Engineering Insights

Why Your JST Connector Harness Keeps Failing (It's Rarely the Crimper)

The call that starts with "we need it by Friday"

In March 2024, a client called on a Thursday at 4 p.m. Their line was down to a few hundred JST XH 2-pin assemblies, and a shipment was booked for Friday morning. Normal lead time on that build was nine days. We found a shop that could re-run the terminations overnight, paid about $600 in expedite fees on top of a $2,400 base, and delivered by 6 a.m. The alternative was a missed ship date and a customer conversation I don't envy.

Here's the part that still bugs me: the original harness wasn't bad because of the crimper. It was bad because the terminals came from a bin that wasn't the bin on the drawing.

I coordinate rush connector and harness orders for electronics manufacturers. Over the last two years I've handled 200+ expedited builds, including 36-hour turnarounds for contract manufacturers in line-down situations. When I'm triaging a rush order, the first question is always the same: what actually failed? Most of the time, it isn't the thing the customer thinks failed.

First, the problem you think you have

If you've ever built your own JST harness and watched it fail continuity on the bench, you probably blamed one of three things: the crimper, the wire stripper, or your own hands. I get it. That's the surface problem, and it's the one that gets all the attention in forums and comment sections.

The jst xh connector 2 pin is about as common a part as exists in small electronics. It's a 2.5 mm pitch, 2-position wire-to-board connector, typically rated 3 A AC/DC and 250 V, with terminals that accept roughly 30 to 22 AWG wire. Because it's so common, it's also the one people learn on. Which is exactly where the trouble starts.

What's actually going on

Mistake #1: treating the series as interchangeable

JST XH, PH, SH, ZH, GH, and VH all read as "JST connector" from across the bench. They are not interchangeable. XH is 2.5 mm pitch, PH is 2.0 mm, SH is 1.0 mm, VH is 3.96 mm. A PH housing will not seat on an XH header—but the two look close enough in product photos that people order the wrong one, then force it. I've seen a forced PH-on-XH joint pass a quick continuity check and fail a vibration test three weeks later. (Should mention: that one came back as a field return, not a bench failure. Bench checks don't catch everything.)

Mistake #2: the terminals aren't the terminals

This one surprised me. For years I assumed the dominant failure mode was tooling. Turns out a lot of the failures we get handed are terminal-sourcing problems. Genuine JST terminals have a defined crimp-height window and specified plating thickness. "Compatible" terminals from secondary bins can be dimensionally close and metallurgically different. On a 2.5 mm pitch part, "close" is not a word you want in your process notes.

If you're building harnesses in-house, buy terminals from a traceable source and keep the lot number on the traveler. That single habit has saved us more rework than any tool upgrade we've made.

Mistake #3: generic crimpers on fine pitch

I'm not 100% sure where the "any ratcheting crimper works" advice started, but it needs to stop. On XH, the crimp-height tolerance is tight—think a few hundredths of a millimeter. On SH and GH at 1.0 and 1.25 mm, it's tighter still. A generic SN-28B-style tool doesn't hold that window consistently across a production run. A crimper matched to the terminal series does. The difference between the two is usually a few hundred dollars, which is the cheapest insurance in this entire process.

The part nobody scopes: strip length and strand placement

Strip length off by a millimeter, a stray strand outside the crimp barrel, insulation creeping into the conductor crimp—all of these pass a gentle tug and fail a pull test. IPC/WHMA-A-620 sets minimum crimp tensile strength by wire gauge and class. If you're building to a customer drawing, their class governs, not your bench standard. Ask before you assume, because the answer changes how much of your process needs to be controlled.

What it costs when it goes wrong

Bench rework is cheap. Field failure isn't. The math we use internally is simple: a failed XH 2-pin caught at the bench costs a few cents in labor and a few minutes. The same failure at a customer's assembly line costs a line stoppage, a containment report, and a shipment that has to be re-screened. At the field level it's a return, a root-cause report, and someone's credibility. The interesting thing is that the cost of prevention sits almost entirely at the earlier, cheaper stage.

I have mixed feelings about in-house harness building. On one hand, for small runs and quick prototypes, it's genuinely the right call—you control the schedule and you learn your own process. On the other, I've watched teams scale a bench process straight into a production schedule without changing the tooling, and that's where the emergency calls come from. I reconcile it by treating in-house and outsourced as two different processes with two different control plans, not one process at two volumes.

Last year we handled a re-termination where the client had built 800 harnesses in-house over two days. About 60 failed the pull test. The decision was whether to re-crimp in-house or buy pre-crimped assemblies at rush pricing. The upside of staying in-house was roughly $2,000 in material savings. The risk was shipping parts that passed a bench check and failed under load. I kept asking myself whether $2,000 was worth potentially triggering a containment event on a customer line. We went with pre-crimped assemblies and ate the difference.

Even after that call, I second-guessed it. What if the rush vendor's crimp height wasn't as consistent as their sample report suggested? The 48 hours until the pull-test data came back were not relaxed hours. The data came back clean, but I wouldn't describe that week as comfortable.

The fix, briefly

I'm not going to turn this into a step-by-step on how to make JST connectors—there are decent videos for that, and if you're reading this you probably already know the mechanics. The fix is process, not technique:

  • Match the whole system. Terminal, housing, and header from the same series and the same datasheet revision. No mixing, no "they look the same."
  • Buy the tool or buy the assembly. If your annual volume on a given series is small, pre-crimped assemblies are usually cheaper than tooling plus the learning curve. If volume is high, buy the matched crimper and a crimp-height gauge and use both.
  • Pull-test to a standard. Not to "feels solid." IPC/WHMA-A-620 or the customer's drawing, whichever applies.
  • Build a real buffer. The March call happened because a nine-day build had a two-day buffer. That's not a buffer, that's a coin flip.

What changed in five years

What was best practice in 2020 doesn't fully apply in 2025. Five years ago, a lot of this category was 2.54 mm pitch on through-hole boards, where "good enough" crimps survived because the geometry was forgiving. Now the same teams are putting 1.0 mm SH connectors into battery packs and 1.25 mm GH into camera modules, and the tolerance window narrowed without the habits catching up.

The fundamentals haven't changed—a crimp is still a cold weld, and a bad one still fails the same way. But the execution has transformed. Tooling, terminal traceability, and the test standard all matter more than they did, and the teams that updated their process are the ones who stopped calling us on Thursday afternoons.

That's the whole point. Your crimper probably isn't the problem. Your process is—and unlike the crimper, the process is fixable.

Leave a Comment

Your email address will not be published. Required fields are marked