Why Working Product Components Fail After Final Assembly [Podcast Ep. 335]

A product’s electronics work correctly on the test bench. Its firmware performs as expected. The mechanical parts fit, and each subsystem appears to have passed its tests.

Then everything is installed inside the final enclosure, and new problems suddenly emerge.

This is where product integration becomes critical. Bringing the electronics, firmware, power system, wiring, mechanical components, and enclosure together creates a very different operating environment. Heat accumulates, airflow changes, components may interfere with one another, clearances become tighter, and access for debugging can disappear.

In this episode of China Manufacturing Decoded, Adrian is joined by Paul Adams from Agilian Technology to explain why individually successful subsystems can fail once they are combined into a complete product.

They discuss the importance of staged integration, retaining access to test points, testing thermal and EMC risks early, and validating the product as its eventual users will experience it. Paul also shares a practical five-step framework for moving from successful bench tests to a safe, reliable, and genuinely usable finished product.

Listen here

Listen to the episode or watch on YouTube

Podcast sections

  • 00:21 – When Individually Working Parts Fail Together
  • 02:28 – Why Everything Works During Bench Testing
  • 07:40 – What Changes When Integration Begins
  • 08:39 – Designing the Product So It Can Be Debugged
  • 10:16 – Integrating One Subsystem at a Time
  • 13:45 – A Technically Working Product Can Still Fail
  • 15:00 – The Hidden Impact of Heat Buildup
  • 20:19 – The First Complete Build Is Only a Mini Milestone
  • 21:39 – Solving an Enclosure Airflow Problem
  • 23:53 – Paul’s Product-Integration Playbook
  • 24:07 – Step 1: Integrate in Stages
  • 24:52 – Step 2: Test Important Risks Early
  • 26:16 – Step 3: Keep Debugging Access Available
  • 27:18 – Step 4: Test the Way the Customer Will Use It
  • 28:20 – Step 5: Treat the Enclosure as Part of the System
  • 29:29 – Final Lessons From the Integration Process

Further content

🎧 Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are on the way, so remember to rate us and subscribe! You can find us on:

Why Version 1 Shouldn’t Be Perfect (And What to Do After You Launch) [Podcast]

In episode 325 of China Manufacturing Decoded, Adrian and Renaud break down a powerful idea from Tony Fadell: “Builders build, ship, then solve what breaks.”

They explore what really happens after a product hits the market, and why chasing perfection before launch can actually kill your chances of success.

You’ll learn:

  • Why over-engineering delays launches (and increases risk)
  • How Version 1 should really be defined: simple, lovable, and complete
  • What real-world users reveal that prototypes never can
  • How to collect meaningful feedback without damaging your reputation
  • Why early adopters are critical for innovative products
  • How smart teams build Version 2 while Version 1 is still launching

Developing a new product in 2026? You’ll have food for thought from this one!

 

Listen here

Listen to the episode or watch on YouTube

Episode Sections:

  • 00:00:13.326 — Episode overview
  • 00:00:37.886 — Tony Fadell’s quote
  • 00:01:37.906 — Why perfection is a trap
  • 00:04:28.242 — Engineering vs speed trade-off
  • 00:06:30.021 — Launch early vs over-engineering
  • 00:07:46.050 — De-risking with Version 1
  • 00:10:30.755 — “Simple, lovable, complete”
  • 00:13:43.044 — Launch isn’t the finish line
  • 00:15:04.650 — Real-world user behaviour
  • 00:17:06.894 — Nest example (unexpected insights)
  • 00:19:36.494 — Managing reviews & early releases
  • 00:21:27.921 — Choosing the right early users
  • 00:24:02.240 — Misinterpreting “ship early”
  • 00:25:47.970 — Lessons from product reliability
  • 00:26:56.210 — Why post-launch work matters
  • 00:28:28.883 — Continuous product development
  • 00:30:25.311 — Key takeaways

Further content

🎧 Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are on the way, so remember to rate us and subscribe! You can find us on:

Are You Building What People Will Actually Buy? How to Validate Demand, Customers, and Features [Podcast]

Many hardware teams don’t fail because their engineering is weak. They fail because they misread the market.
It’s easy to fall in love with a clever solution, a beautiful prototype, or a smart technical idea. The hard part is proving that real customers actually care enough to pay for it.
Adrian speaks with Renaud Anjoran about how product teams can answer three critical questions before investing heavily in tooling, molds, and production:
1. Is there real demand?
2. Who is the target customer?
3. What features do customers truly want?
Here’s what teams need to get right.

