Fupower Custom Auto Parts Manufacturer - Best Racing parts manufactuer in China
In Vlog 05, Andy raised the concern that sits behind almost every custom project: my customers are picky. If the carbon weave comes out distorted, or the fitment against the titanium pipe is loose, the part comes back.
This is not a hypothetical worry. Returns are the most expensive thing in this business. Freight in both directions, a replacement part, and the customer's confidence — which is far harder to rebuild than a part is to remake.
Frank's answer did not start with quality control. It started with a confidentiality agreement. That sequence matters, and it is what this article is about.
"We can sign NDA with you to protect your design."
Most suppliers treat a confidentiality agreement as paperwork a buyer pushes for. In a custom manufacturing relationship it is something else entirely: it is what makes the investment rational.
Think about what a design actually costs a brand. A bespoke carbon intake, tooled and validated for one chassis, represents design hours, prototype cycles, tooling cost and testing. If that design is not protected, the brand has funded the development of a product that anyone can order from the same factory next season. Nobody spends money on development under those conditions — which is why factories with weak design protection attract only the cheapest, least ambitious projects. That is a losing loop for both sides.
An NDA is the mechanism that breaks the loop. It lets a buyer spend on development knowing the result stays theirs.
"If you have your engineer, your engineer can design the styles, then we just make your own products for yourself."
Frank's answer describes two clearly different working relationships, and it is worth being explicit about which is which, because they involve different responsibilities.
Model one: you bring the design, we manufacture it. You have an engineer or a designer on staff. They own the geometry, the styling and the specification. We tool it, validate it and produce it to your drawing, at your tolerance. The intellectual property begins with you and stays with you.
Model two: you bring the idea, we develop it with you. You know the market and what it needs, but you do not have engineering on staff. You describe the product and how you intend to sell it, and we work the development alongside you — concept, modeling, prototype, sample approval, production.
Both are legitimate ways to work. What they have in common is that neither requires you to hand over your commercial thinking to someone who might turn it into a catalogue item. We ask about your idea and your selling plan because we need them to engineer the part correctly — not because they become ours.
Frank's offer was to sign an agreement, not to promise discretion verbally. That distinction matters, because a confidentiality arrangement is only as good as what it names.
Before you send anything, make sure the agreement addresses four things.
The design material itself. Drawings, 3D files, CAD models, ply schedules, material specifications and dimensional data. This is the core of the agreement, and it should name the file types rather than describing the design in general terms.
Tooling and molds. Who owns the tooling made from your design, whether it can be used for any other order, and what happens to it if the relationship ends. Tooling is often the most valuable physical asset in a custom project, and it is the thing most often left unaddressed.
Commercial information. Your pricing, your target market, your customer list and your launch timing. Geometry can be copied; knowing where you sell and at what price is what turns a copy into a competing product.
Duration, and the end of the relationship. How long confidentiality lasts, and what happens to the material afterward — returned, destroyed, or retained under continuing obligations.
We can work from your agreement or ours. What we will not do is treat confidentiality as something that only exists while a purchase order happens to be open.
Frank put it directly: "we not open to clients... this is our working solution in the past 17 years."
In practice, that describes a factory that does not display one customer's work to another. Your part is not a showroom piece for the next buyer through the door. Your drawing is not a reference file pulled up during a competitor's enquiry. Your tooling is not quietly run for somebody else's order.
That is an operational discipline, not a marketing line. It means controlled access to design files, a production floor that is not a public gallery, and a sales process built around a real conversation instead of a portfolio.
For a brand, the practical result is simple: the part in your catalogue is genuinely, commercially yours to sell.
Now the second half of Andy's question. A picky customer notices two things before anything else — a weave that has been pulled out of shape, and a joint that does not sit tight.
Carbon weave does not distort by accident. It distorts when fabric is asked to do something it was not cut to do.
Carbon tow is woven on a grid. When a ply is laid into a compound-curved section of a mold, the fabric either conforms or it fights. If the ply has not been cut to the right shape, someone has to stretch it or bundle it to make it reach the edge — and every bit of that manipulation appears in the finished surface as a pattern that drifts. The same thing happens when a ply is tacked down unevenly, when the vacuum bag pulls before the fabric has properly relaxed, or when one section is allowed to draw more resin than the rest.
What we control for it: plies cut to shape so the weave lands where it is meant to land; orientation marked on the template and respected at layup; fabric allowed to settle before the bag goes down; and inspection under raking light, which is the only reliable way to see distortion — because distortion is nearly invisible under flat, direct lighting.
