Automatic Transfer Switch Commissioning Technique
Why this matters
A whole-house generator that starts on the controller's self-test and runs clean with no load has proven exactly one thing: the engine works. It has not proven the transfer switch will actually move the house's load off a dead utility feed, hold it there without drifting outside a safe voltage and frequency band, and hand it back cleanly once the lineman's work is done. Skipping the commissioning sequence and calling the job finished because the unit "ran fine" is how a shop discovers, during the next real outage and in front of the customer, that the neutral bond is wrong, the transfer delay never got set past the factory default, or the shipping interlock pin never came out. Commissioning is the one chance to find those defects on a day when finding them costs an hour, not a callback during a storm.
Before you touch anything live
Confirm the installation is actually finished before you start the sequence: every lug torqued to the manufacturer's chart, neutral and ground bonded at exactly one point (the bonding rule itself lives in Transfer Switch Installation, not here), and any shipping or mechanical interlock pin physically removed from the transfer mechanism. If you didn't do the install yourself, don't take the last tech's word for it. With the utility source still live, open the ATS door and verify with a meter, not a proximity tester, that nothing is energized on the generator-side lugs. Two-source equipment is exactly where that meter check earns its keep: a switch that can close to both sources at once, even for an instant, can backfeed a line a crew believes is dead, which is the leading cause of lineman fatality tied to standby generators. Wear arc-flash PPE selected per NFPA 70E (edition as adopted by your jurisdiction) for the panel class you're opening, and work this step under 29 CFR 1910.333(b)(2), not 1910.147; the lockout/tagout standard's own scope excludes exposure to electrical hazards on conductors and equipment like this. If the interlock pin or shipping lock is still in place, stop here. Do not proceed to Mode 1 with a switch physically prevented from transferring, because a stalled contactor an hour into the sequence teaches you nothing a five-second check right now would have caught.
Mode 1: prove the resting state before you force a fault
With utility power present and the ATS in AUTO, confirm the switch is actually sitting on the utility contacts, not floating between positions or defaulted to generator from a prior test. Read utility-side voltage at the ATS's own display or its labeled test points with the enclosure otherwise closed, never by reaching into the live compartment to probe lugs directly: 228 to 252 volts line-to-line on a 240-volt split-phase service is the steady-state band, and anything outside it before you've even started is a utility problem, not a transfer-switch defect, and commissioning stops until the incoming feed itself checks out clean. Confirm the generator is NOT running and the controller shows Ready or Standby with no active fault. Every later reading in this sequence is a comparison against this baseline; if you don't know what normal looked like before the test, you can't tell a real transfer-switch defect from a pre-existing utility condition once the load actually moves.
Mode 2: force the outage and time the transfer
Open the utility-side disconnect feeding the ATS (the meter-base disconnect or the main breaker, per how the site is configured) to simulate a genuine outage rather than using the controller's internal test button, which on some models skips real sensing logic entirely. Keep the same PPE from the pre-check on for this operation and stand clear of the panel door as the breaker moves. The moment utility drops, start a stopwatch. Acceptance: the ATS registers the loss and sends a start signal inside its programmed sense delay (commonly 1 to 3 seconds; confirm the value you actually set, not the factory default), the engine reaches rated speed within about 10 seconds of the start signal, and the ATS transfers once the warm-up timer expires, landing at the transfer delay programmed for this install (15 seconds on this job; log whatever value you used). If total time from utility loss to load-on-generator runs past your programmed delay by more than a few seconds, the fault is in the transfer logic or a sticking contactor, not the timer setting, and this commissioning does not pass; work New Install Commissioning: Won't Transfer, ATS vs Controller vs Utility Sense Decision Tree rather than nudging delays until a bad transfer looks acceptable. Never shorten or disable the warm-up timer to speed the test along; it exists so the engine reaches stable speed and voltage before it ever sees load, and skipping it is how a genset gets asked to carry a house at 45 Hz.
