Library of normalized spectral plots

The computational procedure described in the article generates as a sub-product the spectral plots normalized to any desired set of physical units. I choose to represent the horizontal axis in nanometers (nm), and the vertical axis in uE/s/Watt/nm. That is, the plots show effectively how the quantity PAReff per wavelength unity changes as a function of wavelength.

With this particular choice of units for the plot axis, bulbs of different power can be directly compared in efficiency terms. Note that slight differences in efficiency may still exist when comparing bulbs of the same type but different power. For instance, a 250 Watt metal halide can be more efficient than a 150 Watt MH of the same brand/type. The plots do not take effects such as these into account.

Using photon units instead of the usual energy units in the plots makes it possible to directly compare bulbs in terms of how the photosynthesis process "sees" then. As far as I know, no other representation of lamp light spectra in such units was yet published.

Notice that the vertical scale encompasses a different range in each plot. To ease the comparison of curves plotted in separate plots, the Cool White fluorescent is used as a reference and plotted in all figures in black. Thus one can compare spectra in different figures based on how each one compare with the Cool White.

Spectra plotted in these figures were not rebinned to some standard, common spectral resolution. They are shown exactly as they were published, except for the units on the vertical axis. This must be taken into account when comparing sharp spectral features between spectra of dissimilar resolutions. For example, the GE Fresh & Saltwater bulb seems to have a much more intense emission at 600-620 nm than the Triton bulb. However, this is only an artifact created by the lower resolution in which the Triton spectrum was published. What should be compared is the integral under the curve, which is how the tabulated figures in this article were generated.

The horizontal scale encompasses the same range of wavelength in all plots. This is the same range used to define PAR. Bulbs where grouped in each plot roughly according to their efficiency. The plots are quite fine-detailed and do not display well in a web browser. There are links to access both a JPEG and a PostScript version of the plots. The PS format is highly recommended since it can be magnified without loss of information.

The first plot (picture 1) depicts the high intensity discharge bulbs. Note how diverse can be the spectra, with different degrees of "fullness". The low efficiency of the mercury vapor bulb is evident, as it barely produces any output outside a few narrow regions. The Wonderlite MV spectrum looks much better, but still suffers from low efficiency. Also of note is the extremely large amount of red light produced by the HPS bulb.

This plot (picture 2) depicts the five higher efficiency fluorescents included in the sample, plus the Pentron HO bulb. Of note is the fact that they have very similar spectra, typical of tri-phosphor or rare-earth tubes. Note that, in spite of the peaky nature of these spectra, these bulbs are so efficient that even in most of the "valleys" (wavelength ranges with minimal emission) there is still sufficient emission to compete with full-spectrum bulbs.

This plot (picture 3) depicts spectra of somewhat lower efficiency fluorescents that are still dominated by the tri-phosphor signature. Note that the Daylight Deluxe could be included in the full-spectrum family as well. Note also the large amount of blue generated by the Triton bulb, beaten in this regard only by the Aquarelle bulb.

This plot (picture 4) depicts spectra of "plant" bulbs. That is, bulbs designed to match the Photosynthesis Action Spectrum. The main feature of these spectra is the strong red emission peaked around 650 nm. Around and redward of this wavelength, these bulbs are more efficient than any other bulb in this sample, except the HPS. Overall the bulb that achieves the best apparent match with the action spectrum is the original GroLux. The Agrolite bulb is supposed to be a plant bulb as well, but its spectrum is typical of a Warm White Deluxe halophosphor fluorescent.

This plot (picture 5) depicts the full-spectrum fluorescents, and the next one (picture 6) depicts the halophosphor fluorescents. The GE Plant & Aquarium bulb was included here since it seems to be basically a slightly modified version of the generic Warm White Deluxe fluorescent.

This plot (picture 7) depicts the low efficiency fluorescent Philips TL950, the screw-in Ott compact fluorescent, and the halogen bulb. Note how the halogen rates poorly against everything else in this sample. The Ott compact fluorescent has a spectrum that rates quite good when compared with the rest of the sample, making it a suitable replacement for incandescent hoods.

As for the Hagen bulbs (picture 8), despite the uncertainty in the original spectra, it is possible to get some interesting conclusions from the normalized spectral plots. First, one can see that the PowerGlo bulb is just an underpowered version of the Triton bulb. Same phosphor signature (notwithstanding the shifted wavelengths) but lower efficiency. The SunGlo bulb is just a repackaged Cool White (talk about overpricing !) and the AquaGlo is a GroLux-type bulb similar to the Perfecto. The FloraGlo combines low efficiency with a spectrum almost totally devoid of blue and green, which suggests that its color rendering properties cannot be that good.

Printed on DPH with permission of Ivo Busko