险清单（从 PA66-GF 换到低吸水高频体系，要动的东西）
| 环节 | 要动什么 | 容易漏的点 |
|---|
| 基材体系 | 吸水率、熔点、流动都变，配方要重配 | 沿用原体系的界面方案 |
| 模具 | 收缩率与各向异性变了，配合尺寸要复算 | 只按手册通用收缩率补偿 |
| 干燥 | 换除湿干燥机，温度上抬一档 | 沿用 PA66 的干燥参数 |
| 料温 / 模温 | 料温上抬，模温按结晶要求重定 | 拿原工艺参数直接套 |
| 表面处理 | 脱模与润滑体系要配合镀层要求 | 脱模剂残留影响附着力 |
| 量测条件 | 干态、调湿态分别定测试规范 | 只给一个状态的尺寸 |
| 验证顺序 | 干态电气 → 调湿态电气 → 尺寸 → 装配 → 涂层老化 | 前一项未过就往下走 |
一页纸汇报表（给要向上汇报的人）
| 场景 | 推荐路线 | 关键指标 | 验证标准 | 需先确认的条件 |
|---|
| 高频连接器定位件 | PA6T / PA9T 增强 | 介电常数、介质损耗 | 注明频率点的传输线法 | 频率段与插损预算 |
| 光模块外壳，要过炉 | PA6T / PPA | 尺寸稳定、耐回流焊 | 过炉后尺寸 + 调湿态复测 | 过炉次数与实测峰值 |
| 射频段，不过炉 | 改性 PPE / LCP 方向 | 介质损耗 | 多湿度点电气复测 | 耐温与耐溶剂是否够 |
| 需金属化的外壳 | 低吸水体系 + 涂层方案 | 镀层附着、尺寸 | 百格或拉脱 + 湿热后复测 | 涂层体系相容性 |
风险提示：本路线的主要不确定性在吸湿之后的电气与尺寸漂移，不在初始介电常数。
读者常问的三句
问：物性表上写着介电常数 3.5，是不是就能用？
先看两件事：这个数字是在什么频率下测的，以及是什么湿度状态测的。不带频率点和状态的介电数据，参考价值有限。
问：加了玻纤刚性变好，介电为什么反而更差？
因为玻纤本身的介电常数比树脂高。刚性和介电在玻纤这里是两个相反的方向，需要用低介电填料或结构调整来补。
问：外壳不金属化，能不能只靠塑料本身的屏蔽填料解决？
能解决一部分，通常做法是加导电填料形成一定导电性。但它和金属化层不是同一档效果，屏蔽要求高的件建议把两种方案放在一起评估。
结语
回到开头那位客户。
最后我们没有先给他推荐牌号，而是先陪着把通道的插损预算拆了一遍，又拿两组试样在 85℃/85% 湿度里跑了几轮复测。
复测出来，他的问题主要在吸湿之后的那一段漂移，而不是初始介电常数。
后来动的是三处：基材换成吸水更低的一档、玻纤比例做了一点下调、金属化的前处理工艺跟着改。
高频连接器材料的判断链，说到底只有三条：频率段定损耗预算 → 吸湿量定尺寸与电气的余地 → 量测条件定争议的边界。
三年五年之后还那样——高频件上，这句话不是形容词，是能不能交付的分界线。
三条定完，"这个件能不能做塑料"这个问题自然就有答案了。
如果你手上正有一个高频件或光模块外壳要定料，把三样东西发过来就能给方向：频率段与插损预算、是否做金属化、长期湿热条件。
高频件比别的件更认一件事：材料只是链路里的一段，它做不到单靠自己把整条通道救回来。
我们做的事很具体：把 PA6、PA66、PA46、PA11、PA12、PA6T、PA9T 和尼龙合金这些树脂，改成某个件真正能用的样子；顺带做改性 PPO、PPS 和热塑性弹性体。
也经营各大化工巨头的尼龙树脂、副牌料和大包料现货。另：长期收尼龙原料、水口回料与各类尼龙废料，有正规处置渠道。
配方里的助剂体系按件的工况配——常规助剂常备现货，特殊型号按需配套；你报工况和牌号，料和助剂一次配齐。
这类件的选料与试模，可以一起聊。
Last month, a customer who makes optical module structural components sent a very brief message.
