有个做气动元件的客户，第一次联系我们的时候，语气里带着明显的挫败。
他新开发一个阀体，用量不大，一个月大概三百公斤。找了两家大型改性厂，一家说"最少一吨起订"，另一家报了个价——比他预算高出将近一倍。他把报价转给我，附了一句："是不是他们看不上我们这种小单？"
我让他把这两份报价留着，先跟他说了另一件事：这不是看不看得上的问题，是双螺杆挤出机开机那一刻，账就已经定了。
他后来走的路是先试 25 公斤样料，验证方向；然后用 300 公斤跑工艺；最后才谈到 1 吨的量产订单。整个流程比他原来设想的多了两个月，但到最后谈价的时候，他手上有的是一套验证过的参数和一个跑顺的量产节点——这时候议价的位置完全不同。
很多项目卡在最开头这一步：量太小。
几十公斤的料，大供应商不接；接了，单价高得离谱，还得另算打样费。于是采购就在"多买点摊薄"和"少买点但贵"之间来回纠结。
其实这条路的正确走法，不是纠结量，而是分三段走。
一、小批量为什么难：不是态度问题，是成本结构
改性尼龙是双螺杆造粒出来的。开一次机，要先洗机、过渡、调参数。这几百公斤的损耗是固定的。
批量越小，这份固定成本越摊不开。
这里值得把"几百公斤"这个说法拆开，因为它不是随口说的量，是可以算出来的：
洗机：上一次生产留在螺杆和机筒里的料，要用新配方的基础树脂或专用洗机料顶出去。一台中型双螺杆的机筒里，滞留的物料本身就是几十公斤量级，而且不可能一次顶净——要顶到颜色、熔指都稳定为止
过渡段：参数刚调上去的时候，出来的粒子各项指标还在爬坡。这一段的产品既不确定能不能要，也谈不上合格品，通常是打回同类别的低端用途或降级处理
调参：玻纤含量、阻燃分散、熔指窗口，每一项都要取样的过程中调整，每调一次就意味着又一段过渡料
这三部分加起来构成一次开机的门槛。 它对所有的订单都一样——不管你要 25 公斤还是 5 吨，这份门槛是照付的。
所以小批量的高价，不是"看人下菜碟"，是分摊的结果。理解这一点，谈判的方式就变了——不是求人便宜，而是想办法减少分摊项。
这句话可以换算成一笔很直观的账。假设一次开机的固定损耗折合成本大致是 X：
| 订单量 | 分摊到每公斤的固定成本 |
|---|
| 25 公斤 | X ÷ 25 |
| 300 公斤 | X ÷ 300 |
| 3000 公斤 | X ÷ 3000 |
同样是付了那次开机，25 公斤的订单里，固定成本是每公斤单价里最大的一块；到 3 吨的时候，它已经小到可以忽略。这就是"为什么买得越少越贵"的全部答案——贵的不是料，是那次开机被摊在了多少公斤上。
既然如此，能谈的就只有两件事：一是让这次开机少浪费一点，二是让这次开机后面跟着更大的量。
比如：选一个已有的相近牌号做微调，就不用重新洗机试方；先用现货样做初筛，方向对了再谈定制。这些都是把分摊项压下来的做法，比砍价有效得多。
二、小批量分三段，走法完全不同
| 阶段 | 量级 | 目的 | 走法 |
|---|
| 试样确认 | 25-50kg | 确认这个方向能不能用 | 要现货样或分装样 |
| 小批试产 | 200-500kg | 跑通工艺 + 验证性能 | 让改性厂做小批定制 |
| 试产转量产 | 1-3吨 | 验证批次稳定性 | 定牌号、锁批号 |
这三段的顺序不能倒。
这三段各自要解决的是不同的问题，这也是它们不能合并的原因：
第一段解决的是"方向对不对"——这一批次做出来的是不是能用，永远比性能高不高重要
第二段解决的是"能不能复制"——实验室跑出来的参数，在连续生产几百公斤的时候还能不能成立
第三段解决的是"会不会漂移"——连续三批是不是同一个东西
见过不少项目直接跳到第三段——拿一个"看起来对"的牌号直接放量，结果第二批就不一样，只能回头补做验证。省掉的中间一步，最后都要用返工补回来。
