Our Data Sources: Why OEM Specs Beat Guesswork

Search “portable power station wattage chart” and you will find the same table repeated across dozens of sites. Refrigerator: 400 to 800 watts. Air compressor: 1,200 to 3,600 watts. Window AC: 900 to 1,400 watts. These ranges are so wide they tell you almost nothing, and the surge figures attached to them are worse.

Generic charts treat entire device categories as interchangeable. A mini fridge and a 26-cubic-foot French door get the same wattage range. A window AC with a variable-speed inverter compressor and a window AC with a single-speed compressor get the same surge estimate. A pancake air compressor and an 80-gallon shop compressor share a row.

That approach is not just imprecise. For motor-driven devices, it is dangerous. If you size a power station based on a generic chart and the chart underestimates the startup spike, the inverter shuts down the moment you need it most.

This is why every number in GeneratorChecker’s database traces back to a specific, verifiable source. This page explains where our data comes from, how we rank source quality, and why the difference matters when real money and real emergencies are on the line.

The Problem with “Refrigerator: 400–800W”

The biggest failure of generic wattage charts is that they blend fundamentally different products into a single row. Here is what the generic charts say versus what OEM data shows for three common device categories:

Look at that last comparison. A generic chart gives you one number for “8,000 BTU window AC.” In reality, a Midea inverter unit surges at 800W while a Frigidaire standard unit surges at 2,010W. Same BTU rating, same category, same row in the chart. One works with a mid-range power station. The other trips it. A chart that cannot tell these apart is a chart that will cost you money, either because you buy a station that is too small, or because you buy one that is far larger than you need.

The running watts problem is just as bad. Generic charts say refrigerators draw 400 to 800 watts. The actual running wattage for a GE Profile PGD29BYTFS French door model is 195W. For an LG LMXS28596S, it is 207W. If you size your power station for 800W continuous just to run a fridge, you are overpaying by a wide margin. If you use the 400W floor and ignore surge entirely, you are underestimating the startup spike.

Neither outcome is acceptable if you are spending $1,000 to $4,000 on a backup power system.

Our Source Hierarchy

Every number in our database carries a source tag. We rank sources from highest to lowest confidence, and we display that ranking on every compatibility page so you can see exactly how much trust to place in each data point.

The hierarchy is not arbitrary. Each step down introduces more estimation, more assumption, and more room for error. An OEM_MANUAL figure was measured by the manufacturer under controlled conditions and published in the documentation they ship with the product. An ENGINEERING_ESTIMATE is our best-informed projection based on comparable devices, clearly labeled as such.

Here is how each source level works in practice, with real examples from our database.

OEM_MANUAL — The Gold Standard

The device manufacturer publishes the exact running and starting watts in the owner’s manual or installation guide. No estimation required.

Example: The ResMed AirSense 10 CPAP machine lists power consumption in its official battery guide. Running power: 53W typical. Peak: 104W with heated humidifier at max settings. These numbers come from ResMed’s own published documentation, not from a third-party review or a generic “CPAP machines use 30–100W” estimate.

OEM_VERIFIED — Manufacturer-Published, Not in the Manual

The running or starting watts appear on the manufacturer’s product page, spec sheet, support article, or FAQ. The data comes from the manufacturer, just not from the physical manual.

Example: CRAFTSMAN publishes a support article for the CMEC6150 pancake air compressor that lists 1,320W running and 6,800W starting watts. These are OEM numbers, verified on their support site, and they are the same numbers used across our database for this device.

NEC_TABLE — Calculated from Electrical Standards

When a motor lists only horsepower and voltage on its nameplate (common for well pumps, sump pumps, and some air compressors), we derive running watts from the National Electrical Code. NEC Table 430.248 provides full-load amps for single-phase AC motors by horsepower and voltage. Running watts equals voltage times full-load amps. Surge is estimated using standard motor engineering factors, typically 3x for general motors and higher for compressors.

Example: A 1/2 HP, 115V sump pump. NEC Table 430.248 lists the full-load current for a 1/2 HP single-phase motor at 9.8A. Running watts: 115V × 9.8A = 1,127W. Surge is estimated at 3x for a standard motor load. This method is standard electrical engineering practice and is the basis for how electricians size circuits across the United States.

GOVERNMENT_DATABASE — Energy Consumption Converted to Watts

For appliances that publish annual energy consumption (in kWh per year) through ENERGY STAR, the DOE Compliance Certification Management System, or AHRI directory listings, we convert those figures to running watts using a duty-cycle estimate.

Example: The LG LMXS28596S French door refrigerator is listed in the ENERGY STAR database at 724 kWh per year. Dividing by 8,766 hours gives an average draw of 82.6W. But the compressor does not run continuously. At an estimated 40% duty cycle (the compressor runs roughly 24 minutes per hour in normal conditions), the running watts during compressor operation are approximately 207W. This cross-validates with the wattage derived from the nameplate current rating on the unit.

ENGINEERING_ESTIMATE — Informed Projection

When a manufacturer does not publish starting watts and the device lacks a code letter or LRA rating, we estimate surge using a category multiplier derived from a comparable device that does have OEM data. This is our lowest confidence level, and it is always labeled.