Only nine characters plus a question mark: 'Do you have low-dielectric nylon?'
I went back and first asked three questions: Where is the frequency range of this component? What is the budget for insertion loss? Will the casing be metallized? What are the long-term humidity and heat conditions?
The answer he gave is: a section around 28 GHz, the casing needs to be made conductive, and long-term 85°C and 85% relative humidity.
Answered in three sentences, the outline of this matter changed — what he really wanted to resolve was not 'whether the dielectric is low or not,' but 'whether after being kept at 85℃ and 85% humidity for five years, the size, coating, and loss can still all be maintained.'
This article elaborates on the materials of high-frequency connectors: first, look at what constraints this component has, then look at the positions of several base materials, followed by the criteria table, and finally clarify above which frequency range plastics should not handle this work.
1. Low dielectric is a relative term, so first clarify what 'low' is being compared to.
In the materials field, saying 'low dielectric' is a relative term; you need to find a reference first.
Ordinary nylon is not a low-loss system.
PA6 and PA66 have dense amide groups on their molecular chains, which are polar groups. Polar groups will flip in an alternating electric field, and if they can't flip, they generate heat—this is the physical origin of dielectric loss.
So the dielectric constant of PA66 at 1 GHz is roughly between 3.5 and 4.0, and the dielectric loss is on the order of 0.02 to 0.04.
For comparison: the dielectric loss of LCP and modified PPE is between 0.002 and 0.007.
Off by an order of magnitude. This gap cannot be bridged by a formula; it is determined by the molecular structure.
Secondly, water absorption.
The dielectric constant of water is on the order of 80, much higher than that of any engineering plastic.
Every bit of water absorbed by nylon increases the dielectric constant and the dielectric loss of the material. The equilibrium water absorption of PA66 is 8% to 9%, while PA9T is significantly lower—this is why "low water absorption substrates" are highly regarded in RF components, not just for size, but also for electrical performance.
The third thing: fiberglass.
The dielectric constant of E glass is on the order of 6.6, higher than that of resin.
In other words, for every bit of fiberglass you add for rigidity, you are raising the dielectric constant. If you want rigidity and a low dielectric constant, these two directions are opposite in fiberglass.
So the true meaning of the four words 'low dielectric nylon' is: in a system that is not originally low, try to suppress these three things — few polar groups, low water absorption, and little high-dielectric filler.
In a word: the so-called low-dielectric nylon does not turn nylon into an inherently low-loss material; rather, it keeps the increase in loss within the budget.
2. Operating condition six dimensions: What is the high-frequency component clamped by
Frequency and loss budget. This is the first thing to clarify.
Insertion loss budget is usually allocated according to the entire channel, and the share that a connector or housing can get is limited. The higher the frequency, the greater the attenuation calculated from the same dielectric loss.
The frequency goes from 10 GHz to 28 GHz, and with the same dielectric loss, the attenuation magnitude will more than double. So the same material is sufficient for a 100G component, but it is not enough for a 400G one.
Temperature. High-frequency components are divided into two stages: the peak of 240 to 260°C during the few tens of seconds in the furnace, and the long-term operating temperature.
The long-term temperature of the optical module enclosure is usually considered to be between 70 and 85°C, and it will be higher in densely stacked cabinets. The long-term value is what needs to be monitored.
Humidity. This is the dimension that is most easily underestimated in high-frequency components.
85°C and 85% relative humidity are common accelerated test conditions in the industry. Under this condition, the moisture absorption of nylon will approach saturation.
After absorbing water, the size increases, the dielectric constant rises, and the dielectric loss rises; all three things happen simultaneously.
Load. Low-frequency components discuss plug-in and unplug forces, while high-frequency components discuss the positioning accuracy during plugging and unplugging.
The insertion tolerance of optical modules is commonly on the order of 0.05 millimeters; even a slight increase in size can change the insertion and removal force from 'just right' to 'stuck'.
