Latest / Hardware Startups with Fexingo: Devices, Robotics, and Manufacturing Tech Companies / How Hardware Startups Use Wire Harness Assembly to Avoid Field Failures
Transcript
- Lucas: You know the last time a device in your house just stopped working — not dramatic, no smoke, no error code — and you tossed it? I want to talk about what probably caused it. Luna: I am guessing you are about to tell me it wasn't the chip or the software. Lucas: Exactly. In a teardown analysis I came across recently, a smart thermostat that failed after about six months — completely dead, no response — the root cause was a single crimp terminal in the internal wire harness. The wire had pulled loose from the connector because the crimp height was off by something like 0.1 millimeters. Luna: So the whole device was bricked by a connection that probably cost less than a penny to make. Lucas: Right. And that is the story for a surprisingly high percentage of field failures in hardware — especially for startups shipping their first few thousand units. The fancy PCB, the custom enclosure, the firmware — all fine. But the wire harness, which looks like the simplest part of the assembly, is where the defects hide. Luna: Why do you think harnesses get overlooked? I remember in my own prototyping days, we treated wiring almost as an afterthought — just grab some off-the-shelf cables and crimp them. Lucas: Because it looks easy. You cut a wire, strip the end, crimp a terminal, insert it into a housing. It feels like something you learned in middle school shop class. But in production, those manual steps introduce variability — crimp height, pull strength, insulation crimp width, terminal alignment. And because harnesses are typically the last thing assembled, they get rushed. Luna: So what is the single most important test a hardware startup should add to their line to catch harness defects before the product ships? Lucas: A pull test on every crimp. Not a sample — every single one. There are handheld pull-test gauges that cost around three to five hundred dollars. You set it to a force threshold based on your wire gauge — say, 10 pounds for 22 AWG wire. You pull each terminal after crimping. If it slips out before reaching the threshold, you know your crimp tool needs adjustment or your operator technique is off. Luna: But that sounds like it adds time to the line. How many seconds per crimp? Lucas: About two to three seconds per terminal. If your harness has ten wires, that is thirty seconds of cycle time. Compare that to the cost of a field failure — maybe fifty dollars in warranty handling, shipping, and replacement unit. If you have a one percent defect rate on a product that sells ten thousand units, that is five thousand dollars in field failure costs. The pull-test gauge pays for itself in the first batch. Luna: Okay, but what about visual inspection? Can you catch a bad crimp just by looking at it? Lucas: Sometimes. A good operator can spot an exposed conductor or a damaged insulation barrel. But you cannot see the internal deformation of the wire strands. A crimp can look perfect on the outside and still have insufficient compression. That is why pull testing is the gold standard — it measures the actual mechanical bond. Some shops also do a micro-section test on a sample every shift: they cut the crimp in half and look at it under a microscope to check the grain structure. Luna: That seems like a lot for a startup. Is there a lighter-weight version? Lucas: Absolutely. If you are building only a hundred units in a prototype run, you can do a 100 percent pull test manually. That is your baseline. As you scale, you can switch to a sampling plan — pull test every tenth crimp, or every crimp from a new operator. The key is having the data. You want to track pull force over time and watch for drift. If the average pull force drops from 12 pounds to 10 pounds over a week, something changed — maybe the crimp die is wearing out. Luna: I want to go back to that thermostat teardown. Was the crimp height out of spec because the tool was not calibrated, or was it operator error? Lucas: The report I read said the crimp die was set for a slightly larger wire gauge. Someone used a 20 AWG setting on 22 AWG wire. That produced a crimp that was too loose. It passed an initial continuity check because the metal was touching, but under thermal cycling — the thermostat heats up and cools down — the connection eventually failed. That is the insidious part: some harness defects only show up after temperature or vibration. Luna: So continuity testing alone is not enough. You need mechanical testing. Lucas: Exactly. Continuity tells you there is an electrical path right now. It does not tell you whether that path will survive the ride to the customer or a summer in an attic. A pull test gives you the mechanical integrity. And for startups, there is another cheap win: a connector retention test. After you mate two connectors, you gently tug on the wire bundle to make sure the locking tab is fully engaged. That catches a surprisingly common defect where the connector is not fully seated. Luna: I can see how that would happen on a rushed line. So let's say I am a startup founder listening right now. What is the one thing I should do tomorrow to improve my harness quality? Lucas: Buy a crimp pull tester and a connector retention gauge — total cost under a thousand dollars. Train every assembler on the correct crimp height for your wire gauge. Then implement a first-article inspection: the first harness of each shift gets a full pull test on every terminal. Document the results. That alone will catch ninety percent of harness defects before they leave your building. Luna: That is really practical. And it is something that doesn't require a PhD in mechanical engineering. Lucas: Right. And this is exactly the kind of thing we try to cover on this show — the operational detail that makes the difference between a product that works in the field and one that generates support tickets. If today's episode was useful to you, the way we keep these conversations ad-free and focused is through listener support. You can find us at buy me a coffee dot com slash fexingo. That is the simplest way to help us keep digging into topics like this. Luna: Yeah, and honestly, every contribution — even a small one — goes directly to keeping the show independent. We don't have sponsors, so it really is listener powered. Lucas: Exactly. And on that note, let's talk about another common harness failure mode: the strain relief. A lot of startups use off-the-shelf cable glands or simple zip ties as strain relief. But for a device that gets handled or moved, that is often not enough. Luna: I have seen products where the wire bundle exits the enclosure and the only strain relief is a rubber grommet. That seems like a recipe for broken wires right at the exit point. Lucas: It is. The wire flexes at that sharp edge every time the device is picked up. Eventually the copper work-hardens and fractures. A proper strain relief — like a molded boot or a clamp that grips the cable jacket — can extend the flex life by a factor of ten. And it costs maybe an extra dollar in materials. Luna: So the message is: don't treat the wire harness as a commodity. It is a mechanical system that needs just as much design attention as the PCB. Lucas: Exactly. And one more thing: when you are sourcing pre-made harnesses from a contract manufacturer, ask for their crimp pull test data. A reputable harness shop will have it. If they don't, that is a red flag. You want to see a histogram of pull forces across a production run, not just a single number. Luna: That is a great procurement tip. I am going to add that to my list of questions for suppliers. Lucas: And if you are doing your own in-house harness assembly, run a process capability study. Crimp twenty terminals with your tool, measure the pull force on each, calculate the mean and standard deviation. If your process capability index — Cpk — is below 1.33, you need to tighten the process. That is a standard metric in automotive harness manufacturing, and it is a good benchmark for any hardware startup. Luna: So we have gone from a dead thermostat to Cpk. That is a pretty rich vein for something that looks like a simple wire. Lucas: It always is. The boring parts of hardware are where the failures live. Get those right, and the exciting parts — the sensors, the software, the user experience — get to shine. Get them wrong, and the customer never even gets to see the good stuff. Luna: Alright, I am going to check my own prototype's harness tomorrow. Thanks, Lucas. Lucas: Always a pleasure. That is it for this episode of Hardware Startups. We'll be back next time with another deep dive into the details that matter.