← Back to Blog Tuesday 3rd of November 2026

HARTING Connectors and Ethernet Cable: Which of These Three Situations Are You In?

I get asked for "the right connector" about three times a week, and I've stopped answering with a part number. That's not me being evasive. It's because I spent my first two years treating connectors like fasteners — look up the thread, check the current rating, done — and that assumption cost us somewhere around $6,000 in re-crimps, expedite freight, and one very long Friday afternoon.

What I learned instead is that there's no universal answer, but there are basically three situations. Which one you're in decides almost everything downstream: the connector family, the cable jacket, how much tooling to buy, and how many spares belong on the shelf.

I've been handling connector and cable orders for a panel shop and integrator for nine years. I've documented 23 mistakes that were mine, totaling roughly $14,000 in wasted budget — wrong coding, wrong contacts, wrong jacket, one shipment that went to the wrong customs zone. Now I keep the checklist our team runs before anything gets released. This is the short version of it.

Three situations, three different orders

Before you look at a catalog, figure out which of these describes you:

  • You own the spec. New machine or new line. You're writing the bill of materials, and if you want a different panel cutout, you can have one.
  • You inherited the spec. Retrofit, expansion, or spare parts for a line designed in 2011 by someone who has since left the company.
  • There is no spec. Prototype, lab bench, a single machine, or commissioning work where you need it working by Thursday.

Most bad orders I've seen come from using the third situation's logic in the first situation's context — grabbing what's on the shelf when you actually had six months to plan. Or the reverse: over-engineering a one-off prototype into a $4,000 tooling commitment.

Situation 1: You own the spec

If you control the design, the single highest-leverage decision is when you pick the connector. Not which one — when.

The teams that get this right choose the connector family before the enclosure layout is frozen. The teams that get it wrong — and I was on one — pick the enclosure, then pick a connector that fits the remaining cutout area, then discover eighteen months later that they need four more signal paths through a hood that has room for two modules.

For a typical mixed power-and-signal machine interface, the families sort out roughly like this. Heavy-duty rectangular connectors — the Han series being the obvious reference point in the HARTING range — handle power and signal in one hood and let you add or swap modules later. M12 handles sensors, actuators, and network drops where you want a sealed, keyed, idiot-resistant interface. DIN 41612 stays at the board and backplane level, where it's been doing its job since long before any of us got here.

Here's the part that surprises people: the modular rectangular hood is often the cheaper long-term choice even when it looks expensive on the line-item quote. A fixed-pin-count connector locks your I/O count. A modular one doesn't. I've watched a team re-machine 40 panel cutouts because a machine gained a second Ethernet drop and a brake circuit halfway through the project — a change a modular hood would have absorbed with two extra modules and a longer cable gland plate.

One thing I'd push on harder than most people do: the network cable. If there's any chance the machine needs more than 100 Mbit/s within its service life — and in 2025 there usually is — specify X-coded M12 for the Ethernet runs rather than D-coded. X-coded assemblies run meaningfully higher per unit in my experience, often 30–60% more than D-coded equivalents at 2 m and 5 m lengths depending on jacket and coding (verify with a current quote; copper and jacket materials have both been volatile). But that delta is small against re-pulling cable through a cable carrier after installation. Price it against the rework, not against the D-coded part.

The mistake I made here

In 2019 we built a second machine that was "basically the same" as the first. I approved the panel cutout drawing from memory instead of pulling the as-built. The cutout was 4 mm off on one axis. Fourteen hoods, $890 in replacement glands and hoods, and a week of schedule we never got back. I knew I should check the as-built. I told myself we'd built the same machine six times and what were the odds. The odds caught up with me on a Tuesday.

Situation 2: You inherited the spec

Retrofit and spares work is where I see the most money burned, because the temptation is to buy what the catalog says rather than what's actually on the machine.

Inventory first. Walk the line with a phone camera and photograph every connector interface, including the cable side. Count them. Note the coding. If you're dealing with M12, remember that A-coded, D-coded, and X-coded do not mate, and neither do the various keying variants within a family — and no, the fact that they screw on doesn't mean the pinout matches. "Compatible" isn't a specification. It's a marketing word, and it's the one that costs you a shift.

This is exactly the same discipline as reading symbols on a blood pressure monitor instead of assuming what the little icons mean. The cuff markings tell you whether it's an airtightness indicator or a movement warning, and guessing is not a strategy. Industrial connectors are no different: the IP rating, the coding mark, the temperature class, the approval logos — those are the markings that tell you whether the part in your hand will survive the washdown you're about to put it through.