Appearance and craftsmanship. The casing often needs to be metallized or sprayed with conductive coating, and the adhesion of the plating is a critical factor.
Once there is a transfer layer on the surface, the coating cannot hold—it’s the same issue as the surface precipitation mentioned in the article about the glue core.
Compliant. Flame-retardant V-0, halogen-free, RoHS, as well as additional requirements from some customers regarding volatility and leaching.
In these six dimensions, frequency and humidity are unique to high-frequency components, while the remaining four dimensions share the same logic as ordinary connectors.
3. Several routes, each with its own merits
Lay out the substrates that can make this part; there are roughly five routes.
| Route | Dielectric constant (on the order of 1 GHz) | Magnitude of dielectric loss | The position on this piece |
|---|
| PA6T / PA9T Reinforced | About 3.4–3.8 | About 0.01–0.02 | Balances reflow soldering resistance with dimensional stability; dielectric constant increases with high glass fiber content |
| PPA (Semi-Aromatic Copolymer) | About 3.3–3.8 | Approximately 0.008–0.015 | Balance temperature resistance and low absorption, and make the processing window a bit wider |
| Modified PPE / PPO | About 2.5–2.9 | Approximately 0.002–0.007 | Good RF performance; temperature resistance and solvent resistance need to be evaluated separately |
| LCP | About 3.0–3.3 | About 0.002–0.005 | Extremely thin walls and advantages in high-frequency bands are obvious; toughness and flow directionality are strong |
| PPS / PEEK | About 3.0–3.2 | About 0.002–0.005 | Good temperature resistance and chemical stability; price and molding process involve trade-offs |
Don't make a 'who is better' conclusion, just talk about the differences.
The situation with PA6T and PA9T is: they can withstand enough heat, have low water absorption, and stable dimensions, making them a compromise that customers can accept in high-frequency components; the cost is that their intrinsic losses are not low, and as the frequency increases, it depends on whether the specific component can handle it.
The account for modified PPE is: the RF aspect is indeed good, that's its intrinsic advantage; however, its temperature resistance, solvent resistance, and dimensional stability are not as good as the nylon family, so for parts that need to go through the oven, it needs to be re-evaluated.
The characteristics of LCP are: it has both ultra-thin walls and low loss; the trade-off is that it is highly anisotropic, so assembly and drop conditions need to be verified separately.
There is also one route worth mentioning separately: replacing part of the fiberglass.
Replacing high glass fiber with low-dielectric fillers such as minerals and glass beads can bring down the dielectric constant, at the cost of reduced rigidity.
This route is very cost-effective for parts that are 'slightly lacking in dielectric but have excess rigidity,' but it is not a universal solution—the trade-off point between rigidity and dielectric must be determined by the part.
A reminder: Switching the substrate system is equivalent to changing the entire set of interface solutions and verification plans. Don't use the loss data from system A to infer parts of system B.
4. Selection Criteria Table (This page is worth saving)
Translate the above constraints into verifiable indicators. The threshold values in the table are directional suggestions, not acceptance criteria—the actual values must be determined by specific projects, specific working conditions, and actual measurements.
| Indicator | Directional threshold | Verification Method / Standard | Common Failures | Common solution | Corresponding auxiliary agent system |
|---|
| Dielectric constant | Based on the channel impedance budget, it is usually taken in the range of 3.0–3.8 | Resonant cavity method or transmission line method, specify the frequency point | Impedance deviation, increased reflection | Low water-absorption substrate Low dielectric filler | Coupling agent |
| Dielectric Loss | According to the insertion loss budget, the high-frequency band often needs to be reduced below 0.01. | Same as above, and the test frequency must be specified | Insertion loss exceeds budget, local overheating | Reduce the density of polar groups to decrease water absorption | Coupling agent |
| Electrical drift after moisture absorption | The difference between moisture-conditioned state and dry state is calculated per piece | Test again after 85°C/85% RH treatment | It's only over the limit when installed | Request the supplier to provide the data in the conditioned state | Without additives |
| Dimensional stability (plug-in tolerance) | Common 0.05 mm scale | Measured before and after humidity adjustment / ISO 294 | Increased insertion and extraction force, sticking | Low water-absorption substrate Moisture-adjusted delivery | Without additives |
| Coating adhesion | Paige or Lathrop method qualified | Conduct part-level tests according to the customer's coating system | Blistering, peeling, localized loss of gloss | Control surface migration Confirm coating compatibility | Lubricant |
| Long-term thermal oxygen retention rate | After 85℃ × 1000 h, determine per item | ISO 527 | whitish and brittle | Stabilization system | Antioxidant |
| Flame retardant | V-0 (reported according to the minimum wall thickness per piece) | UL94 / IEC 60695 | A departing fire does not extinguish itself | Halogen-free flame retardant system | — |
How to use this table: Do not score line by line. First, look at the first two rows.
