Some plots of the time-waveforms and (low) harmonic spectrum were made, and a calculation of true power consumption was made: both were very close to the rated power. My hand-based heat detection seems to have been wrong.
In the following, the time-graphs show voltage (blue) and current (red). The bar-graphs show the low harmonic spectrum, by an FFT ("fast fourier transform").
Test: mains_L2 P : 7.46W Vrms : 232V |V|mean : 210V Vpk/sqrt(2): 236 form: 1.11 Irms : 0.052A |I|mean : 0.030A Ipk/sqrt(2): 0.15 form: 1.71
Test: mains_L3 P : 18.7W Vrms : 232V |V|mean : 210V Vpk/sqrt(2): 240 form: 1.105 Irms : 0.131A |I|mean : 0.0688A Ipk/sqrt(2): 0.306 form: 1.91
The original data is here, just in case you want it! Each file is CSV data with the first column being time and the second the v or i measurement. v is through a x10 probe (i.e. multiply it by 10 to get Volts), i is the voltage across a 3.33 Ohm series resistance that carried the lamp current (i.e. divide i by 3.33 to get the current in Amps). In case it's of use to know this, the voltage measurement included the drop across the shunt.
The (matlab) script used to read and plot the data is here.
Page started: 2005-03-13
Last change: 2014-02-22