Ethernet cable is where this gets subtle. If you're replacing a HARTING Ethernet cable on a moving axis, the jacket matters as much as the category rating. PUR jackets for continuous flex and oil exposure; PVC only where nothing moves and nothing splashes. A Cat6A assembly with the wrong jacket will pass every electrical test in the shop and fail in the cable carrier in four months. I've seen that movie twice and I didn't enjoy it either time.

And keep spares. Two of the assemblies you use most, on the shelf, labeled with the machine they belong to. Not because anything is unreliable — because the difference between a 45-minute changeover and a three-day wait is about $180 of inventory.

On multimeters, since people ask

I get asked for the best multimeter for automotive work more often than you'd expect, usually from techs doing AGV or vehicle-adjacent commissioning. My honest answer: the features that matter aren't the ones on the box. You want true RMS, min/max capture (transient drops are what actually break things), a low-impedance mode for ghost voltage, and a decent CAT rating for whatever panel you're opening. Automotive-specific functions — dwell, duty cycle, RPM — are nice when you need them and dead weight when you don't. Buy once for the panel work and borrow the specialty functions.

Situation 3: There is no spec

Prototype, lab bench, single machine, commissioning. The rules here are almost inverted.

The counterintuitive one: for anything under about 50 pieces, hand-crimping your own assemblies is usually the expensive option. Not because the crimping is hard, but because you'll buy the tool, buy a reel or bag of contacts, get the strip length wrong twice, scrap a dozen contacts, and then discover you need a different contact for the wire gauge on the second half of the job. Buy pre-assembled cable assemblies in the lengths you need and put the money into a good ratcheting crimp tool for the sizes you genuinely build in volume. The tool is worth it once you're past a few hundred terminations a year; a bag of the wrong contacts is worth nothing once it's open.

One bench-level detail that has bitten me: if you're doing USB power delivery while recording — long test logs on a laptop or logger running off a USB-C supply — the supply becomes part of your measurement setup whether you planned for it or not. A marginal adapter that renegotiates its power profile mid-log will drop your session and you'll spend an afternoon blaming the instrument. Use a supply you've actually load-tested at the profile you need, and log to local storage as a backup.

How to tell which situation you're in

Five questions, honestly answered:

  1. Can you still change the panel cutout or enclosure drawing without a change order? If yes, you're in Situation 1.
  2. Is there an as-built drawing, and does it match the machine in front of you? If it doesn't match — or doesn't exist — you're in Situation 2, even if the project is technically "new."
  3. Will more than one of these get built? One unit is a prototype regardless of what the project plan calls it.
  4. Is the interface already installed in a cable carrier, a washdown area, or a Class I environment? Then inheritance applies, and you buy to the existing standard.
  5. Does anyone besides you have to maintain this in three years? Then the specification matters more than the unit price.

If you answered "yes" to the first question and "no" to the last one, congratulations, you're building a prototype and should stop reading catalogs and start ordering assemblies.

The question that saves money in all three situations

Whatever situation you're in, ask what's not in the quote before you ask what the price is.

Tooling. Setup charges. Minimum order quantities. Contact and housing sold separately. Freight and customs. Crimp tool calibration. Spare contacts. Whether the "cable assembly" price includes the labor or just the components.

I've learned to ask that question first, every time. The supplier who lists all the fees up front — even when the total looks higher — has come out cheaper for us in the end more often than not, because there's nothing left to surprise you at invoice time. The number you can't reconcile isn't a bargain, it's a liability.

What actually changed for us

There's something genuinely satisfying about a pre-release checklist that works. After all the re-crimps and the wrong-jacket orders and the 4 mm cutout, we finally have a two-page review that takes twenty minutes and has caught 47 potential errors in the past eighteen months — wrong coding, mismatched jacket, missing strain relief, one order that would have shipped to a site we no longer service.

Twenty minutes to avoid a week of rework. That's the whole argument, and it works in all three situations.

Tooling, cable assembly and connector pricing referenced in this article reflects January 2025 observations and personal project costs; verify current pricing and availability with a current quote or the manufacturer's catalog before ordering. Standards references (IEC 60529 for IP ratings, IEC 61076-2-101 and -109 for M12 coding, IEC 60603-2 for DIN 41612) should be checked against the current edition for your application.

Leave a Reply