Picture this. You've got 2,400 general purpose relays scheduled into a build. The line runs next Tuesday. Your regular distributor quotes six weeks. A second source quotes three weeks at roughly 14% less. You split the order.
The relays land Tuesday morning. Boxes, labels, part numbers — everything looks right. You pull forty units for incoming inspection, they all operate on the bench, and you release them to production.
Six weeks later, eleven of them are intermittent. Two have welded shut. The machine's been throwing phantom faults for a week and nobody's connected the dots yet.
That's when the phone calls start.
The Problem You Think You Have
If you've been through this, you already have a theory. The relays were counterfeit. Or they were genuine but came through a grey channel. Or the seller shipped factory seconds and relabeled them.
Any of those can be true. But after four years of reviewing incoming material for an industrial control distributor — roughly 200 line items a year, across maybe 40 manufacturers — I can tell you that the counterfeit theory is the one people reach for first because it has a villain. It's satisfying. It's also usually only part of the story.
The decision that caused the failure happened months before the shipment arrived. And it didn't feel like a bad decision at the time.
The Problem You Actually Have
You're buying a relay the way you buy a bolt
A bolt is a bolt. Grade, size, finish — that's the whole specification. Confirm those three things and the cheapest source is the correct source.
A general purpose relay isn't a bolt. It's an electromechanical assembly with a coil, an armature, spring contacts, and a housing — and every one of those has a tolerance band. Two relays with the same part number printed on the case can behave differently in month six if they came from different production lots, or if the coil wire gauge drifted, or if contact plating thickness moved from three microns to one and a half.
None of that shows up on a bench test. It shows up at 10,000 operations, in a panel at 45°C, running sixteen hours a day.
The specs that matter are the ones nobody reads
Open any Omron relay datasheet and you'll find electrical durability curves. Not a single number — curves. Operations at rated load, operations at low load, operations at cos φ = 0.4 versus cos φ = 1.0. The distance between the top of that curve and the bottom is usually one or two orders of magnitude.
IEC 61810-1 governs how those durability ratings are tested and expressed for electromechanical relays. It doesn't tell you what number to expect. It tells you that the number is only meaningful next to the load conditions it was measured under. A relay rated "100,000 operations" with no qualifier attached is marketing, not engineering.
Mechanical life and electrical life aren't the same thing. Mechanical life is the armature cycling with no load on the contacts. Electrical life is the contacts actually doing work. On a typical general purpose relay, the first number can be 50–100x the second. It's the number on the front of the catalog page because it's bigger.
The socket is half the circuit
Here's the one that catches people. A relay doesn't sit in a panel by itself — it plugs into a socket, and the socket is a set of spring-loaded contacts carrying the full load current, day after day, through thermal cycles.
A poor socket doesn't fail. It degrades. Contact resistance creeps up. The coil draws slightly more current. The contacts arc a little harder on break. Eight months later the relay gets blamed for a failure that started in a three-dollar base.
When we audit an incoming lot now, the socket or base ships in the same inspection as the relay. Same lot, same paperwork, same rejection criteria. Buying general purpose Omron relays from one channel and sockets from another — even when both are genuine — is how you end up with a mismatch nobody can trace.
It isn't just relays
The same pattern runs through everything else in a controller catalog. Bulk PLC purchases get negotiated on price per unit and then sit in a cabinet for a decade. A contactor specification guide gets skimmed for the amperage rating while the utilization category gets ignored — and the gap between AC-3 and AC-4 duty, which IEC 60947-4-1 defines separately for good reason, is the difference between a contactor that lasts a decade and one that lasts eighteen months.
The part number didn't change. The application did. Nobody asked.
What It Costs When You Find Out Too Late
Let's be concrete, because this is where the 14% gets embarrassing.
The diagnosis. Intermittent faults on a running line are the worst kind. Not a hard failure you can find with a meter — a symptom that moves. Two electricians, a maintenance tech, and an engineer on the phone for three days is not unusual. At fully loaded labor rates, that's already more than the entire price difference on the order.
The downtime. Every hour the line is down, the cost of that hour doesn't care what the relay costs. It's throughput. It's overtime. It's the customer notification you're about to send.
The collateral. A welded relay contact can take an input card with it. If it's in a safety circuit or a safety PLC segment, you're not replacing a part anymore — you're re-validating a function. Different budget, different calendar.
The scramble. This is the piece left out of the post-mortem. Once you've decided the stock is suspect, you need replacements. Not next week. Now. You call every channel you have. You pay whatever's asked. You pay overnight freight on a pallet.
When you're standing in that gap, the price of a relay stops mattering entirely. I've watched a team pay triple list price plus air freight for two hundred units of a part they normally buy on a quarterly plan. The premium wasn't for the relay. It was for the certainty that it would be on the dock before the line started Monday.
We used to treat "probably on time" as good enough. After two real misses, we don't. Our purchase terms now separate the price of the product from the price of the schedule, and we budget for the second one on purpose — because a delivery window you can't verify isn't a delivery window, it's a hope.
Honestly, I still don't fully understand why some channels hold their quoted lead times on the same part while others drift by a week or two. My best guess is it comes down to how much confirmed stock they're actually holding versus how much they're counting on inbound. I've never gotten a straight answer from anyone who'd have reason to know.
That first bad batch saved us roughly 14% on unit price. Cleaning it up cost more than every unit-price saving we booked that year, combined.
What We Changed
Not a lot, and none of it is clever.
- Paperwork before pallets. Any new channel produces traceable documentation before the first unit ships. Not a photo of a datasheet — lot-level origin, channel of record, and a return path that isn't a phone number in a different country.
- Sample testing at the failure point, not the bench. We cycle relays under load, hot, in a fixture that looks like the real panel. Bench testing at room temperature passes almost anything.
- Relay and socket as one line item. Specified together, inspected together, rejected together.
- Contingency budget for lead time. We set aside money for expedited freight on critical lines before we need it. It stops being an emergency cost and becomes a line item.
If I remember correctly, we put the incoming-inspection protocol in place in early 2023 after a near miss, though I might be off on the exact timing. The specifics matter less than the rule: nothing goes to production that we can't trace back to a source we've verified. Not on price. Not on urgency.
The unit price of a relay was never the cost. The cost is the line that stops. What you're actually buying when you source industrial control components is a delivery you can plan around — and when a supplier can give you that, it's worth paying for. When they can't, the savings are imaginary and you're just borrowing against a bad afternoon.