 

Listen here

Listen to the episode or watch on YouTube

Episode Sections:

  • 00:00 – Intro: The big question — are you building what people will actually buy?
  • 01:04 – Is there real demand? (customer discovery first)
  • 09:40 – Who is the target customer? (segmentation beats ‘everyone’)
  • 15:35 – What features do customers actually want? (listen for patterns)
  • 24:30 – Three lessons before you spend on tooling.
  • 25:25 – Close & resources.

Further reading

🎧 Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are on the way, so remember to rate us and subscribe! You can find us on:

Why Your New Product Development (NPD) Partner Really Matters [Podcast]

Bringing a product from idea to mass production is never a straight line. In this episode, Adrian is joined by Paul Adams, who leads the New Product Development (NPD) team at Agilian Technology, to explain why the strength of your NPD partner can make or break your product.

They break down the full NPI journey, explain what a good NPD team actually does day-to-day, and highlight the risks importers face when working with “order-taker” factories instead of true development partners.

Listen here

Listen to the episode or watch on YouTube

Episode Sections:

  • 00:00 – Introduction & episode context
    Why NPD partnerships matter when going from idea to mass production
  • 01:55 – Overview of the NPI / NPD journey
    Why new product development is a process, not a single milestone
  • 02:36 – The six NPI phases explained
    Feasibility → Prototype → Tooling → Validation → Pre-production → Mass production
  • 05:00 – Why pre-production runs are critical
    Real example: catching a potential 30% failure rate before mass production
  • 07:30 – What an NPD team actually does
    Acting as both the customer’s voice and the company’s representative
  • 11:10 – Managing scope, budget, and expectations
    Why scope creep quietly kills timelines, cost, and quality
  • 14:10 – Transparency as a core NPD responsibility
    Why “telling customers what they want to hear” creates long-term risk
  • 16:35 – Embedding risk mitigation into every phase
    Living risk registers, phase gates, and cross-functional reviews
  • 21:00 – Risk goes beyond engineering
    Budget limits, internal constraints, and customer readiness
  • 24:00 – Benefits of a strong NPD partner
    Faster time-to-market, built-in quality, and reliability by design
  • 27:05 – Intellectual property protection and trust
    Why IP protection is foundational to long-term partnerships
  • 30:10 – Order-takers vs true manufacturing partners
    What importers should look for when choosing a contract manufacturer
  • 31:25 – Closing remarks & where to learn more

Further reading

🎧 Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are on the way, so remember to rate us and subscribe! You can find us on:

Can You Afford to Manufacture Your Idea? Budget Truths from Idea to Mass Production [Podcast]

In episode 305 of China Manufacturing Decoded, Adrian and Paul dig into a question that most new hardware founders don’t really want to face: can you actually afford to take your product from idea to mass production? They unpack the most dangerous cost myths (like “MOQ × unit price is my total cost” or “we’ll fix reliability later”), then walk through Sofeast/Agilian’s structured NPI phases to show where the money really goes, from early engineering work and tooling through pilot runs and ramp-up.
Along the way, they share hard-won lessons from projects that ran out of cash at the worst possible time, and offer simple “readiness checks” so you can judge whether you’re genuinely prepared to launch, or need to rethink your plan before burning money.