If the dielectric and loss are unacceptable, no matter how well the dimensions and coatings are made later, it is meaningless because the signal simply cannot pass through.
A reminder: For the rows 'Dielectric Constant' and 'Dielectric Loss', the test frequency points must be included in the protocol. The same material has different values at 1 GHz and 28 GHz, and data without frequency points is equivalent to no data.
Five, four common failures and their real root causes
Failure 1: The sample stage meets the loss standard, but the insertion loss increases after mass production.
The root cause of this phenomenon is often not the formula, but moisture absorption.
The samples in the laboratory are in a dry or semi-dry state, so the measured loss is a very optimistic scenario. Once they leave the door, are stored in the southern warehouse for two weeks, and then installed in the cabinet for operation, the values change after absorbing moisture.
Common solution: write the loss data of the conditioned state into the technical agreement, and only record the dry state data as a process record. This is much more effective than changing materials.
The timeline for this kind of matter is usually shaped like this:
Starting point: The sample was tested in the laboratory, in a dry state, and the loss just stayed within the budget line, so the plan was approved.
Latent: The mass-produced items left the factory and were stored in the southern warehouse for two weeks. They absorbed a little water slowly, but no one measured it.
Outbreak: When installed in the cabinet and running, the insertion loss of a certain section of the channel exceeded the budget, and the customer first suspected the PCB.
Traceback: Take out the problematic batch number and sample, run them through a cycle in the humidity and heat chamber, and the data matches after moisture absorption.
Settlement: The plan did not change the materials; what was changed were the delivery status and acceptance conditions — losses in the conditioned state are included in the agreement, while those in the dry state are only recorded.
The most expensive step in this process is 'tracing,' because it requires using samples to deduce the state from several months ago.
Failure 2: After thermo-oxidative aging, there is variability in the loss among the same batch of parts.
The root cause lies on the additive side: uneven dispersion of the stabilization system.
If the antioxidant is not evenly mixed during the mixing stage, areas with high local concentration will age slowly, while areas with low concentration will age quickly, and the consumption will naturally disperse.
First check the blending process and masterbatching; don't rush to change the substrate.
Failure 3: After metallization, the coating shows localized blistering.
The root cause lies on the additive side: the amount of external lubricant is relatively high.
It migrates to the surface, forming a very thin migration layer. If it is not compatible with the coating, the adhesion will not improve.
This item is often mistakenly judged as 'pre-plating cleaning not done properly,' but actually the first thing to check is the lubrication system.
Failure 4: The insertion and extraction force is good on the new part, but increases after being installed on site for a few months.
The root cause is moisture absorption and expansion. A tolerance of 0.05 millimeters, spread over a 30-millimeter casing, only requires a change of one-thousandth seven to consume it.
Here is something that needs to be said directly: the item most likely to be omitted in this failure analysis is the measurement conditions.
Measuring the same part in a drying cabinet, in the workshop, and in a temperature- and humidity-controlled room yields three different dimensions. If the measurement conditions are not written into the agreement, disputes can never be clearly resolved.
A straightforward statement: The order for inspecting high-frequency components is—first determine the frequency points and measurement conditions, then look at the humidity history, and only finally suspect the grade.
If the order is reversed, a 'status problem' will be treated as a 'material problem'.
6. Processing and Verification: Several Things That Must Be Decided in Advance
Drying. The drying window for substrates like PA6T and PA9T should be determined according to the actual measured moisture content, and a dehumidifying dryer must be used.
