-
The short version
-
The receipts
-
Five mismatches, in the order I made them
-
1. Eaton 5S UPS: the VA number is marketing, the watt number is physics
-
2. Eaton BRNSURGE whole-house surge protector: panel space is not breaker space
-
3. Push-pull battery disconnect switch: peak amps are a sales number
-
4. "What does a charge controller do" — the answer that isn't the answer
-
5. Rocketfish universal soundbar mounting system: "universal" fits many, guarantees none
-
1. Eaton 5S UPS: the VA number is marketing, the watt number is physics
-
The twenty-minute check
-
Where this advice stops being good advice
The short version
In nine years of buying power and AV gear for client installs, I've logged twelve mistakes that cost my team real money. Eleven of them had nothing to do with the product being bad. They were spec mismatches — right category, wrong number on page three.
Three from 2024 alone: an Eaton 5S UPS sized on VA instead of watts, an Eaton BRNSURGE whole-house surge protector that sat on a shelf for six weeks because nobody counted breaker slots, and a push-pull battery disconnect switch rated to a peak current we would never actually run. Total: about $4,800, plus one phone call I'd rather not repeat.
The rule I follow now: buy the datasheet, not the description. The product category tells you what a part is for. The installation page tells you whether it fits your box, your panel, and your cable lugs.
Four numbers are worth twenty minutes of reading:
- Continuous current at your actual ambient temperature
- Watts, not volt-amps
- Voltage at the worst case (coldest morning, highest load), not the nominal case
- The physical envelope — breaker slots, wire length, stud size, bolt length
The receipts
I run operations for a small systems integrator. I handle procurement for solar/battery, backup power, and AV fit-outs, alongside one very patient installer. I've made twelve documented mistakes totaling roughly $31,000 in wasted stock, rework, and rush shipping. My first one was in 2017, and it was embarrassingly basic: I ordered 60 enclosures with the wrong knockouts because I matched the part number prefix and stopped reading.
I built our pre-purchase checklist in Q1 2024 after the third return in eighteen months. It has flagged 47 problems since. It also failed me once in Q3 2024, which I'll get to.
Five mismatches, in the order I made them
1. Eaton 5S UPS: the VA number is marketing, the watt number is physics
We needed backup for a network rack and a NAS. Measured load: 940 W. I ordered a 1500 VA unit because 1500 sounded like plenty, and because it was in stock with a confirmed next-day ship date.
The Eaton 5S line runs 550 VA to 1500 VA, and the watt ratings are meaningfully lower than the VA numbers — the 1500 VA model is rated around 900 W. We were over that before the rack fans kicked in. On battery, the thing transferred and then dropped load in under two minutes.
Then the second half of the mistake. The client said they needed "about 15 minutes." I sized for 15 minutes at half load, because that's the column people read. At full load, the runtime curve is a different (shorter) story.
I said "about 15 minutes." They heard "15 minutes guaranteed." Thirty-eight minutes of downtime on a $9,000 billable day. We paid $340 in rush freight to get the right size in 24 hours, which I'd do again — the certainty was worth more than the money. The original UPS was the wrong purchase, not the wrong product.
2. Eaton BRNSURGE whole-house surge protector: panel space is not breaker space
I asked the electrician whether there was room in the panel. He said yes. He meant physical room inside the enclosure. I heard space for a dedicated two-pole breaker. Those are not the same thing, and it took six weeks and one very annoyed site visit to discover it.
Second issue: lead length. A panel-mount Type 2 SPD is only as good as its connection, and NEC 285.12 calls for conductors to be as short as practicable and installed without unnecessary bends. Ours ended up with roughly four feet of wire because the free breaker slots were on the wrong side of the bus. Long, looping leads cost you clamping performance — you're paying for a device and then throwing away part of what it does.
Third: a whole-house SPD protects at the service entrance. It does not replace point-of-use protection for the expensive stuff. To be fair, Eaton's documentation is clear about that. I just didn't read past the box.
3. Push-pull battery disconnect switch: peak amps are a sales number
Ordered a switch with a 275 A continuous / 455 A intermittent rating for a 48 V inverter that pulled about 180 A sustained. Looked like a 50% margin. It wasn't, for two reasons.