Example: The Metabo HPT EC914S pancake air compressor does not list starting watts anywhere in its published materials. It is the same motor class and comparable capacity to the CRAFTSMAN CMEC6150, which does publish starting watts (6,800W at 5.15x running watts). We apply that 5.15x ratio to the Metabo HPT’s running watts to estimate its surge. The estimate is clearly flagged as ENGINEERING_ESTIMATE on every page where this device appears.

When V×A from the Nameplate Is NOT Running Watts

Reading a device nameplate seems straightforward: find the amps, multiply by the voltage, get the watts. In many cases, this works. But there are four common traps that catch people regularly, and all of them lead to sizing errors.

The first trap is the most common. Many devices list the circuit breaker requirement on the nameplate, not the actual operating current. A window AC that says “circuit requirement: 15A” does not draw 1,800W while running. The 15A figure tells you what size breaker it needs for code compliance, which is always higher than the operating draw to account for startup inrush. The actual running draw might be 5.9A (670W for the Frigidaire FHWC084WB1).

The PSU trap is especially relevant for gaming PCs and desktop workstations. A computer with a 1,000W power supply does not draw 1,000W from the wall. The PSU is rated for its maximum capacity. Actual draw during gaming is typically 600 to 800W, and at idle it drops to 100 to 150W. Using the PSU rating for sizing will lead you to buy a power station that is significantly larger than necessary.

The multi-voltage trap affects any device with a universal power supply (common on laptops, phone chargers, and networking equipment). A label that reads “100–240V, 2.0A” means the device draws 2.0A at 100V, the lowest and most demanding voltage in its range. At 120V, the same power delivery requires less current. Using 2.0A × 120V = 240W overstates the actual draw.

Cross-Validation: How We Catch Errors

No single data source is perfect. Manufacturers sometimes publish inconsistent numbers across different documents. Government databases occasionally contain data entry errors. NEC tables give theoretical values that real-world devices may deviate from. Our process addresses this with cross-validation: checking each data point against at least one independent source before it enters the database.

The most important cross-validation applies to refrigerators and freezers, because these are the single most common device people want to run during a power outage, and because their actual running watts are dramatically lower than what generic charts claim.

Here is how the cross-validation works for the LG LMXS28596S French door refrigerator:

Source 1: ENERGY STAR database. Annual consumption: 724 kWh/year. At 40% duty cycle, this yields approximately 207W running during compressor operation.

Source 2: Nameplate current. If the nameplate lists a specific amperage at 120V, we multiply to get watts and compare against the ENERGY STAR derivation.

Source 3: GE/LG published spec sheets. When available, manufacturers publish the rated wattage on their product specification documents. For the comparable GE Profile PGD29BYTFS, the ENERGY STAR derivation yields 195W, which aligns with the nameplate data.

When all three sources converge within a reasonable range (typically within 10%), we have high confidence in the number. When they diverge, we investigate further and default to the most conservative (highest) figure.

For motor loads, we cross-reference against NEC Table 430.248 as a sanity check. If a manufacturer claims their 1/2 HP motor draws 15A at 120V (1,800W), but the NEC table says a 1/2 HP single-phase motor should draw 9.8A (1,127W), that discrepancy triggers a review. It does not mean the manufacturer is wrong. Motors vary. But it means we look closer before accepting the number.

We also reference established wattage references from generator manufacturers (Generac and Honda both publish appliance wattage tables) as a supplementary sanity check. These are not our primary data source because they suffer from the same genericization problem we described earlier. But if our OEM-derived figure falls completely outside the range that major generator manufacturers publish, we recheck our work.

What “OEM-Verified” Means on Our Site

Every compatibility page on GeneratorChecker displays a confidence badge next to each data point. This badge tells you where the number came from, not just what the number is.

If you see OEM_MANUAL or OEM_VERIFIED, the number was published by the device manufacturer. If you see NEC_TABLE, we derived it from the National Electrical Code using standard electrical engineering methods. If you see ENGINEERING_ESTIMATE, we projected it from a comparable device.

This transparency exists for a reason. A SAFE verdict is not equally reliable across all source levels. A SAFE backed by two OEM_VERIFIED data points (both the power station and the device have manufacturer-published specs) is a stronger result than a SAFE where one side relies on an engineering estimate. Both pass our three-gate check with buffers applied, but the first has a narrower margin of error.

We do not hide estimates behind a clean interface. We label them. If that makes our results look less polished than a competitor’s simple green checkmark, that is a trade-off we accept. You are making a purchasing decision that could involve thousands of dollars and, in some cases, the reliability of medical equipment during an outage. You deserve to see the work behind the verdict.

Sources: CRAFTSMAN CMEC6150 starting and running watts from CRAFTSMAN Support (1,320W running, 6,800W starting). Frigidaire FHWC084WB1 specifications from Frigidaire product support (670W at 5.9A, 115V). Midea MAW08V1QWT specifications from Midea product page. LG LMXS28596S annual consumption from ENERGY STAR product data (724 kWh/year). GE Profile PGD29BYTFS annual consumption from ENERGY STAR (681 kWh/year). ResMed AirSense 10 power data from ResMed Battery Guide (53W typical, 104W peak). NEC 2023 Article 430, Tables 430.248 and 430.7(B). Metabo HPT EC914S surge estimated from CRAFTSMAN CMEC6150 proxy (5.15× ratio applied to 1,620W running = 8,343W estimated surge).