High-frequency parts have stricter requirements for this aspect than ordinary parts, because moisture affects both dimensions and wear.
Mold temperature. Mold temperature affects crystallinity, which affects shrinkage and also influences moisture absorption rate.
The mold temperature is set too low, resulting in insufficient surface crystallization. Moisture absorption will be faster than normal parts — this will gradually become apparent in long-term field operations.
Glass fiber orientation. High-frequency parts are mostly elongated or thin-walled, and the gate position determines the fiber orientation, which in turn determines the shrinkage difference in two directions.
If the long direction of the part is not the flow direction, dimensional stability basically relies on luck.
Metalized pretreatment. The adhesion of the coating should be considered during the molding stage: draft angle, surface roughness, and the way mold release agents are used.
Wait until the part is made before thinking about the coating; the room for adjustments will be very limited.
Verification order. It is recommended to arrange it like this; do not change the order:
1. Dry-state electricity: dielectric constant and dielectric loss, standard frequency point
2. Humidity-conditioned electrical: test the same group again after treatment at 85℃/85% RH
3. Dimensions: Key fitting dimensions before and after humidity adjustment
4. Component-level assembly: insertion/extraction force, positioning accuracy, retention force
5. Coating and aging: coating adhesion, electrical re-measurement after aging
If the previous item fails, just move on; the subsequent data has no explanatory value.
Here's an insider detail: for wear data of high-frequency parts, the same set of samples needs to be measured in at least three humidity conditions—dry state, during moisture adjustment, and near saturation.
Only when three points are connected in a line can one understand how sensitive this component is to humidity. Data from a single point can only indicate that moment.
7. Boundaries: Which high-frequency parts should not use modified nylon
This section might be more valuable than the previous six sections.
First, RF channel components with dielectric loss budget reduced below 0.005.
This scale has already entered the territory of LCP, modified PPE, fluoroplastics, and even ceramics. The intrinsic loss of the nylon family determines that it is not a suitable option at this level; trying to force it with formulations will cost much more than switching to a different system.
Secondly, matching parts that need to maintain dimensional accuracy for a long time at 85°C and 85% humidity.
Nylon's moisture absorption is inherent; formulations can reduce the rate, but cannot eliminate the final amount. For combinations that require absolutely stable dimensions over the long term, you either need to modify the structure to allow for extra tolerance, or switch to a low-moisture-absorption system.
Third, it needs to bear the main structural load of the shell.
The rigidity of thin-shell parts relies on fiberglass, and fiberglass also raises the dielectric constant. These two requirements are opposing in the same part, and if the part also has to bear the main load, the trade-off becomes very ugly.
Fourth, it is necessary to make the parts fully shielded.
What usually determines the shielding effect is the metallized layer or metal components; the plastic here provides shape and insulation. Relying on the plastic's own formulation for shielding requirements is the wrong direction.
Writing these four points first is not to discourage, but to save time.
The verification cycle for high-frequency components is long and costly. Projects that go smoothly at the sample stage often get stuck at the wet heat aging retest, and in the end, the entire channel often has to be re-evaluated—the cost of rolling back is much higher than not doing it in the first place.
There is one more thing to clarify: high-frequency connectors and optical module housings are not the same thing. Connectors focus more on dielectric properties and mating alignment, while housings focus more on dimensions, plating, and heat dissipation. The materials for these two types of components may come from the same source, but the criteria for evaluation are prioritized differently.