为什么"看起来对"常常不对？因为牌号相同不等于批次相同。同一个牌号在两个生产批次之间，原料批次、螺杆磨损程度、环境湿度都可能不同。但这些差异只有在连续生产、连续取样的时候才看得出来——25 公斤的试样跑得再漂亮，也只能说明那 25 公斤是好的。
三、打样阶段必须问清的六件事
1. 用途与工况——不要说"我要 PA66"，要说做什么件、受什么力、什么温度、什么介质
2. 目标指标——灰分、耐温、阻燃、CTI，能写清就写清
3. 要样料还是试产料——25kg 和 300kg 是两件事，报价逻辑都不同
4. 打样费怎么算、谁承担——先谈，别做完再谈
5. 打样周期——急项目尤其要提前确认
6. 试产成功后的量产衔接——牌号能不能锁定、批号能不能延续
把这六件事一次说全，能省掉好几轮回合。来回一次，往往就是一周。
这里特别说一下第 1 条，因为它最常被敷衍过去。很多人以为"用途工况"是客套话，实际它决定了后面五件事的答案：
没说温度，就定不了用 PA6 还是 PA66（这一步选错，后面全废）
没说介质，就不知道要不要抗水解（碰上冷却液、乙二醇、热水，这是两年内必出事的一条）
没说受力，就定不了玻纤含量（GF15 和 GF30 的刚性差着一档）
没说是不是外观件，就漏掉了浮纤和色差这两个最常见的打样失败原因
所以"把工况说全"这一件事，本质上是在替你自己省掉两到三轮试错。而在双螺杆开机这个语境里，每一轮试错意味着一次开机的钱。
四、怎么把打样成本降下来
一是选相近的现有牌号做微调，不要从零开发。 从零开发的打样费和时间都是另一个量级，微调通常便宜一个数量级。
原因在于：成熟牌号的工艺是跑过的，螺杆组合、温度分区、加料顺序都不用重新试；所谓微调，通常是在玻纤含量、色母比例、某一项助剂的加量上动一动，这些都是在线参数，不需要重新洗摸索。
二是先用现货样做初筛。 方向对了再谈定制，避免在错的方向上做定制。
三是一次把工况说全。 减少反复试错，每一轮试错都是时间和料钱。
四是有后续量的，把量讲清楚。 很多供应商愿意把打样费在量产订单里抵扣。
这一条要提前讲，不要藏着。很多采购担心一说量小就被怠慢，结果反过来：把后续规划讲清楚的采购，往往更容易拿到打样费的宽免。 因为对方判断的不是你现在买多少，是这一单值不值得投入。
这四条里，最有效的是第一条。微调比开发省得多，前提是你的工况说得足够清楚。
五、打样阶段最容易犯的两个错
第一个：先要料，再想工况。 结果是拿到料也不知道对不对，试了等于没试，还多花一轮时间。
第二个：小样跑通了就直接量产。 小样和量产料的工艺与批次都可能不同。中间缺了"批次验证"这一步，风险就全压在量产线上——而量产线上的损失，是小样的几百倍。
这个"几百倍"不是修辞。25 公斤试样废掉，损失是几百块钱和两天时间；1 吨量产跑出不确认的批次，损失是料钱、返工工时、下游客户的停线索赔，以及一段很难修补的信任。两者之间隔着的不只是金额，还有不可逆性——小样废了可以重来，量产废了客户可能就不再来了。
批次验证的具体做法，不是再买一吨，而是在同一牌号下连续要两到三个不同批号的样料，测关键指标的一致性。批次差异小的，量产风险就低；差异大的，说明这个牌号在这个来源上本身就不稳，要提前换方向。这一步成本很低，但能挡掉后面最大的风险。
这里有个顺序上的巧劲：批号验证适合安排在第二段（小批试产）做，而不是拖到第三段。 因为第二段的 300 公斤本来就是分几批出的，正好可以拿来对比批次差异——同一订单内部的批次一致性，本身就是一条几乎免费的信息。等到第三段再验证，发现问题的时候能做的选择已经很少了：模具已经定了，客户已经定了交期，留给你换料的窗口可能只有几天。把这一轮验证往前挪一段，代价接近于零，价值却完全不同。
六、试样阶段要做的三个验收动作
很多人把"试样"当成"拿到料试一下"，其实试样阶段有三个动作要做完整，否则试了也白试：
一是对照。 拿试样件和现有件做对比测试，而不是只看"能不能做出来"。目标指标达不到，做出来了也不能用。
