 |
| rue keyenveld 48, ixelles, brussels. |
 |
| little heaven <3 |
 |
| OLD books everywhere! |
 |
| i hope old, hard bound books like these would find their way into my own library someday. |
 |
| avoc and the kind grandpa who allowed me to play with avoc :) |
::
Thursday, July 10, 2014
big dogs and bookshops
::
Saturday, July 5, 2014
optical design with ray tracing
Q1. describe how you would define the following structure on ASAP. in addition, create an ASAP script which generates the surface, the only necessary condition being n1 != n2.
A1. the provided volume could be defined using 5 surfaces: elliptical air-n1 interface, tubular air-n1 interface, elliptical n1-n2 interface, tubular air-n2 interface, and elliptical air-n2 interface. the material composition is defined by defining a database of material parameters, particularly the refractive index. the INTERFACE command is necessary to define not only the material refractive index but the reflection, transmission, and absorption properties as well.
Q2. (a script for a light pipe is provided) what is the intensity at the output of the light pipe? calculate the etendue at the entrance and end facets of the light pipe.
A1. the etendue is a pure geometric concept that does not say anything about the energy distribution of the light but determines the positional and angular extent of a bundle of light. it is important for calculating how much light can be coupled from a source to a detector theoretically. the etendue is a function of the refractive index, detector/emitter area, as well as the acceptance/exit solid angle. the light pipe exit angle could be extracted from the simulation by using the ANGLES command.
Q3. (this question was about a comparison of the simulation time between SPLIT and SPLIT MONTECARLO commands using the light pipe scripts previously provided) which simulation model would you use for optimization?
A1. (add $TIC command to script to get quantitative values of the simulation time) the SPLIT command would generate much more rays than the SPLIT MONTECARLO because it could generate more than one child rays with power distributions according to INTERFACE and MATERIAL settings. SPLIT MONTECARLO, on the other hand, generates only the child ray with the highest probability of occurrence and in fact conserves the number of rays created within the simulation. this being the case, it also follows that SPLIT MONTECARLO simulations are faster than normal SPLIT commands. for design and optimization of light pipes, since we are only interested in getting a homogeneous distribution of light at the output, using SPLIT MONTECARLO would be acceptable since it's faster and gives us the information we need, anyway. on the other hand for ghost analysis, for example, one should use the SPLIT command in order to study where unwanted rays could come from or arrive on your detector.
Q4. describe how you would design an hot mirror. no need to write a script.
A1. a hot mirror is a reflective filter, composed of layers of dielectric materials, which reflects IR light (around 750-1200 nm) and allows visible light to pass. as a starting point we can first define a low-pass filter with central wavelength around 750nm or so. this generates not a perfect low pass filter but some sort of a band pass filter with a bandwidth of about 200nm. since the range of visible light is larger (380-750nm) we need to increase the bandwidth to allow all visible light to pass. this can be done by adding another set of layers of dielectric material to create low pass filters centered on the wavelengths you wish to suppress. ripples in the spectrum may be smoothed out by optimizing the widths of the quarter-wave plates. furthermore, hot mirrors are extremely sensitive to the incident angle of light. in order to account for this sensitivity in the design (and minimize it as much as possible), one may use materials of higher refractive index.
::
A1. the provided volume could be defined using 5 surfaces: elliptical air-n1 interface, tubular air-n1 interface, elliptical n1-n2 interface, tubular air-n2 interface, and elliptical air-n2 interface. the material composition is defined by defining a database of material parameters, particularly the refractive index. the INTERFACE command is necessary to define not only the material refractive index but the reflection, transmission, and absorption properties as well.