 

Listen here

Listen to the episode or watch on YouTube

 

Episode Sections:

  • 00:00 – Intro & who this episode is for
  • 07:02 – Mythbusting: YouTube & “$10k product launch” myths
  • 12:13 – The Sofeast/Agilian 6-phase NPI process
  • 21:18 – How your budget is split across the phases
  • 29:00 – What to expect in each phase & readiness checks
  • 37:31 – Tooling, NRE, and why half a tooling budget is worse than none
  • 43:42 – Budgeting properly and adding contingency
  • 45:21 – Call to action & how Sofeast/Agilian can help

 

Further reading

 

🎧 Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are on the way, so remember to rate us and subscribe! You can find us on:

Before You Hire Freelance Hardware Engineers: 4 Safeguards

A common pattern we see: a startup hires several low-hourly-rate freelance hardware engineers (mechanical, electronics, and/or firmware engineers) from different countries like Jordan, Pakistan, Vietnam, etc. Work seems to progress…until it doesn’t. Typical failure modes include:

  • Designs that can’t be manufactured, so large chunks must be redone.
  • Partial hand-offs where someone claims “the job is done, it is ready for a factory,” but it is only 30% done.
  • Poor cross-discipline collaboration that turns the project into a mess.

The fix is not “work harder,” it’s “set the right safeguards” to prevent this from happening in the first place. Here are the 4 safeguards that we suggest…

 

1. Qualify the freelance hardware engineer’s prior experience in designing manufacturable products

Don’t select freelance hardware engineers purely on the hourly rate. Ask candidates to show prior Design For Manufacturing/Design For Assembly examples, for example, such as the improvements shown here: DFM for plastic injection molding or here: DFM for PCBA. Let them explain what they changed to hit yield, cost, and cycle-time targets. If they can’t describe how the manufacturing process will work in detail, that might be a red flag.

If you pick the lowest-cost designers, they might do work that later has to be scrapped entirely and restarted from scratch. For example, we have seen a number of enclosures that can be 3D printed but can’t be mass-produced for a reasonable cost.

For deeper context, see Agilian’s practical DFM tips and comparisons.

2. Ensure you get the deliverables a manufacturer will need

Freelancers often “finish” without producing factory-ready outputs. The problem is, many manufacturers just want to produce and don’t have the will or the capability to do a nice job fine-tuning your product.

Up front, list exactly what you expect to receive: an engineering BOM (ideally with second-sourced options for critical components), documented evidence about pre-certifications of critical-to-compliance components, 3D drawings, tolerance-controlled 2D drawings with the usual mentions (CTQ points, CMF details), functional test reports, function test jigs, the QC checklist, and so on. Use a stage-gate mindset and review deliverables before paying milestones.

Sofeast’s NPI resources and deliverables review service helps you confirm that you receive what the manufacturer will need.

You may also find this video, Typical Deliverables You Should Expect From Product Designers During Product Development, helpful:

Typical deliverables you’ll want

Mechanical Design

  • 3D CAD files: For enclosure, housing, and mechanical components.
    • Typical format: .STEP
  • 2D drawings: For manufacturing and assembly.
    • Typical format: .PDF

PCB Design (Electronic Hardware)

  • Schematic files
    • Native CAD files (e.g., .SCHDOC for Altium, .DSN for OrCAD) + .PDF
  • PCB layout files
    • Formats: Native CAD files (e.g., .PCBDOC, .BRD) + .PDF
  • Gerber files: For PCB fabrication.
    • Typical format: .GBR
  • Bill of Materials: List of electronic components.
    • Format: .XLSX
  • Pick and Place files: For automated assembly.
    • Formats: .CSV, .TXT

Firmware / Software

  • Source code: For microcontroller or embedded system.
    • Typical format: .C
  • Compiled binaries: For flashing onto the device.
    • Typical format: .HEX
  • Documentation: API references, setup guides.
    • Typical format: .DOCX or written in markdown in Github

Packaging Design

  • Artwork files: For labels, box design, inserts.
    • Typical formats: .AI (Adobe Illustrator), .PSD, .PDF
  • Die-line files: For packaging structure.
    • Typical formats: .PDF, .DXF, .AI

3. Ensure you have access to native files and IP

Make sure your contract states you will get all native design files (mechanical drawings in .step format, PCB schematic + Gerber files, firmware source code & documentation if needed, and so on) and that they’re handed over at each milestone for review, not just at the end.

Avoid situations where R&D is done “in-house” by a party that won’t release files; these could then be used to produce your products without permission or, more commonly, at that party’s selected factory only and at the price they decide. Pair that with the right IP protections and, if needed, custody for critical tooling/data.

Go deeper into IP protection by reading IP Protection in China when Developing Your New Product [Importer’s Guide].