二是记录。 把这次的干燥条件、模温、注射参数记下来。试样最大的价值不是结果本身，是这套参数能不能复制到量产。
这里多说一句为什么"记录"值钱：试样和量产的差别往往不在材料，在工艺。干燥温度差 10℃、模温差 15℃、保压时间差 0.5 秒，出来的件在尺寸和外观上都可能差一个等级。试样阶段是唯一有机会把参数试到位的阶段——因为这时候改参数是免费的，量产后改参数是要停线的。
三是留样。 每一批留一份料样和件样。后面出问题，没有留样就没有对照，排查只能靠猜。
这三个动作做全了，试样才叫完成；只做出一个件，那只是"打过样"。
七、试样之后：把小样转成量产的四件事
试样通过了，还有四件事要把小样和量产连起来：
一是有没有替代方案。 万一这个牌号后面断供或涨价，有没有已验证的第二选择。试样阶段是问这个问题最省事的时候——因为工艺参数还在手上，再试一版成本最低。
二是工艺文件化。 把试样的干燥、模温、注射参数写成正式工艺卡，不能只留在工程师脑子里。人一换，参数就丢了。
这一条在注塑厂特别容易出问题。试样的参数往往掌握在一两个老师傅手里，他们说"加两秒保压"是靠手感，不是靠记录。等到换班、换人，同一批料打出来的件就不一样，然后第一个被怀疑的就是材料。而实际上材料没变，是人没了。
三是包装与标识规范。 小样和大货的包装、标识要统一，避免产线上拿错料。
四是价格与供货确认。 试样成功后再谈价格和年用量，比一开始就谈更有底气。
这四件事做完，小样才算真正转成量产。 少了哪一件，后面都会以别的方式补回来。
结语
回到开头那个客户。
他后来养成了一个习惯：每开发一个新件，先跟我们把工况和年用量说清楚，再决定从哪一段切入。上个月他那个阀体已经跑到量产的第三批，中间没有返工。
而当初让他纠结的那两份报价，一份是把固定损耗摊在了 25 公斤上，一份是把起订量设成了他吃不下的一吨——两份都对，只是都不适合那个项目当时所在的阶段。
小批量不是"少了点"，是"顺序不一样"。
先试样、再试产、后放量——这三步的顺序，能省掉的返工，比省下的料钱多得多。
手上有正在打样的项目，把工况发过来，我们帮你看看该从哪一段切入。
There is a customer who makes pneumatic components. When they contacted us for the first time, their tone was clearly filled with frustration.
He newly developed a valve body, the usage is not large, about three hundred kilograms per month. He approached two large modification factories, one said 'minimum order is one ton,' the other quoted a price — nearly double his budget. He forwarded the quote to me, with a sentence attached: 'Could it be that they look down on small orders like ours?'
I asked him to keep these two quotations and first told him about another matter: this is not a question of whether it is valued or not, the moment the twin-screw extruder starts, the deal is already made.