then
SYSTEM NEW
RESET
UNITS MM
MISSED 0.5
WAVELENGTH 1550 NM
SPLIT 1
MEDIA
1.4 'N1'
MEDIA
1.5 'N2'
DIA=2
L1=2
L2=2
!! Grin substrate
SURFACE
PLANE Z 0 ELLIPSE (DIA/2) (DIA/2)
OBJECT 'N1.EDGE'
INTERFACE 0 1 N1 AIR
SURFACE
TUBE Z 0 (DIA/2) (DIA/2) (L1) (DIA/2) (DIA/2) 0
OBJECT 'N1.TUBE'
INTERFACE 0 1 N1 AIR
SURFACE
PLANE Z (L1) ELLIPSE (DIA/2) (DIA/2)
OBJECT 'N1N2INTERFACE'
INTERFACE 0 1 N1 N2
SURFACE
TUBE Z (L1) (DIA/2) (DIA/2) (L1+L2) (DIA/2) (DIA/2) 0
OBJECT 'N2.TUBE'
INTERFACE 0 1 N2 AIR
SURFACE
PLANE Z (L1+L2) ELLIPSE (DIA/2)
OBJECT 'N2.EDGE'
INTERFACE 0 1 N2 AIR
!! DETECTOR
SURFACE
PLANE Z (L1+L2+1) 0 ELLIPSE 2@((DIA)/2)
OBJECT 'DETECTOR'
!!SOURCE
WINDOW Y Z
PLOT FACETS OVERLAY
EMITTING DISK Z 0 -(DIA/2) (DIA/2) 100 5 5
!!SOURCE DIR 0 0 1
FLUX TOTAL 100
TRACE PLOT
CONSIDER ONLY DETECTOR
WINDOW X Y
!!SPOTS POS
!!PLOT FACETS OVERLAY
!!STATS
$VIEW
Q2. (a script for a light pipe is provided) what is the intensity at the output of the light pipe? calculate the etendue at the entrance and end facets of the light pipe.
A1. the etendue is a pure geometric concept that does not say anything about the energy distribution of the light but determines the positional and angular extent of a bundle of light. it is important for calculating how much light can be coupled from a source to a detector theoretically. the etendue is a function of the refractive index, detector/emitter area, as well as the acceptance/exit solid angle. the light pipe exit angle could be extracted from the simulation by using the ANGLES command.
then
SYSTEM NEW
RESET
UNITS MM 'WATT'
MEDIA
1.48 'ACRYLIC'
PIXELS 101
SPLIT 1 MONTECARLO
FRESNEL AVE
SURFACE
PLANE Z 0 ELLIPSE 1 1
OBJECT 'FACE_ENTRANCE'
INTERFACE COATING BARE_SUBSTRATE AIR ACRYLIC
SURFACE
TUBE Z 0 1 1 10 2 2 0 0
OBJECT 'TUBE'
INTERFACE COATING BARE_SUBSTRATE AIR ACRYLIC
SURFACE
PLANE Z 10 ELLIPSE 2 2
OBJECT 'FACE_EXIT'
INTERFACE COATING BARE_SUBSTRATE AIR ACRYLIC
EMITTING DISK Z -0.01 1 1 500000 90 90
FLUX TOTAL 100
WINDOW Y Z
PROFILES OVERLAY
TRACE PLOT 10000
!!TRINIDAD START
PIXELS 101
WINDOW X Y
CONSIDER ONLY FACE_EXIT
!!SELECT ONLY COS[25] C
SPOTS DIR
STATS
DISPLAY
ANGLES
PICTURE
RETURN
!!TRINIDAD END
$VIEW
Q3. (this question was about a comparison of the simulation time between SPLIT and SPLIT MONTECARLO commands using the light pipe scripts previously provided) which simulation model would you use for optimization?
A1. (add $TIC command to script to get quantitative values of the simulation time) the SPLIT command would generate much more rays than the SPLIT MONTECARLO because it could generate more than one child rays with power distributions according to INTERFACE and MATERIAL settings. SPLIT MONTECARLO, on the other hand, generates only the child ray with the highest probability of occurrence and in fact conserves the number of rays created within the simulation. this being the case, it also follows that SPLIT MONTECARLO simulations are faster than normal SPLIT commands. for design and optimization of light pipes, since we are only interested in getting a homogeneous distribution of light at the output, using SPLIT MONTECARLO would be acceptable since it's faster and gives us the information we need, anyway. on the other hand for ghost analysis, for example, one should use the SPLIT command in order to study where unwanted rays could come from or arrive on your detector.
Q4. describe how you would design an hot mirror. no need to write a script.
