Commission a 300 W xenon source as an acceptance test, not a switch-on. Six checks carry it: a parts audit, lamp seating, coaxial alignment, a 10–15 min warm-up, irradiance calibration at the sample plane, and a stability run. Each one needs a written pass criterion. Until all six pass, you own a lamp rather than an instrument.
Each check isolates one failure mode, so a failure tells you where to look instead of only that something is wrong. Run them in the order below; every step assumes the previous one has passed.
| Check | What it proves | Pass criterion | Failure signature |
|---|---|---|---|
| Parts audit | The shipment matches the configuration you will actually run | Every line of the packing list accounted for; lamp wattage matches the order; quartz parts and filters uncracked | A missing filter or the wrong lamp class, which invalidates every later comparison |
| Lamp seating | The arc sits where the reflector expects it | Lamp fully seated, no rocking, contacts clean and dry | Output that will not reach its expected level at full power |
| Coaxial alignment | Lamp axis, reflector, and exit port form one line | Beam concentric at two distances along the axis | Elliptical or offset spot; irradiance that never peaks as you adjust |
| Warm-up | Output has reached thermal equilibrium | 10–15 min at the power you will use, before the first measurement | Irradiance still climbing through the first data points |
| Irradiance calibration | The recorded number describes the sample plane | Three consecutive meter readings agree at the same position | Readings that change with small movements of the sensor |
| Stability run | The source holds output for the length of one experiment | Logged output flat across the longest run you plan | Monotonic decline or step changes inside the run |
Two rows are usually skipped and should not be. The parts audit catches configuration errors that no amount of alignment will fix, and the stability run is the only check that tests the source over the time scale you care about. A source that passes five checks and fails the sixth is not eighty percent commissioned.
Xenon output rises for the first several minutes after ignition, so any reading taken inside that window describes the ramp rather than the lamp. That makes the 10–15 min figure a floor for a cold start, not a suggestion. A re-strike after a brief shutdown recovers faster but still takes minutes, and the recovery is rarely to exactly the same level.
Verify it instead of assuming it. Log output every 30 s for 20 min at the power you intend to use, then plot the series. The check passes when the last several readings sit inside the noise band of your meter, not when the clock runs out. That log also gives you a reference for the day the ramp starts taking noticeably longer, which is an early signal of lamp aging.
The practical rule for a campaign: ignite the source, start the log, and do everything else — cell loading, purge, baseline acquisition — while it stabilizes. By the time the reactor is ready, the source should already be flat. Background on ignition, arc behavior, and the run-in period is covered in how a xenon lamp behaves during warm-up.
Irradiance belongs at the sample plane, not at the exit port, because everything between those two planes — a filter, a fiber, a window, a water jacket — modifies the beam. A port reading quoted as an experimental condition is a number with no defense when someone asks how the light reached the catalyst.
A 300 W source in this class delivers 50 W total optical power, and lamp power is adjustable from 150–300 W. Those are source-level figures; they say nothing about the mW/cm² at your catalyst until you measure it. For sensor selection and placement, see choosing and using an optical power meter.
A source is only as trustworthy as its longest logged run, so run it once under real conditions and keep the record. Use the reactor in place, the filter installed, the cooling on, and the duration of your longest planned experiment. Ambient temperature belongs in the log too, because a warm room and a cold room produce visibly different curves on the same hardware.
Judging the result is straightforward. Flat within the meter noise passes. A slow monotonic decline points at cooling, airflow, or an aging lamp; a step change points at a connection, a ballast fault, or something mechanical shifting. Each has a different remedy, which is why the distinction is worth recording rather than averaging away.
If the calibrated irradiance sits well below what the source rating implies, check alignment and lamp age before questioning the rating itself. The relationship between electrical input and delivered light is covered in the 300 W xenon source selection guide.
Plan most of one working day. The physical checks take minutes each; the time goes into the warm-up log and the stability run, which should last as long as your longest experiment. Compressing those two steps is the usual reason a first dataset cannot be reproduced later.
No, because the setpoint is electrical and the experiment depends on optical. Lamp aging, window haze, and a changed filter all alter the relation between the two. Calibrate at commissioning, then re-check after any optical change and at fixed intervals during a long campaign.
Any monotonic trend across the run is a stop signal, because it biases every rate you calculate from that session. A step change is different: it usually indicates a connection or ballast fault that can be fixed and re-tested the same day. Log the shape of the drift, not just the endpoints.