The route he later took was to first test 25 kilograms of sample material to verify the direction; then use 300 kilograms to run the process; and only finally discuss a 1-ton mass production order. The entire process took two months longer than he originally envisioned, but by the time it came to negotiating the price, he had in hand a set of verified parameters and a smoothly running mass production node—at this point, his position in bargaining was completely different.
Many projects get stuck at the very first step: the quantity is too small.
For dozens of kilograms of material, big suppliers won't take orders; if they do, the unit price is outrageously high, and a separate sample fee must be calculated. So the purchaser is caught between 'buy more to dilute the cost' and 'buy less but pay more'.
Actually, the correct way to walk this path is not to worry about the distance, but to walk it in three sections.
1. Why small-batch production is difficult: It's not an attitude problem, it's the cost structure
Modified nylon is granulated using a twin-screw extruder. Each time the machine is started, it needs to be cleaned, transitioned, and the parameters adjusted. The loss of several hundred kilograms is fixed.
The smaller the batch, the less this fixed cost can be spread out.
It is worth breaking down the term 'hundreds of kilograms' here, because it is not a casually mentioned amount; it can be calculated:
Machine cleaning: The material left in the screw and barrel from the last production needs to be pushed out using the base resin of the new formula or a special cleaning material. In the barrel of a medium-sized twin-screw machine, the material that remains can be dozens of kilograms, and it is impossible to push it all out at once — it must be pushed out until the color and melt index are stable.
Transition phase: When the parameters are just set, the particles that come out are still ramping up in various indicators. Products from this phase are neither certain to be usable nor considered qualified products; they are usually sent back for lower-end uses of the same category or downgraded for processing.
Parameter adjustment: glass fiber content, flame retardant dispersion, melt flow index window; each item needs to be adjusted during the sampling process, and every adjustment means another batch of transition material.
These three parts together constitute the threshold for starting up. It is the same for all orders—whether you want 25 kilograms or 5 tons, this threshold is paid in the same way.
So the high price for small batches is not 'charging differently for different people,' but the result of cost allocation. Understanding this point changes the way you negotiate — it's not about asking for a lower price, but about finding ways to reduce the allocated costs.
This sentence can be converted into a very straightforward calculation. Suppose the fixed loss cost for a single startup is roughly X:
| Order volume | Fixed cost per kilogram |
|---|
| 25 kilograms | X ÷ 25 |
| 300 kilograms | X ÷ 300 |
| 3000 kilograms | X ÷ 3000 |
Just like paying for that startup, in a 25-kilogram order, the fixed cost is the largest part of the price per kilogram; by the time it reaches 3 tons, it has already become negligible. This is the whole answer to 'why buying less is more expensive'—what is expensive is not the material, but how many kilograms that startup cost is spread over.
Since that's the case, there are only two things to discuss: one is to make this startup waste less, and the other is to ensure that a larger volume follows this startup.
For example: choosing an existing similar grade for minor adjustments means you don't have to clean the machine and test a new formulation; using stock samples for preliminary screening first, and only discussing customization if the direction is correct. These are all ways to reduce shared costs, which are much more effective than haggling.
2. Small batches are divided into three stages, with completely different approaches
| Stage | Magnitude | Purpose | Way of walking |
|---|
| Sample Confirmation | 25-50 kg | Confirm whether this direction can be used | Do you want a spot sample or a split sample? |
| Small batch trial production | 200-500 kg | Run the process Verify performance | Have the modification factory make small-batch custom orders |
| Pilot production to mass production | 1-3 tons | Verify batch stability | Specify batch number, lock batch number |
The order of these three sections cannot be reversed.
Each of these three sections addresses a different problem, which is also the reason they cannot be combined.
The first section addresses whether the 'direction is correct'—whether this batch is usable is always more important than whether its performance is high.
The second part addresses whether it can be replicated — whether the parameters obtained in the lab can still hold when producing several hundred kilograms continuously.
