Figure 1 shows the spectral power distribution for a MR16 20/12 BAB/FL/40 lamp operating on slightly reduced voltage to provide 50 lux in the experimental situation with a CCT of 2857 K. Note the characteristic smooth, continuous curve, climbing towards the long wavelength end of the spectrum. The decline above 700nm is due to the dichroic reflector, which is transparent to TX and extreme long wavelength visible radiation.
While a practical 3-band light source for museum use will comprise a single lamp, with or without a filter, for the experimental situation the three bands were provided by separate lamps, each with a band-pass filter. The band center wavelengths identified by Thornton are at 80nm intervals, and three types of 50mm diameter, 40nm band pass filters were obtained. The 450nm filter was available as a stock item, but the 530nm and 610nm filters were custom items. The filters were mounted in compact industrial luminaire housings fitted with 50-watt MR lamps. Because of the low radiant power from these lamps at short visible wavelengths, it was found necessary to use two 450nm sources.
Figure 2 shows the spectral power distribution for these lamp and filter combinations with their outputs balanced to match the CCT of the source shown in Figure 1. The difference in the SPD curves is strikingly obvious: not only are the end parts of the visible spectrum missing, but also there are two deep notches in the curve. Data for the MR and the 3-band sources are given in Table 1, from which it can be seen that the radiant luminous efficacy for the 3-band source is 70% higher that that for the MR lamp. This means that at equal illuminances (lux) the 3-band source will produce 41% less irradiance (W/m2).
The process of adjusting the balance of the three wavebands of the experimental source was tedious and time-consuming. However, the experimental procedure required the 3-band source to be dimmable, and this created a problem: how to vary the output of the source while maintaining the balance of the wavebands? The problem was solved by feedback device that was developed specially for this experiment. Each lamp housing was fitted with a light sensor directed towards the lamp. The sensors were connected to the feedback device, which continually monitored the outputs of the lamps and maintained them in constant ratios. The control operated by the subject changed the output of the mid-waveband lamp, and the feedback device made instantaneous proportional changes to the lamp outputs for the other two wavebands. By turning a switch outside the test room, the experimenter selected whether the subject’s control operated the 3-band source or the MR lamp, so that the subject was given no obvious indication of the type of light source that was in use.
The first phase of the experiment was completed using low CCT (approximately 2850 K) sources. For the second phase, the regular MR lamps were replaced with 12-V 50-W “SoLux” MR lamps, which were dimmed to provide 50 lux at approximately 4200 K. Figure 3 shows the spectral power distribution, and the loss of long wavelength power to achieve the higher CCT is apparent. The 3-band source was adjusted to match this CCT, and the spectral power distribution is shown in Figure 4. Data for these sources are shown in Table 1, and for this higher color temperature the radiant luminous efficacy for the 3-band source is 46% higher than for the MR lamp. The main reason for this difference being less than for the low color temperature case is that the radiant luminous efficacy of the MR lamp is higher. However, this is not because this is a more efficient lamp, but is due to the lower radiant power emitted at the long wavelength end of the visible spectrum.