4. If the product is not simple & straightforward, appoint a tech lead

From the start of the project, you may hear the electronic designer fight with the mechanical engineer over the space allotted to the PCB. That’s just an example. An experienced tech lead can make informed decisions and allow you to move forward with confidence.

The tech lead can also review the design work and point out issues or risks. That’s priceless. Bear in mind that design work is often completely restarted from scratch because of seemingly small oversights. Picking the wrong microcontroller chip for a given application may result in wasting 50,000 USD and 3 months of work.

In addition, electro-mechanical products need one accountable integrator (yours or external) who will integrate all elements of the design. To simplify, the firmware must work well with the electronics parts and within the mechanical elements. You want to prevent the “it works on my bench” discussions among the engineers, which usually lead to a dead-end.

That integration work can be done by the manufacturer themselves, if they have an R&D team in-house that can handle this.

Bonus: Final checks before engaging factories

Do a quick review: are designs mature, risks mapped, tests planned, and pilot run criteria clear? If not, keep developing before you approach manufacturers; this shortens time-to-stable-mass-production and avoids expensive backtracking.

Fitness Wearable Development Challenges: What You Need to Know Before Manufacturing in China

Smart wearables that collect vital data, such as fitness trackers, smartwatches, and medical-grade health monitors, are becoming more and more popular in the marketplace. They are, for the vast majority, made in China. As they start to be produced in very large numbers, they are following the model of other mass-market electronic products, and buyers need to be aware of the implications of that model before starting their fitness wearable development project.

 

The General Model of R&D vs. Assembly in China

Not many people know how wearable technology manufacturing (and other electronics such as smartphones and laptops) is organized in China.

Some companies do the R&D, designing the PCB, writing the firmware, and selecting components, and then sell the whole solution as “kits” to other companies. Let’s call them the R&D companies.

Those other companies buy the kits, then assemble and test the final products. We’ll call them the assemblers.

Assemblers do not:

  • Have access to the source code.
  • Decide what key components to use.
  • Control the sensor integration.

All of these decisions are made by the R&D company.

So, when you get in touch with a manufacturer of a smartwatch, chances are, all they can do is offer relatively light customization. The software may have a slightly different feel. Maybe the casing will have a different shape. But it usually doesn’t go much deeper than that.

What Does That Mean When It Comes to Customization Options?

When made in China outside of big brands like Apple or Huawei, these products tend to be low-cost with intense competition between assemblers, and very little differentiation.

Any of the models will often look like many others that are made from the same kit. A kit typically includes:

  • The electronic design and components.
  • The firmware is loaded on the motherboard.
  • Key peripherals like display, memory, and ports.

All these elements have been tested to work together. Customizations are possible, but they require additional payment, which goes mostly to the R&D company. Typically, you pay the assembler, who then pays the R&D company.

A real-world customization example

A German company orders a smartwatch with slight customization. They pay an assembler in Shenzhen who is using a kit from a Shenzhen R&D company. The assembler doesn’t have the source code or own the IP, and simply passes on requests (and payments) to the R&D company. In the end, the German company cannot get the source code, bill of materials, or key technical data, even if they forced the assembler to share everything they had.

The Extra Implications for Fitness/Medical Wearables That Collect Vitals Data

If you are developing a medical device or fitness wearable and want to build your own intellectual property, you need raw biometric data from the device’s sensors.

Look at the Apple Watch health functions:

  • Blood oxygen
  • ECG
  • High/low heart rate notifications
  • Irregular rhythm notifications
  • Low cardio fitness notifications
  • Sleep apnea notifications

All these rely on raw sensor data, which is then processed through a proprietary algorithm. To compete, you must create your own formulas, tuned for your niche application.

Without raw data, you can’t…

  • You can’t develop your own analytics.
  • You can’t run population-level data analysis.
  • You won’t have a strong IP to attract investors.
  • You may not be able to meet medical device certification requirements.

As noted earlier, assemblers do not have the raw data because they don’t have access to the source code.

Is It Impossible to Have Access to the Raw Data?

In our experience, only a small minority of companies will provide raw data.

Why?