The third section addresses 'Will it drift?' — whether three consecutive batches are the same thing
I have seen quite a few projects jump directly to the third stage—taking a brand that 'looks right' and scaling up, only to find the second batch is different, forcing them to go back and redo the verification. The intermediate step that was skipped inevitably ends up needing rework to make up for it.
Why does "looking correct" often turn out to be incorrect? Because having the same grade does not mean the same batch. Even for the same grade, different production batches may have variations in raw material batches, screw wear, and environmental humidity. But these differences can only be seen during continuous production and continuous sampling—no matter how perfect a 25-kilogram sample runs, it only shows that those 25 kilograms are good.
3. Six things you must ask during the proofing stage
1. Purpose and working conditions — Don’t say 'I want PA66,' say what part it is for, what force it will endure, what temperature, and what medium.
2. Target indicators — ash content, temperature resistance, flame retardancy, CTI; write clearly if possible
3. Do you want sample material or trial production material — 25kg and 300kg are two different things, and the pricing logic is different for each.
4. How the proofing fee is calculated and who bears it — discuss it first, don't talk about it after it's done
5. Proofing cycle — urgent projects especially need to be confirmed in advance
6. Connection from trial production success to mass production — whether the grade can be locked and whether the batch number can be continued
If you cover these six things all at once, you can save several rounds. Going back and forth once often takes a week.
I want to specifically talk about Item 1 here, because it is the one most often brushed aside. Many people think that 'usage conditions' is just a formality, but in fact, it determines the answers to the next five things:
If the temperature isn't specified, you can't decide whether to use PA6 or PA66 (if you choose wrong at this step, everything afterward is wasted)
If the medium isn’t specified, you won’t know whether it needs hydrolysis resistance (coming into contact with coolant, ethylene glycol, or hot water, this is likely to cause problems within two years).
Without mentioning the stress, the glass fiber content cannot be determined (the rigidity difference between GF15 and GF30 is one level).
It wasn’t mentioned whether it was an exterior part, but it missed the two most common reasons for sample failures: floating fibers and color differences.
So, 'fully describing the operating conditions' is essentially a way to save yourself two to three rounds of trial and error. And in the context of starting up a twin-screw machine, each round of trial and error means the cost of one startup.
4. How to lower sample production costs
First, choose similar existing grades for fine-tuning; don't develop from scratch. The cost and time for prototyping from scratch are on a whole different scale, and fine-tuning is usually an order of magnitude cheaper.
The reason is: mature grades have already been passed through processes; screw assembly, temperature zoning, and feeding sequence don't need to be retested; Fine-tuning usually means tweaking glass fiber content, masterbatch ratio, or dosing of a certain additive—these are online parameters and don't need to be re-examined.
Second, use in-stock samples for initial screening. Only discuss customization when the direction is right, avoiding customization in the wrong direction.
Third, explain the entire working condition in one go. Reduce repeated trial and error; each round is a cost of time and materials.
Fourth, there is follow-up volume; clearly state the quantity. Many suppliers are willing to deduct the sample fee from mass production orders.
This point should be explained in advance; don't hide it. Many buyers worry that if they mention a small quantity, they will be neglected, but the opposite happens: purchasers who clearly explain subsequent plans are often more likely to get exemption from the sample fee. Because the other party is not judging how much you buy now, but whether the order is worth investing.
Among these four points, the first is the most effective. Fine-tuning is much cheaper than development, provided your working conditions are clearly explained.
Fifth, the two most common mistakes during the prototyping stage
First: first ask for materials, then consider working conditions. The result is that even when you get the material, you don't know if it's right; trying it is like not trying, and it takes an extra round.
Second: Once the sample works, mass production goes straight to production. The process and batch sizes for small samples and mass-produced materials may differ. Without the "batch validation" step, all the risk is placed on the mass production line—and losses on the production line are hundreds of times greater than for small samples.