First, the switch went into a sealed enclosure that ran near 60 °C in summer. Continuous ratings are specified at a reference temperature, and that margin evaporates fast when your ambient is 35 °C above the test bench.
Second, the studs were M10 and our lugs were 5/16" — a $14 part that cost us a two-day delay and a set of re-crimped cables. The cheap fix would have been reading the terminal spec line.
One more thing worth knowing if you're buying these for lockout purposes: a keyed disconnect and a push-pull disconnect solve different problems, and neither is meant to interrupt a loaded DC circuit under normal operating conditions. Shut the loads down first.
4. "What does a charge controller do" — the answer that isn't the answer
Everyone gets told a charge controller regulates charging. True. Useless. The question that actually matters is whether your array's open-circuit voltage stays under the controller's maximum PV input on the coldest morning of the year.
NEC 690.7 requires maximum PV system voltage to be calculated using the lowest expected ambient temperature — not the 25 °C nameplate number. Panel Voc rises as temperature falls, typically around -0.3% per °C for silicon modules.
Our four-panel series string measured about 152 V at nameplate conditions. On a -12 °C morning it would have hit roughly 169 V. The MPPT controller's limit was 150 V. We caught it during commissioning, not during a warranty claim — barely. Nobody was hurt and nothing burned, but that's luck, not process. On the PWM vs MPPT question: if your array voltage is close to battery voltage, PWM is fine and cheap. If it isn't, MPPT isn't an upgrade, it's a requirement.
5. Rocketfish universal soundbar mounting system: "universal" fits many, guarantees none
The smallest mistake on this list taught me the most. Conference room TV, wall-mounted, customer wanted the soundbar on the TV rather than the wall. Simple job. Ordered a Rocketfish universal soundbar mounting system because it said universal.
Here's what the bracket actually does: it sandwiches between the TV and the wall-mount plate. Which means the M8 bolts that came with the TV mount are suddenly too short. Ours were 20 mm; we needed 35 to 45 mm. The retailer didn't stock them. The hardware store two blocks away did, for $6.
We also had to check the soundbar's width against the bracket's arm span, its weight against the rating, and whether the bar would block the TV's IR receiver or bottom-mounted ports. It blocked the IR receiver. We moved the bar down an inch and lost a row of ports.
Total cost: $66 and four hours. It's the mistake that made me write the checklist, and it's the reason I now read the installation page before the product page on anything that mounts to anything.
The twenty-minute check
Four questions, in this order. They've caught 47 problems in eighteen months.
- What is the continuous rating at my operating temperature, and what's the derating curve?
- What's the worst-case voltage or current — coldest morning, heaviest load, highest inrush — not the nominal figure?
- What physical space does this need once it's installed? Breaker slots, lead length, stud size, bolt length, bend radius.
- If I'm wrong, what's the lead time to fix it?
Question four is the one people skip. It's also the reason I'll pay a premium for a confirmed ship date instead of an estimated one. The rush freight is not buying speed. It's buying certainty, and in an install window certainty is the whole product.
Where this advice stops being good advice
If your deadline is soft, the certainty premium is a bad trade. I've watched teams pay $300 to compress a schedule that had 30 hours of slack in it. That's not risk management, that's anxiety with a purchase order.
If the load is non-critical and the failure mode is cheap to fix, buy the cheapest thing that meets the printed spec and move on. Two of my twelve mistakes were over-engineering, not under-specifying. "Probably fine" was, in fact, fine. I paid extra to feel careful.
And the checklist isn't a guarantee. In Q3 2024 I shipped a switch with the correct amperage and the wrong torque spec, because torque wasn't one of my four questions. I've added it. There will be a fifth question eventually.
One last thing, for the record: none of these products were defective. The Eaton 5S delivered what its datasheet said. The BRNSURGE did what a Type 2 SPD does. The soundbar bracket fit everything it claimed to fit. To be fair to all of them, the equipment held up its end. I just read page one and assumed page three didn't exist.
The expensive part was never the equipment. It was the assumption.