The joint where a carbon part meets a titanium pipe is where two difficult materials have to agree.
They behave differently under heat. A carbon laminate barely changes dimension as it warms; titanium expands appreciably more. That means a joint that is snug at room temperature is a different joint at operating temperature. A joint designed without accounting for that will either loosen when hot or load the carbon part when cold — and both failures look like a manufacturing defect to the customer, even though neither one is.
Then there is alignment. A fitting is not one dimension; it is a relationship between bore, wall centerline, flange face and the direction the pipe runs. Measuring a part and measuring the interface it must meet are two different jobs. Getting the first right while ignoring the second is exactly how you end up with a component that measures perfectly on a bench and still will not seal in the car.
What we control for it: mating interfaces measured where they will actually meet rather than in isolation; jigs and fixtures that hold geometry constant across a production run instead of relying on each operator's eye; prototype verification before any metal is cut; and a fit tolerance agreed with you up front, so both sides are checking against the same number.
Frank's reassurance to Andy — that Fupower's clients have reported good results over many years — is worth less to you than your own inspection. So here is what to check on the sample, before you approve it for production.
Weave under raking light. Turn the part so light skims across the surface instead of hitting it straight on. Distortion, drifting pattern and resin-rich patches all become visible at a low angle, and disappear under flat, direct light.
Pattern alignment against the design intent. If the weave was meant to run on the centerline, check the centerline. If a colored aramid band was meant to sit at a specific width, measure it rather than eyeballing it.
Edge quality. The trim line on a hybrid laminate is where process control shows itself. A clean, even cut with no fuzz and no torn tow tells you the part was cut with the right tool at the right stage.
Joint concentricity and fit against the mating part. Fit the sample against the actual pipe or manifold it will meet — not a spare — then check it again after the part has been through a heat cycle.
Repeatability. Where the project allows, ask for two samples rather than one. A single good sample proves a part can be made; two matching samples prove a process can be repeated, and it is the process you are buying.
That last point is the whole reason prototype verification exists. Approving one sample and assuming consistency is how a good prototype becomes a bad production run.
It is tempting to read Frank's answer as two separate policies — a confidentiality rule and a quality process. They are one thing.
A brand only invests in the harder, slower, more expensive version of a part when that part is defensible in its market. A brand only asks for a tighter tolerance when that tolerance is a competitive advantage rather than something a competitor can read off a photograph.
Confidentiality is what makes quality worth paying for. That is why the NDA came first.
To move a confidential project forward, send us:
Confidentiality, tolerance and market plan are all first-conversation items. Bring all three and the project starts on the right footing, instead of arriving there by correction.
Andy's worry was returns, so it is worth saying what happens when one occurs.
The first step is diagnosis, not replacement. A distorted weave and a loose joint have entirely different root causes, and treating them the same way guarantees the next shipment fails the same way. Send photographs of the failure as installed, the chassis details, and the dimension that is out of tolerance.
From there the real question is whether the fault sits in the design, in the tooling, or in that individual part — because those three answers lead to three different fixes. Reproducing the fault deliberately, rather than simply remaking the part, is what stops it coming back a third time.
🔒 NDA on Request — we can sign with you, or work from your own agreement, before any file changes hands
🖐️ Two Working Models — bring a finished design and we manufacture it, or bring an idea and we develop it with you
🙈 Your Product Is Not Our Showroom — we do not display, share or reference one client's design to another
🧵 Weave Distortion Control — plies cut to shape, orientation marked and respected at layup, inspected under raking light rather than flat light
📐 Fit Tolerances Agreed Up Front — one number both sides measure against, instead of an argument after delivery
🔥 Titanium-to-Carbon Joints Engineered for Heat — carbon barely moves dimensionally with temperature, titanium does, and the joint has to account for it
🎯 Mating Interfaces Measured Where They Meet — a part that measures to spec and still will not seal is a measurement failure, not a manufacturing one
🏭 Jigs and Fixtures for Repeatability — geometry held constant across a production run, not left to any individual operator's eye
🔬 Prototype Verification Before Metal Is Cut — tolerances proven on a real part rather than assumed from a drawing
🤝 Your Design Stays Your Design — your geometry, your styling, your specification, your IP
⏳ 17 Years of This Working Method — the same operating model, refined across years of custom production
📦 Built for Brands That Resell — parts and processes that survive a picky customer's inspection, not just a shipping notice