Blog

The Crimp Failures We Blamed on Technique Were Actually a Cable Problem

In March 2022, our crew installed 200 crimped connections on a commercial fit-out just outside Manchester, NH. By the end of the first week, 47 of them had failed pull tests. Not marginal failures—complete pull-outs. The GC flagged it in writing. We spent the next six days redoing work we'd already billed for.

I was the one handling the cable orders on that job. Nine years in, and I'd never seen a failure rate like that. My first instinct was the crimper. Then the technique. Then the connectors themselves.

It took me about six months to figure out the real cause. It wasn't any of those things.

What We Blamed First

If you've ever looked up how to crimp connectors, you've seen the same list I followed back then. Use the right die. Match the connector to the wire gauge. Don't twist the strands. Pull-test every connection before you move on.

We did all of that. I rewrote our SOP in April 2022. Bought a new ratcheting crimper (about $340, which felt like a lot at the time). Retrained the two apprentices. For the next job, we treated every connection like it was going into a hospital.

The failure rate dropped—but not to zero. Maybe one in forty still pulled loose too easily. That's when I stopped blaming the tool and started asking a different question: if our technique is consistent, why is the cable behaving inconsistently?

The Real Cause Was the Cable Itself

I pulled the spec sheets for three lots we'd used in that stretch. Same part number, three different suppliers. The conductor stranding counts matched on paper. But when I sat down with a caliper and counted actual strands on a 10-foot sample from each lot, two of the three were short by two to four strands per conductor.

That doesn't sound like much. On a #12 THHN, it changes how the strands compress inside a crimp barrel. Fewer strands means more void space, which means the crimp has to work harder to achieve the same pull strength—and often it doesn't.

The insulation diameter was also inconsistent. On two of the lots, it was running about 8% over spec. That's enough to throw off the die seat on a hand crimper. The die closes on schedule, but it's not fully seating on the metal. You get a crimp that looks fine and fails under load.

I've got mixed feelings about blaming the supplier here. On one hand, the spec sheet said the right things. On the other, nobody checks strand counts on a paper spec—they trust the lot.

Why Small Orders Get the Bad Batches

Here's the part that took me longest to understand. The two bad lots both came from orders under $600. Our larger orders in the same period were fine.

That's not a coincidence. Large distributors rotate stock and pull from current production runs. Small orders often get filled from whatever's sitting in the back of the warehouse—including returns, job leftovers, and stock that's been on the shelf for 18 months. The bigger the order, the more likely you get fresh material.

This is the thing about being a small contractor that nobody warns you about. You're not just paying a higher unit price. You're also more likely to get the material nobody else wanted.

I get why it happens. Warehouses optimize for throughput, and a $400 order takes the same picking time as a $40,000 order. But it's still a bad way to treat small accounts. Today's $400 order is tomorrow's $40,000 order, and the suppliers who understand that are the ones I still buy from.

What It Actually Cost

Let me add up the damage from that one job, because I want to be honest about the scale:

  • Rework labor: two electricians for six days—about $3,900
  • Replacement cable and connectors: roughly $600
  • Schedule penalty from the GC: $1,200 (they were within their rights)
  • Lost follow-on work: the GC used a different sub on the next two buildings

The direct hit was around $5,700. The lost work is harder to price, but over two buildings it was probably close to $18,000 in margin. All of that traced back to crimp failures that we initially blamed on our own technique.

So glad I pulled those sample strands. If I'd just replaced the crimper again and moved on, we'd still be shipping bad connections and wondering why.

The Choice I Almost Got Wrong

When we reordered cable, I went back and forth between going back to the big regional distributor and switching to a smaller local supplier for two weeks. The big distributor had the reliability track record and the brand name. The local supplier had a 3-day lead time and answered the phone on the second ring.

On paper, the big distributor made sense. My gut said the local shop would be more careful with a small order because they actually wanted the relationship.

I went with the local supplier for a trial 500-foot run, and I'm still using them. It wasn't a completely rational decision—I didn't have data on their failure rates yet. But I'd decided that responsiveness on a small order was a signal worth trusting.

To be fair, the big distributor probably would have fixed the lot issue if I'd pushed. I just didn't want to spend the next three months pushing.

What I Check Now

I'm not going to pretend this is a comprehensive QA protocol. It's five things I do before any cable order goes out, and it's caught problems on 11 orders in the past 18 months:

  1. Ask for a lot number and manufacture date before confirming the order. If the supplier can't provide it, that tells you something.
  2. Order a 20-foot sample first on any new supplier, even if it feels silly. Count strands on a #12 conductor with a cheap pair of calipers. It takes five minutes.
  3. Measure insulation diameter against the spec sheet on that sample. If it's over by more than 3%, call it out before the full order ships.
  4. Pull-test on a scrap length with the exact connector and die you plan to use on the job. Not the spec sheet's recommended setup—yours.
  5. Note the lot number on the job paperwork. If a failure pattern shows up later, you know which batch to pull.

None of this requires special equipment. The caliper was $22. The sample lengths have cost me under $200 total in three years. It's a much better deal than the $5,700 I spent in 2022 learning the same lesson.

The Part I'd Push Back On

If you're a small contractor reading this, I know the temptation. You feel like you should just be grateful a supplier will talk to you at all. I've been there. My first year, I was ordering $180 worth of SOOW cord and getting put on hold for 20 minutes.

But small doesn't mean unimportant. It means potential. The suppliers who treat a $400 order with the same care as a $40,000 order are the ones worth keeping, because the failure modes on the small stuff are the same failure modes that will wreck you on the big stuff later.

And the cable doesn't know how big your purchase order was. It only knows whether it was made right.

Rowan Whitaker
Rowan Whitaker

Rowan Whitaker is a fiber-optic systems analyst covering SFP and QSFP transceivers, OLT, ONT, ONU, passive splitters, optical amplifiers, and CWDM and DWDM platforms. He applies IEC 61280-4-2 and IEC 61300 methods while examining insertion loss, return loss, optical power budget, bit error rate, wavelength drift, dispersion, channel spacing, and transmission reach. His guides help carriers, data-center teams, system integrators, and sourcing specialists compare capacity, interoperability, link margin, serviceability, and migration paths.

Leave a Reply

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