So it occurred to me: what effect were the HPS and CFL having on the plants? Which was most important, and what was the ideal height of the box relative to the canopy in order to maximize light use?
This was also initiated by the fact that my canopy got a little toasty in the center if I kept the light too low. My reflector is home made from a sheet of aluminum, so I know it's not great. The whole HPS with reflector cost $60 though, so no complaints.
So I decided to do an experiment to compare the two at various heights. Here are the details:
Equipment:
The systems are rated approximately equally in terms of lumens, although the CFLs pull significantly more power.
System 1:
6 x 42W CFL :: 252W total
3 cool white (at left)
3 warm (at right)
Each bulb rated at 2700 lumens: total lumen rating of ~16200 lumen
System 2:
1 150W HPS
Approximate lumen rating of 16000
Light meter:
Hydrofarm digital footcandle meter
Procedure:
The box top was mounted atop pots over a rug, which served as a map for measuring intensity at various positions. Here is a closer look at the top, and of the setup:
The box top was raised and lowered, to explore a range of heights that one might use to optimize coverage and intensity. The intensity reading in footcandles was measured at many points, by tilting the meter to maximize the intensity reading. This simulates the plant's tendency to turn toward the light. Readings taken with the meter flat showed the same trends, but are not practically relevant.
Results:
CFL Illumination:
Shown below are the intensity maps at the heights indicated on the plots, as measured by the vertical distance from the top of the meter to the edge of the spiral tube (~4,7, and 9"). Note that the vertical color scales are not the same on the first plot as the second two. Purple is 300-400 klux in the first and 150-200 klux in the second.
HPS Illumination:
Shown below are the intensity maps at the heights indicated on the plots (~7,9, and 10.5"), as measured by the vertical distance from the top of the meter to the filament:
Observations:
Optimal height:
Noon sun is 100-150 klux. With this in mind (and my toasted tops at 7"), the lights should be adjusted to a height that keeps the maximum intensity at the canopy at 300-400 or below (purple in the first CFL plot and all HPS plots).
We can see that of the heights sampled, we would choose 4" (or possibly a bit closer) for CFL and 9.5" or possibly a tad closer for HPS. The problem here is that the filament of the HPS is 2" closer to the canopy than the CFL spirals, meaning 4" for CFL means 2" for HPS, and 9.5" for HPS means 11.5" for the CFL. This is obviously a flaw in my design, and an illustration for why this need be considered when designing a system like this.
Coverage:
Assuming we choose the 4" and 9.5" separation for CFL and HPS, respectively, what kind of coverage do we get?
As you would expect when comparing a point source to a distributed source like an array of CFLs, you get much poorer coverage for a flat canopy. From the first plot, you can see that at the same height, the intensity is much more even. As you pull the lights up, the CFL distribution remains relatively even, but weakens quickly as expected. As you pull up the HID, the coverage first gets marginally better - due to the reflector getting more light to smaller angles - and then light quickly weakens with distance.
NB: The crease in the center of the plot is caused in part by the HPS reflector, which very partially obstructs the CFL/light meter line of sight when the lights are about 4" or closer.
In other words, if you want a flat canopy, the CFL setup might work better than a single small HID. On the other hand, this is easily corrected for by arranging plants of differing heights so as to even out distance from the bulb, ala continual harvest SOG.
It also wouldn't hurt to have a more efficient reflector than the approximately parabolic one I constructed. Furthermore, peripheral reflective material will assist in redistributing light in both setups, but more so in the case of CFLs, as the HID light simply does not reach the corners of the measured area.
The main point is that you need to consider the power and location of your lights when designing a grow setup.
The results also further support the fact that lumens DO ADD. If they didn't, what you would get from the CFLs is a constant intensity across the center near the bulbs that decreased as you move away.