Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
J. A. Shaw F12
J. A. Shaw F12
Non-monochromatic conversions
When a source does not emit at a single wavelength, the spectral distribution function
must be considered when converting between radiometric and photometric values. For
example, the LTST-C190CKT LEDs spectral emission function is plotted in Figure 1 and
the photopic visual response function is plotted in Figure 2 over the same spectral range.
Figure 2. The photopic visual response function over the LEDs spectral range.
J. A. Shaw F12
The human-perceived LED output is obtained by integrating the product of these two
curves, as is shown in Figure 3. Here it is clear that the dominant wavelength for human
perception is shorter than 660 nm (it is not clear that the best dominant wavelength value
is 638 nm, which requires looking at the colorimetric coordinates of the emission on a
CIE color diagram).
Figure 3. Product of LED emission spectrum and the photopic visual response
function, showing that the human-perceived LED radiation is dominant at shorter
wavelengths than the actual LED spectral emission function (Fig. 1).
So the question naturally arises, which wavelength gives the best monochromatic
approximation of the radiometric output from the photometric data sheet value? To
answer this rigorously requires inverting the integral equation (1). However, finding the
radiometric value embedded inside of the integral from a single photometric number is
not trivial. Doing this requires at least knowing the spectral distribution of the radiometric
quantity. There are many people who make careers out of inverting integral equations to
estimate an unknown quantity from imperfect measurements of another quantity.
A non-monochromatic conversion can be performed at least approximately if we use the
known LED spectral emission function together with the known photometric luminous
intensity, as follows. Lets express the radiometric intensity as a single value I [w/sr]
multiplied by a dimensionless spectral distribution function f(), which is the curve
shown in Figure 1. In this manner we can write
(4)
I v 683I f ( )v( )d.
The values of f() and v() used to create the graphs in Figures 1, 2, and 3 are listed in
Table 1.
J. A. Shaw F12
Table 1. Values of the LED emission curve and the photopic response function
_______________________________________________________
[nm]
f()
v()
610
0
0.503
620
0.08
0.381
630
0.20
0.265
640
0.35
0.175
650
0.60
0.107
660
1.0
0.061
670
0.60
0.032
680
0.35
0.017
690
0.20
0.0082
700
0.08
0.0041
710
0
0.0021
________________________________________________________
We can use a tool like Matlab to numerically approximate the integral in eq. (4), being
careful to normalize so that the integral of f() over this spectral range gives a value of
one, allowing the actual numerical value of the LED emission to be contained within I.
i f i vi
(5)
I v 683I f ( )v( )d 683I
.
fi
i
(6)
It is useful to note that the monochromatic conversion at the peak wavelength of 660 nm
gave I = 0.48 mw/sr, while the monochromatic conversion at the dominant wavelength of
approximately 640 nm gave I = 0.17 mw/sr. So apparently the peak wavelength is the
better choice, but appears to slightly overestimate the radiometric intensity.