Material Change Risk List (From PA66-GF to low moisture absorption high-frequency system, things that need to be changed)
| link; segment; part | What needs to be moved? | Points that are easy to overlook |
|---|
| Substrate System | Water absorption, melting point, and flow have all changed, the formula needs to be rebalanced. | Interface solution based on the original system |
| Mold | The shrinkage rate and anisotropy have changed, and the matching dimensions need to be recalculated. | Only compensate according to the general shrinkage rate in the manual |
| Dry | Replace the dehumidifying dryer and raise the temperature by one level | Continue using the drying parameters of PA66 |
| Material Temperature / Mold Temperature | Raise the material temperature, and reset the mold temperature according to crystallization requirements | Directly apply the original process parameters |
| Surface treatment | The demolding and lubrication system must be matched with the coating requirements | Release agent residue affects adhesion |
| Measurement Conditions | Testing specifications are defined separately for dry state and humidity-conditioned state | Only give the dimensions for one state |
| Verification order | Dry-state electrical → Humidity-conditioned electrical → Dimensions → Assembly → Coating aging | If the previous item fails, just move on. |
One-page report form (for people who need to report upwards)
| Scene | Recommended Route | Key indicators | Verification Standard | Conditions that need to be confirmed first |
|---|
| High-frequency connector positioning component | PA6T / PA9T Reinforced | Dielectric constant, dielectric loss | Transmission line method with specified frequency points | Frequency Band and Insertion Loss Budget |
| The optical module casing needs to go through the oven. | PA6T / PPA | Dimensionally stable and resistant to reflow soldering | Dimensions after firing Retest in conditioned state | Number of passes through the furnace and measured peak value |
| RF band, but not the oven | Modified PPE / LCP direction | Dielectric Loss | Electrical re-inspection at multiple humidity points | Is the temperature resistance and solvent resistance sufficient? |
| Shell that requires metallization | Low Water Absorption System Coating Solution | Coating adhesion, dimensions | Bai Ge or La Tuo Retest after damp-heat | Coating system compatibility |
Risk warning: The main uncertainty of this route lies in electrical and dimensional drift after moisture absorption, not in the initial dielectric constant.
Three questions readers often ask
Question: The material property table lists a dielectric constant of 3.5. Does that mean it can be used?
First, look at two things: the frequency at which this number was measured, and the humidity condition under which it was measured. Dielectric data without frequency points and conditions have limited reference value.
Q: Adding glass fiber improves rigidity, so why does the dielectric property get worse?
Because the dielectric constant of glass fiber itself is higher than that of resin. Stiffness and dielectric properties are two opposite directions in glass fiber, so low-dielectric fillers or structural adjustments are needed to compensate.
Question: If the casing is not metallized, can the shielding filler in the plastic itself solve the problem?
It can solve part of the problem; the usual approach is to add conductive fillers to achieve a certain level of conductivity. However, it is not equivalent to a metallized layer. For components with high shielding requirements, it is recommended to evaluate both options together.
Conclusion
Back to the first customer.
In the end, we didn't recommend the grade to him first. Instead, we went through the insertion loss budget of the channel together and ran a few rounds of retesting on two sets of samples at 85°C/85% humidity.
Upon retesting, his problem mainly lies in the drift after moisture absorption, rather than the initial dielectric constant.
Later, three changes were made: the substrate was replaced with a grade that absorbs less water, the fiberglass ratio was slightly reduced, and the metallization pretreatment process was also modified.
The chain for determining materials for high-frequency connectors ultimately comes down to three things: frequency range sets the insertion loss budget → moisture absorption determines the allowable size and electrical margins → measurement conditions define the boundaries of disputes.
If it's still the same three or five years later—for high-frequency components, this is not an adjective, it is the dividing line for whether it can be delivered.
Once the three are determined, the question "Can this part be made of plastic" naturally has an answer.
If you currently have a high-frequency part or optical module housing to select materials for, just send over three things and we can give direction: frequency range and insertion loss budget, whether to use metallization, long-term hot and humid conditions.
High-frequency components are more aware than others of one fact: the material is only a segment in the link; it cannot alone save the entire channel.
What we do is very specific: turning PA6, PA66, PA46, PA11, PA12, PA6T, PA9T, and nylon alloys into forms that can actually be used for a given component; we also handle modified PPO, PPS, and thermoplastic elastomers.
We also deal in large chemical manufacturers' nylon resins, secondary brands, and bulk in-stock materials. Additionally: we have long-term collection of nylon raw materials, sprue scraps, and various nylon wastes with proper disposal channels.
The additive systems in the formulations are tailored to the working conditions of each part—common additives are kept in stock, special types are provided as needed; you report the working conditions and grade, and the materials and additives are prepared all at once.
The material selection and mold trial for this type of part can be discussed together.