A1. a hot mirror is a reflective filter, composed of layers of dielectric materials, which reflects IR light (around 750-1200 nm) and allows visible light to pass. as a starting point we can first define a low-pass filter with central wavelength around 750nm or so. this generates not a perfect low pass filter but some sort of a band pass filter with a bandwidth of about 200nm. since the range of visible light is larger (380-750nm) we need to increase the bandwidth to allow all visible light to pass. this can be done by adding another set of layers of dielectric material to create low pass filters centered on the wavelengths you wish to suppress. ripples in the spectrum may be smoothed out by optimizing the widths of the quarter-wave plates. furthermore, hot mirrors are extremely sensitive to the incident angle of light. in order to account for this sensitivity in the design (and minimize it as much as possible), one may use materials of higher refractive index.
::
asap
just some cool stuff one can do with ASAP:
Dove Prism
POFcoupler1_Shao_Trinidad
POFcoupler2_Shao_Trinidad
Free-form TIR reflector_Shao_Trinidad
Tapered Light Pipe
Projector_Shao_Trinidad
GhostAnalysis
LCD Diffuser Design
and a lot of cooler ones like modeling interference..by ray tracing!
MICHELSON_INTERFEROMETER
SLITDIFFPATTERN
ProjectionEngine_Pro6_Shao_Trinidad
::
Dove Prism
POFcoupler1_Shao_Trinidad
POFcoupler2_Shao_Trinidad
Free-form TIR reflector_Shao_Trinidad
Tapered Light Pipe
 |
| view from the light sources (4 LEDs) and the homogeneous distribution of light at the tapered end. |
Projector_Shao_Trinidad
GhostAnalysis
LCD Diffuser Design
and a lot of cooler ones like modeling interference..by ray tracing!
MICHELSON_INTERFEROMETER
SLITDIFFPATTERN
ProjectionEngine_Pro6_Shao_Trinidad
 |
| red (bottom), green (left), and blue (top) light sources focused on a polarizing beam splitter designed to transmit s-polarized light. |
 |
 |
| very good homogeneity of red, green, and blue light at the detector. |
SELFOC
 |
| see the parabolic bending of light in the lens? beautiful! boyang did ask why we had to let the light propagate in free space instead of just using one GRIN lens to couple light into two fibers and i was, like, so we'd have something to optimize, of course! haha. indeed, using just one lens would be very advantageous as you can get rid of some of the alignment issues, for one. seriously, though, things like these actually exist in the market but they've fixed the alignment pretty well. |
Thursday, July 3, 2014
i think i'm drunk haha
prof. gunther was around for a while in the morning. andreas was also around for the whole day but apparently he has a girlfriend. tsktsk. i'm still allowed to like him, though, yes? but that's not why i'm drunk. haha.
we had a mini joyride with floris to het pand where the gala dinner was held. and since we arrived 30 minutes early, he treated us to a beer at his friend's bar. so we lounged for a while with satadal in front of the leiss river and just enjoyed the wonderful gent afternoon. we've been enjoying a very good weather for the duration of the summer school and i hope it will still be as good tomorrow.
anyway, floris bought us beer and apparently he suggested one with relatively high alcohol content for me. i couldn't even remember the name. during dinner they served wine then another wine then champagne. haha. soooo i think i'm..maybe not exactly drunk..tipsy. and now it's 11pm and i still need to get back to brussels and wake up early tomorrow to catch the train back here to gent. haha. i can HARDLY wait for this week to end and i am TERRIBLY tired. goodness!
i was at a table with a group of friends, some phd students and gordon, from some company in the UK, and we talked about culture and research and industry trends and how to differentiate wines and a lot of other stuff. xinyu asked about my plans after masteral and when i mentioned i'm also considering getting a phd, he was pretty surprised and said - but phd takes a long time and a lot of effort, shouldn't you get married first? hahahahahahaha. yeah thanks for the concern, but i should get a boyfriend first, though, haha.
but then again, according to him apparently a lot of the chinese phd students here found me really pretty compared to filipino celebrities, take note - celebrities, haha which is reaaaally surprising hahaha, so that's good news i suppose hahahah.
i laugh too much..i must be really tipsy..
::
Wednesday, July 2, 2014
Subscribe to:
Posts (Atom)