This "hundreds of times" is not rhetoric. If a 25-kilogram sample is wasted, the loss is a few hundred yuan and two days; If a 1-ton mass production runs an unconfirmed batch, the losses are material costs, rework hours, downstream customer downtime claims, and a trust that is hard to repair. Between the two is not just the amount of money, but also irreversibility—if a small sample is scrapped, it can be redone; if mass production is scrapped, customers may not come.
The specific method for batch verification is not to buy another ton, but to continuously take two to three samples of different batch numbers under the same grade to test the consistency of key indicators. If the batch variation is small, the risk of mass production is low; If the difference is large, it means the grade is already unstable at the source and needs to change direction early. This step is very costly but can mitigate the biggest risk later.
There is a clever sequential trick: batch number verification is best done in the second stage (small batch trial production), rather than delayed until the third stage. Because the 300 kg batch in the second batch is originally produced in several batches, it can be used to compare batch differences—the consistency within the same order is almost free information. By the third stage of verification, when problems are found, there are very few options left: the mold is set, the customer has set the delivery date, and you may only have a few days to change the material. Moving this round of verification forward one step costs nearly zero, but the value is completely different.
Sixth, three acceptance actions to be done during the sample stage
Many people think "samples" are like "getting the material and testing it," but in fact, there are three steps to complete during the sample stage; otherwise, trying is pointless:
First, comparison. Compare the sample with existing parts, not just "can it be made." If the target indicators are not met, even if made, it cannot be used.
The second is to record. Record the drying conditions, mold temperature, and injection parameters for this time. The greatest value of a sample isn't the result itself, but whether the parameters can be replicated for mass production.
Here's one more thing to say why "recording" is valuable: the difference between samples and mass production often lies not in the material, but in the process. A 10°C difference in drying temperature, 15°C in mold temperature, or 0.5 seconds in holding time can result in parts that differ by a whole level in size and appearance. The sample stage is the only time to test parameters correctly—because changing parameters is free, but after mass production, the line must be stopped.
Third is sample retention. Each batch keeps one sample and one sample. If problems arise later and no sample is retained, there is no comparison; troubleshooting can only be guessed.
Only when all three steps are done is the sample considered complete; If only one piece is made, it's just "proofing."
7. After the sample: Four things to convert the sample into mass production
Sample passed, and there are still four things to link the sample to mass production:
First, is there an alternative? If this grade is cut off or the price increases, is there a proven second option? The sample stage is the easiest time to ask this question—because the process parameters are still in hand, and trying another batch is the cheapest.
Second, process documentation. Write the sample drying, mold temperature, and injection parameters into the official process card, not just in the engineer's mind. Once the person changes, the parameters are lost.
This point is especially prone to problems in injection molding plants. Sample parameters are often controlled by one or two experienced technicians, who say "add two seconds of pressure" depends on feel, not on record-keeping. When shifts or personnel change, the same batch of material produces different pieces, and the first suspect is the material. But in reality, the material hasn't changed; the person is gone.
Third is packaging and labeling standards. Packaging and labeling for small samples and bulk goods must be consistent to avoid taking the wrong materials on the production line.
Fourth is price and supply confirmation. After the sample is successful, discussing price and annual usage is more confident than discussing it from the start.
Only after these four tasks are done can the small sample truly be converted into mass production. If any one is missing, it will be compensated for later by other means.
Conclusion
Return to the customer at the beginning.
He later developed a habit: after developing a new part, he first explained the working conditions and annual usage to us, then decided which section to start with. Last month, his valve body had already reached the third batch of mass production, with no rework in between.
The two quotations that made him hesitate at the time were: one spread the fixed loss over 25 kilograms, and the other set the minimum order quantity at one ton he couldn't handle—both were correct, but neither suited the project at the time.
Small batch isn't "a bit less," it's "different sequence."
First sample size, then trial production, then volume ramp-up—these three steps save much more rework than material cost saved.
has a project in the making, send me the working conditions, and we'll help you see where to start
