The 9-Point Test Checklist I Use for Every Industrial Power Conditioner

When this checklist applies

I use this before commissioning any industrial power conditioner — single-cabinet wall-mount, floor-standing 100 kVA, whatever sits between the utility feed and the load. It works for stand-alone power conditioners, for the conditioning stage inside an industrial UPS system, and for industrial DC power supplies that need clean input. Same nine points, same order.

Here's the thing: most commissioning checklists I've seen are either too abstract ("verify proper operation") or lifted straight from the manual ("follow manufacturer instructions"). Neither tells you what to actually do with a meter and a clipboard. So this isn't a manufacturer's checklist. It's mine, built from 180 installs and two expensive misses.

In my first year, I installed a 30 kVA conditioning unit without verifying the grounding conductor termination. It read fine on the meter. Three weeks later, the customer's CNC controls started glitching. Loose bond on a grounding lug that "passed" a basic continuity check. That one cost $4,200 in service calls and a credibility hit I couldn't invoice.

The second one was worse. I approved a batch of industrial power conditioners without checking no-load output regulation. Every unit passed the loaded test. Only after deployment did we discover four of them drifting 3% high at zero load — enough to trip downstream protection. $8,000 in units plus $2,500 in field labor. I still kick myself for that one.

The nine-point test sequence

1. Cross-check the nameplate against the spec sheet — and the site

Nameplate mismatches are more common than you'd think. In the past 18 months I've caught 11 — wrong kVA rating, mismatched input voltage window, one unit labeled 50 Hz that shipped with a 60 Hz plate. The nameplate has to match the purchase order, the spec sheet, and the actual site conditions. Three sources, one truth. If they don't align, stop.

2. Visual and mechanical inspection before you energize anything

Loose terminal blocks, shipping damage, debris inside the enclosure, cracks in the insulation. You'd think this is obvious, but when the crate shows up at 4:30 on a Friday, obvious goes out the window. Ten minutes with a flashlight now saves a service call later. Check torque on accessible terminations — I've found factory connections that were hand-tight at best.

3. Grounding and bonding — continuity, then resistance

This is the step most people skip. Don't. Per NEC Article 250 (for US installations), grounding and bonding isn't optional, and a continuity beep doesn't prove a solid bond. You need to measure resistance to ground. I learned that lesson in 2017 — the continuity checker sang, the bond was loose, and the customer's controls paid for it three weeks later.

4. Input voltage window — sweep it, don't spot-check it

Check the unit's rated input window against the actual supply. Then sweep it: low line, nominal, high line. If the nameplate says 190–250 V, feed it 190 and 250 and watch what the output does. Spot-checking at nominal tells you nothing about the edges where most power-quality problems actually live.

5. No-load output regulation — the step I wish I'd never skipped

Here's the counterintuitive one. Most people test a power conditioner under load and call it done. But some units drift at zero load — sometimes 2–3% above nominal, which is enough to trip downstream protection when the load drops off. I found this out the hard way on an eight-unit order in 2022. Now I measure output at no-load first, then at 25%, 50%, 75%, and 100%.

6. Full-load regulation and thermal stability

Run it at rated load for at least 30 minutes — longer if you can. Watch the output voltage drift as the unit warms up. Some conditioners hold steady cold and drift hot. Others do the opposite. The spec sheet says "±1% regulation" but doesn't always tell you at what temperature. Log the output every 5 minutes and compare the trend.

7. Waveform quality — THD in, THD out

Feed it a distorted input and measure the output THD. IEEE 519 is the usual reference here, though the exact limit depends on your system. The question to ask: does the conditioner actually clean up the waveform, or does it pass distortion through? Some cheaper units look great on a clean sine wave and fall apart when the input has real-world noise.

8. Dynamic response — voltage sag and swell recovery

Simulate a sag (drop to 80% or less for 100–200 ms) and watch how fast the output recovers. Same with a swell. The spec sheet will quote a response time, but "response time" means different things to different vendors. Measure the actual time from event to stable output, not just the correction initiation. If you're feeding a data center's modular UPS, this step matters even more — downstream loads are less forgiving.

9. Protection and alarm verification

Force each protection threshold: overvoltage, undervoltage, overcurrent, overtemperature. Does the unit trip at the right point? Does the alarm fire? Does the bypass engage correctly? I've seen units that passed every performance test but had a miswired alarm contact that would have left operators blind during a fault. Alarms are cheap to test and expensive to ignore.

Common mistakes and things I still get wrong

Three things I've learned the hard way:

• Testing only the "good" scenarios. Nominal voltage at 50% load looks great on every unit. The failures happen at the edges.
• Not recording a baseline. If you don't log what "normal" looked like on day one, you can't tell whether a drift six months later is a problem or just how the unit runs.
• Trusting the vendor's QA sheet. Many say "tested and passed." Passed what, exactly? They don't say. I stopped trusting those sheets after finding a loose bond on a unit whose QA sticker was still intact. The most frustrating part: the same issue recurring across batches from different vendors. You'd think written specs would prevent it, but interpretation varies wildly.

One more thing: my experience here is based on roughly 180 installs, mostly 10–100 kVA units in light industrial settings. If you're commissioning MV-class equipment, medical-grade conditioning, or data-center-scale modular UPS systems, your protocol should be stricter than this. I can't speak to those from direct experience.

Honestly, I'm not sure why some units drift more under no-load than others. My best guess is that it comes down to how the regulation circuit is tuned at the factory — but I've never gotten a straight answer from any manufacturer on that point. If someone has better insight, I'd genuinely like to hear it.

We run PowerSentinel power conditioner industrial units through this checklist now. Most pass without drama. But "most" isn't "all," which is exactly why I still check every point, every time. Five minutes of verification beats five days of correction — that's the whole reason this list exists.