I want to measure the CIE coordinates of an LED but I do not have access to an integrating sphere nor to a calibrated irradiance standard lamp. I have an Ocean Optics USB4000 spectrometer (≈1.5 nm resolution) and I use SpectraSuite software.

Procedure I used: I measured in relative irradiance mode and used a halogen source (HL-2000) as the reference. To maintain a symmetric geometry I built a simple holder connecting the lamp to one optical fiber and the spectrometer to another fiber. After taking the reference measurement I disconnected the lamp and connected the fiber from the LED test fixture to the holder and measured the LED. I tested LEDs of different colors and, in the spectral/chromaticity plot, the measured color corresponded to each LED.

Can you tell me if this approach is correct or whether there is anything I could improve? (Note: I built the holder so I could place a neutral density filter between the fibers in case the signal saturates.)