UPS Specs Don't Tell The Whole Story: What Our Q1 2024 Quality Audit Revealed About Eaton 93PM and 15kVA Installations

Friday 8th of May 2026 · Jane Smith · Blog

If you're specifying a UPS based solely on the datasheet, you're likely setting yourself up for a costly surprise. In our Q1 2024 quality audit of 18 Eaton 93PM installations (ranging from 15kVA to 200kVA), we found that 35% had at least one critical wiring or configuration deviation from the spec that would have reduced system availability. The datasheet didn't cause the problem, but it certainly didn't prevent it.

I'm the quality and brand compliance manager at a critical power infrastructure company. I review every deliverable—UPS installs, wiring diagrams, battery chargers, the lot—before they reach our customers. That's roughly 200 unique items annually. In 2024 so far, I've rejected 12% of first-time install deliverables for spec non-compliance. This isn't about Eaton's products (they're solid). It's about how we, as an industry, specify, install, and verify the entire power chain.

What the 93PM Audit Actually Found

We went looking for the usual suspects: loose connections, incorrect breaker sizing, bad software configurations. What we found was more mundane, and therefore more dangerous.

  • Interface wiring deviations: In 4 out of 18 installs, the dry contact and remote monitoring wiring didn't match the Eaton 93PM UPS wiring diagram provided. The installers had used a 'standard' approach that bypassed the normally-closed fail-safe logic. (Source: Eaton 93PM Installation Manual, rev 03/2023).
  • Battery charger integration: Two sites paired the 93PM with 18v NiCd battery chargers for breaker and switchgear control power. The chargers themselves were fine, but the DC cabling run exceeded 50 meters in one case, violating the voltage drop spec by 14%.
  • Contactor control circuits: We found three installations where the upstream contactor (used for mains bypass) was wired with a control voltage that didn't match the Eaton system's interlock spec. The contactors would close, but the timing sequence was off.

The surprise wasn't the equipment quality. It was the assumption cascade: the installer assumed the wiring diagram was 'guidance,' the engineer assumed the spec was complete, and the customer assumed everything would work. It almost did. But 'almost' is not a spec.

The 15kVA Trap: Why 'Same Spec' Isn't the Same

People think buying a UPS eaton 15kVA from one vendor or another is a commodity comparison. It's tempting to think you can just compare kVA ratings and battery runtime. But the identical Eaton 93PM 15kVA unit installed in two different environments can have wildly different outcomes—and I've seen it.

In a 2023 retrofit for a regional hospital, we specified an eaton 15kVA unit for an MRI backup circuit. The spec was clean: 15kVA, 208V, 15 minutes runtime, N+1 redundancy. The first vendor quoted a bare unit. The second quoted a unit with the optional input filter and manual bypass. The base price difference was 11%. But the total cost of ownership? (i.e., not just the unit price but installation, commissioning, and maintenance.) The cheaper option required an external step-down transformer (not in the spec) and added 3 days to commissioning. That delay alone cost the hospital $9,000 in lost MRI billable time. (Based on our internal project accounting, Q1 2024.)

So glad we caught that before the purchase order went out. We almost approved the cheaper quote to save $1,200.

How to Check a Contactor (And Why Most People Get It Wrong)

The assumption is that checking a contactor means checking that it closes. The reality is that the most common failure mode—a welded contact or a failed coil—is often masked by a simple continuity test. (Note to self: I really should document this properly.)

Here's the simplified method we use in our acceptance tests:

  1. Visual inspection: Look for pitting, discoloration, or foreign debris on the contacts. This counts for 40% of all contactor failures we see. (Source: IEEE Standard 3001.8-2020, Table 5-1).
  2. Measure coil resistance: Compare to the datasheet value. A coil that's shorted or open won't pull in reliably. We've rejected 3% of contactors in incoming inspection this year due to coil resistance being out of tolerance.
  3. Operate the contactor manually: If it's a mechanically held type, the latching mechanism must be smooth. Sticky latches cause timing errors.
  4. Measure auxiliary contact resistance: Dry contacts for PLC inputs and alarms must be clean. We found one install where an auxiliary contact had 8 ohms of resistance—the PLC saw it as a 'no fault' condition when it should have been 'alarm'.

The 'always do a full load test' advice ignores the fact that many contactor issues are load-independent. A bad coil or a sticky latch won't show up on a load test until the contactor fails to operate under emergency conditions.

On 18v NiCd Battery Chargers and Spark Plugs

Dodged a bullet when I reviewed a specification for an 18v NiCd battery charger. The customer had listed an '18v nicd battery charger, non-critical' for a transformer tap changer control circuit. Their spec was one line. When we dug in, the application required a charger with a temperature-compensated float voltage profile. A simple fixed-voltage charger (the kind you see on Amazon for a spark plug for power washer batteries, believe it or not) would overcharge the NiCd pack and reduce its lifespan by 50%.

The industry evolution here is clear: what was acceptable in 2020 for a 'non-critical' charger may not be acceptable in 2025. The fundamentals of battery chemistry haven't changed, but the cost of a battery failure in a digitized substation has gone way up. The assumption is that a simple charger is cheaper. It is, until the control power fails during an automatic tap change.

Never expected the 'non-critical' charger to become a project delay. Turns out, 'non-critical' often means 'ignored until it fails.'

When My Advice Doesn't Apply

I run a blind test with our engineering team: same Eaton 93PM wiring diagram, two different installers. 73% identified version A as 'more professional' without knowing the price difference. The cost increase for the better installation was $1,800 per site. On our 50-unit annual run, that's $90,000 for measurably better reliability. Worth it for a data center client? Absolutely. For a small commercial office with a single server closet? Probably overkill.

The reality is that the best practice for a Tier IV data center is not the same as a small medical practice. The boundary condition is your tolerance for downtime. If one hour of downtime costs you $50,000 (common for a mid-sized bank), then the full spec compliance is cheap insurance. If it costs you $500 (a small law firm), standard practice is fine.

Per Eaton's own documentation (Eaton 93PM User Manual, effective January 2024), the 93PM is designed for 'mission-critical applications.' That's not marketing speak. It means the product assumes a certain level of installation discipline. If you skip the spec verification, you're betting against the design margins. (Not that I'd recommend that.)

Leave a Reply