  • IP protection – Raw data access could lead to reverse engineering of their algorithms.
  • Performance concerns – If raw data is compared to certified medical devices or even an Apple Watch, discrepancies might be exposed.
  • Business model – R&D companies want to keep customers dependent on their integrated solutions.

By only providing processed or calculated values, they avoid scrutiny.

What This Means for Your Medical or Fitness Wearable Development Strategy

If you want to own your product and work based on an existing device, here’s what I recommend:

  • Find and work directly with the R&D company.
  • Negotiate raw data access upfront.
  • Verify they can actually provide it (ask for sample data or API documentation).
  • Lock IP rights into your contract.

The problem is that it’s quite difficult and it often is impossible.

If you have time & funding, and if you want a device unique to your company, there is an alternative. Start the product design from scratch with your own team or a “work for hire” design firm. That’s how you have full flexibility, provided your design architecture and your development contracts allow for it.

For a medical device in the EU or the USA, these steps are essential for achieving regulatory compliance.

You may also like…

  • OEM, ODM, Contract Manufacturers: Which Chinese Supplier to Choose?
  • EU MDR Compliance When Developing a New Medical Device in China: High-Level Steps

Getting It Right from the Start: Essential Guide to Working with Manufacturers [Podcast]

Are you gearing up to manufacture a new electronic / hard good product? In this episode, Renaud and Adrian break down exactly what you need to prepare before starting with a manufacturer. These insights could save you time, money, and major headaches.

 

Listen to the episode here

Listen: to the podcast episode

Watch: on YouTube

 

Episode sections

  • 00:00: Welcome and Introduction
  • 01:13: Sourcing Existing Products: Know What You’re Buying
  • 03:26: Transferring Production: Matching Capabilities
  • 06:12: Developing a New Product: What to Avoid
  • 10:04: Design First, Then Involve the Factory
  • 13:06: Involving the Manufacturer in Development (Correctly)
  • 16:07: Avoiding Design Integration Pitfalls
  • 20:32: NPI: New Product Introduction Essentials
  • 24:05: Small Factories & Unstructured NPI: A Hidden Risk
  • 27:01: Wrapping Up: Do the Work Now to Save Later

 

Related content…

 

Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are coming, so remember to rate us and subscribe! You can find us on:

Freedom to Operate Analysis: Reducing Patent Infringement Risk

Imagine dedicating months, even years, to developing a groundbreaking product, only to discover it infringes on an existing patent, preventing its launch. The potential legal and financial repercussions could be devastating. This article explains why a Freedom to Operate (FTO) analysis—a risk management tool that ensures your product can be commercially made and sold in your target markets without infringing on existing intellectual property (IP) rights – is so important.

Continue reading “Freedom to Operate Analysis: Reducing Patent Infringement Risk”

Angell e-Bikes Rise and Fall: A Cautionary Design and Manufacturing Tale [Podcast]

Adrian and Renaud are looking into the Angell e-bike brand today. This once-promising French e-bike startup soared in popularity and even had a partnership with the MINI car brand, but is now facing bankruptcy. What went wrong? Was it poor manufacturing choices, a flawed design, or something else?
We unravel the complexities of this case, exploring the partnerships, design decisions, and supply chain dynamics that led to this downfall.
If you’re in product development or manufacturing, this episode offers essential lessons on choosing the right partners and ensuring you go through a solid product validation process.

 

Listen to the episode here

Listen: to the podcast episode

Watch: on YouTube

 

Episode sections

  • 00:00 – Introduction
  • 01:52 – Changes in Shipping Small Consignments to the USA
  • 09:09 – The Background of Angell
  • 12:30 – Their Business Model and Partnerships
  • 16:10 – Manufacturing and Design Flaws
  • 23:52 – Costly Recall and Bankruptcy Crisis
  • 27:28 – Blame Game and Legal Battles
  • 33:44 – Key Takeaways: Lessons for Entrepreneurs and Manufacturers
  • 37:38 – Conclusion: A Harsh Lesson for Startups

 

Related content…

 

Listen, rate, & subscribe to the ‘China Manufacturing Decoded’ podcast on your favorite provider

More episodes are coming, so remember to rate us and subscribe! You can find us on:

*Hero image of Angell MINI e-bike courtesy of autoevolution.com