Dear all,
A bared cylinder lamp (with 60cm length and 3.2cm diameter) is installed on
the wall and illuminance values were measured at some points and to some
directions as shown in the linked image. Unfortunately, the illuminance
values calculated with Radiance are different from measured values,
especially when the sensor is near the lamp or the incident angles are high
(K~O).
Link image <http://iga.iis.u-tokyo.ac.jp/tmp/test.gif>
What can be supposed as the cause of the discrepancy?
rtrace command was as follow.
cat pts.pts | rtrace -I -oov -ar 128 -ad 512 -as 256 -ab 2 -ds 0.02 test.oct
test.dat
and all the walls, floor and ceiling was covered with non-glossy black
paper. (reflectance 0.05)
Sung.
Hi Sung,
Try adding the following parameters to your rtrace command:
-dj .7 -dt 0
How sure are you of your surface reflectances? Did you actually measure the lamp output, or are you assuming the manufacturer's lumen specification is correct? How far off are your values?
Since you notice the most errors near to the lamp and at low angles, it's possible that the Fresnel effects are important. The emission of the fluorescent phosphor may be considered diffuse, but the actual output of the lamp will be a function of angle due to Fresnel reflection of the glass bulb. You can account for this (approximately) by applying the "winxmit.cal" function included with Radiance, like so:
void brightfunc Fresnel_effect
2 winxmit winxmit.cal
0
Fresnel_effect light lamp_light
0
3 X X X
lamp_light cylinder fluorescent_tube
(etc...)
I think you get the idea.
-Greg
···
From: Minki Sung <[email protected]>
Date: April 1, 2009 11:27:29 AM PDT
Dear all,
A bared cylinder lamp (with 60cm length and 3.2cm diameter) is installed on the wall and illuminance values were measured at some points and to some directions as shown in the linked image. Unfortunately, the illuminance values calculated with Radiance are different from measured values, especially when the sensor is near the lamp or the incident angles are high (K~O).
Link image
What can be supposed as the cause of the discrepancy?
rtrace command was as follow.
cat pts.pts | rtrace -I -oov -ar 128 -ad 512 -as 256 -ab 2 -ds 0.02 test.oct > test.dat
and all the walls, floor and ceiling was covered with non-glossy black paper. (reflectance 0.05)
Sung.
Hi Sung,
This then explains the discrepancies you were seeing with your previous comparisons between simulations (Radiance) and measurements.
You could use the new rsensor program to replicate your UV sensors response to get a match that way if you cannot modify the sensor.
-Greg
···
From: Minki Sung <[email protected]>
Date: April 6, 2009 8:40:35 AM EDT
Yes, I checked cosine sensitivity but the results showed that UV sensor did not calibrated by cosine law. I have two UV sensors made by different maker, but both showed almost same responses to incident angle. I am going to ask the maker on this problem and I think the UV sensors might not cosine calibarated with UV.
http://iga.iis.u-tokyo.ac.jp/tmp/test05.gif Fraction of value in 0 degree
http://iga.iis.u-tokyo.ac.jp/tmp/test06.gif Fraction divided by cosine