Mode 3: hold the load and read what the controller is telling you
Before letting the load sit on the generator for any length of time, confirm the enclosure's intake and exhaust louvers are clear of anything left from the install: packaging, offcuts, a ladder leaned against the cabinet. A loaded run pulls real airflow through the enclosure, and a partially blocked path raises cabinet and exhaust temperature fast, on top of the carbon monoxide hazard any blocked exhaust path already carries on its own. Hold the load for 15 to 20 minutes, long enough for the engine to reach full operating temperature under real demand instead of the brief no-load splash of a factory self-test. Read voltage and frequency off the controller display rather than opening the load-side compartment to meter it directly. Acceptance band once load stabilizes: 228 to 252 volts and frequency within 2 percent of 60 Hz, 58.8 to 61.2 Hz. Frequency or voltage that drifts outside that band, hunts instead of settling, or an ATS that re-transfers on its own with no real utility event, means governor, AVR, or ATS-sensing work before you sign anything, not a note to revisit at the first annual service.
Mode 4: force the return and confirm the system re-arms itself
Close the utility disconnect you opened in Mode 2, again with PPE on and standing clear as the breaker seats. The ATS should detect return within its sensing window, hold it through the full return delay (20 minutes on this install; log the actual value), transfer back, run the generator through its cooldown period (90 seconds here) with the load already back on utility, then shut down clean with no fault. Confirm the ATS is sitting solidly on utility contacts again and the generator controller has returned to Ready with no active alarm. This closes out the disconnect you opened two steps ago: if the switch is still showing generator, or straddling both positions, do not leave the site, because that's a two-source condition, not a documentation gap. The return delay exists to keep an intermittent utility restoration from bouncing the load back and forth; don't shorten it just because the customer is standing there waiting.
A commissioning run that fails, and how it recovers
Kohler 20RESAL, 200A whole-house ATS, day 3 of a fresh install. Mode 1: utility reads 238 volts, ATS on utility, controller Ready, no faults. Mode 2: utility disconnect opened, stopwatch running; ATS registers loss at 2 seconds, engine cranks and reaches rated speed at 9 seconds, warm-up timer expires, ATS transfers at 15 seconds total, matching the programmed delay. Mode 3: load held. At 6 minutes into the run the controller display starts showing frequency swinging between 57.1 and 62.6 Hz, well outside the 58.8 to 61.2 band, hunting roughly every 20 seconds instead of settling. Stop rule: this commissioning does not pass. Retransfer to utility immediately rather than let the ATS discover the fault on its own delay logic, and do not sign the acceptance record. Inspection finds the governor linkage binding against a zip tie left over from shipping, restricting throttle travel under load. The tie is cut and the linkage moves freely through its full range by hand. Before re-attempting Mode 3, the fix gets its own verification, not a note that it's resolved: re-run the full sequence starting at Mode 2, not a spot-check of frequency alone, because a linkage that was binding could also have affected transfer timing under load. Second attempt: transfer again at 15 seconds, frequency holds 59.6 to 60.3 Hz for the full 20-minute hold, Mode 4 retransfers at the programmed 20-minute delay, cooldown runs the full 90 seconds, controller returns to Ready with no fault. Only this second, complete run gets logged as the passing commissioning.
Confirming the commissioning is complete
Before you leave the site or hand the job to the orientation visit, the log needs all four modes present, not a subset: Mode 1's baseline voltage, Mode 2's transfer time against the programmed delay, Mode 3's voltage and frequency across the full hold with no drift outside band, and Mode 4's return time against the programmed delay with a clean cooldown, plus the interlock and shipping-pin check from before you started. A commissioning missing any one of those four is a start test with extra steps, not a commissioning. Log the actual delay values you set, not the factory defaults; that log is what the next tech reads before touching this unit, and it's the record Post-Install Customer Orientation and Exercise Schedule Standard treats as a prerequisite before that visit even begins.
References
- NEC Article 700, Emergency Systems, and Article 702, Optional Standby Systems, transfer-equipment requirements.
- NFPA 110, Standard for Emergency and Standby Power Systems, commissioning and acceptance-test provisions, edition as adopted by your jurisdiction.
- UL 1008, Standard for Transfer Switch Equipment.
- 29 CFR 1910.333(b)(2), electrical safe work practices for this class of work.
- NFPA 70E, arc-flash PPE selection, edition as adopted by your jurisdiction.
- See related: Transfer Switch Installation; New Install Commissioning: Won't Transfer, ATS vs Controller vs Utility Sense Decision Tree; Post-Install Customer Orientation and Exercise Schedule Standard.