The One Thing I Check Before Buying Any Generator (And It's Not the Price)

I've rejected more generator shipments over voltage regulation tolerance than any other single spec. Not the price. Not the brand—and I've sent back Caterpillar ends and Westinghouse units alike. A ±5% tolerance might be 'within industry standard,' but it could also be the difference between your Hilti 22V battery charger functioning under load or tripping out.

As a quality & brand compliance manager, I review roughly 200+ industrial power delivery items annually. In our Q1 2024 audit, we rejected 11% of first deliveries for voltage stability issues. The pattern was consistent: vendors spec'ed tight regulation on paper, but actual production units drifted. The trigger event that changed how I think about this was a $22,000 redo on a 50,000-unit annual order where the AC voltage output failed under a simulated critical load test. The spec said ±3%. The delivered unit registered ±5.8%.

Why Voltage Regulation Is the True Cost Driver

It's tempting to think a generator is a generator. You compare kW, fuel type, and price. But the 'just compare unit prices' advice ignores the total cost of ownership when power quality mismanagement damages downstream equipment.

When I implemented our verification protocol in 2022, I ran a blind test on Caterpillar generator ends versus an alternative budget brand. Both claimed ±5% voltage regulation at rated load. Under a calibrated resistive load bank, the Caterpillar unit held a steady ±2.2%. The budget unit hit ±6.8% before the automatic voltage regulator (AVR) overshot correction. That variation is invisible on a basic multimeter test of AC voltage at idle. You have to test under load (ugh, that's where most buyers cut corners).

For a Westinghouse natural gas generator, the risk profile shifts slightly. Natural gas units typically have slightly tighter regulation (±2-3%) compared to diesel counterparts at ±3-5%, because the fuel delivery is more consistent. But that assumes the AVR is adequately cooled. In the same audit, we saw one Westinghouse unit where the transformer was underspec'ed for the enclosure's heat rise. The regulation widened by 1.3% after 45 minutes of continuous run. The vendor argued this was 'within industry standard'—and it was, barely. We rejected it anyway. The cost to rewound the transformer: $380 per unit. On a 200-unit project, that's $76,000 in potential future repairs the buyer wouldn't have budgeted for.

The Hilti 22V Battery Charger Exception

You might wonder, what about small equipment like a Hilti 22V battery charger? Here's the oversimplification: people assume that a portable charger is less critical. In my experience, the AC input stage on a Hilti charger is actually more sensitive to voltage sag than a large generator's control board. The reason is the charging algorithm relies on precise AC zero-crossing detection for its power factor correction. If the AC waveform distorts (a common consequence of loose regulation), the charger derates itself. I've seen chargers take 40% longer to charge because the generator was 'working fine' in terms of lights running, but not in terms of sinusoidal purity.

The industry simplification that 'generators just need to make power' ignores the critical difference between linear load (heaters, lights) and non-linear load (chargers, switching power supplies). I didn't fully understand the value of this distinction until I visited a job site where a crew ran six Hilti chargers off one underspec'ed generator. The chargers kept faulting. The crew assumed the chargers were defective. They weren't. The generator's AVR couldn't maintain the required voltage stability under the pulsed current draw of the chargers' rectifier circuits.

How to Test AC Voltage With a Multimeter (The Right Way)

Most online guides tell you to measure AC voltage between hot and neutral. That's correct for a baseline, but you don't have enough data to assess regulation. The 'check for 120V' advice ignores the fact that voltage stability is measured over time under load. Here's the test protocol I use:

• Step 1: Measure no-load voltage. Record it. For a 120V circuit, you'd expect 120–125V AC. Anything above 130V at no load is a red flag.
• Step 2: Apply at least 50% of the generator's rated load. Use a load bank or known heavy equipment (space heaters, compressors). Measure AC voltage again. The difference between no-load and load voltage is your regulation band.
• Step 3: Monitor for drift over 10 minutes. Write down the voltage at 30 seconds, 1 minute, 5 minutes, and 10 minutes. If it shifts more than 2% from the initial load measurement, the AVR is likely undersized or poorly matched to the alternator.

Note: This test only works if your load is resistive. Inductive loads (motors, pumps) cause additional voltage drop due to inrush current, which is normal. The above test isolates regulation quality from load characteristics.

Boundary Conditions—When Regulation Isn't Everything

Of course, voltage regulation tolerance isn't the only spec. If you're powering only resistive loads (like space heaters), a wider tolerance (±5-8%) is usually harmless. The cost premium for tighter regulation (±2%) might be 15-20% on the generator end. On a 50,000-unit annual order, that's a substantial line item. I've approved budget generators for non-critical applications when the buyer openly acknowledged 'I just need the lights on.' That's fine. The danger is when someone buys a 'good enough' generator and expects it to run sensitive electronics.

Small doesn't mean unimportant—it means potential. The vendors who treated my small test orders (circa 2022, just trying a new product line) with proper spec documentation are the ones I now use for 50,000-unit runs. If a vendor can't provide you with a third-party test certificate for voltage regulation under load, that's a red flag, regardless of the order size.

The lesson from 400+ reviews is this: voltage regulation tolerance is the spec that separates a power source from a power quality solution. The price difference is usually small. The cost of ignoring it—in equipment damage, project delays, and facility downtime—is not.