Author Topic: EM Drive Developments - related to space flight applications - Thread 3  (Read 1875432 times)

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2335
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2982
  • Likes Given: 2601

plane).




Exy, Eyy, Ezy, Hxy, Hyy, Hzy,   (I need all 6 to compute the Poynting vector)

in each of them

Dr Rodal,
It will be quite interesting to see the differences in the poynting vectors with meep vs your other simulation runs in COMSOL.
http://forum.nasaspaceflight.com/index.php?action=dlattach;topic=36313.0;attach=827007

Offline deuteragenie

  • Member
  • Posts: 71
  • Germany
  • Liked: 22
  • Likes Given: 0
I took a quick look through the MEEP source code and didn't see any assembly optimizations. By identifying bottlenecks and hand writing SIMD assembly routines, you can often improve performance on the order of several hundred percent (it's my day job). If you folks have a set of representative input data, I'll try to take a look with a profiler in search of low-hanging optimization fruit.

Kitsuac & Meep users

to make things even more confusing, for me at least there is another version at :

https://launchpad.net/ubuntu/wily/+source/meep

[last updated on 30Jun15]

Version meep 1.3-1, maintained by "Thorsten Alteholz"

I was wondering which one to go for and try and install?

[I have in no way the math skill as demonstrated in this Thread by others but am willing to help out in doing Meep runs if asked - Two Laptops running Linux Mint].

Regards

Dear 99th ZuluMoon,

If you have some time it would be interesting to try recompiling the latest sources of Meep, document the process or otherwise automate it, then create Ubuntu packages with the binaries. I tried this for the past couple of days but I am stuck with the HDF5 dependency, because the HDF5 libraries have not been properly designed on the one hand, and because the dev team changed APIs in recent versions on the other hand, which makes Meep API incompatible with new HDF5 libraries, which forces it to use a workaround.  It would be good to upodate the Mee p sources to use the new HDF5 Api.

Only when we can re-compile Meep from its sources can we start looking at improving it (ie speed optimizations, bug fixes etc).

Only when we have clean, "click to install" packages can we expect more people to start modelling and simulating things.

Right now, because of the complexity of the setup, all of the above is a bit difficult.

Quote from: Confucius
Life is really simple, but we insist on making it complicated.


« Last Edit: 07/03/2015 12:31 PM by deuteragenie »

Offline deuteragenie

  • Member
  • Posts: 71
  • Germany
  • Liked: 22
  • Likes Given: 0
Dr. Rodal,

I have generated csv data sets for time slice 8, 9, 10, 11, 12, and 13 and uploaded them to the csv folder. Time slice 13 is the same data set as you have been working with, but now they are all named consistently and in the same place.

It only takes a few seconds for h5totxt to run, and about 2 minutes to upload each time slice data set so I spend more time fooling around structuring the folders and moving files than actually running the computer.

Enjoy.

Added: As you can see from my time numbers, I could generate and upload time slices 0 thru 7 without to much trouble. It gets easier as I do it repeatedly.

aero if I am not mistaking you were previously searching for an MPI enabled version of Meep.  Do you have access to a cluster ? Otherwise, the "serial" version of Meep appears to be multi-threaded and will run as fast if not fasted on multi-core machines. 
See here for example: http://biowulf.nih.gov/apps/meep.html

Quote from: unkwown
Direction is so much more important than speed. Many are going nowhere fast.
« Last Edit: 07/03/2015 12:34 PM by deuteragenie »

Offline VAXHeadroom

  • Full Member
  • *
  • Posts: 196
  • Whereever you go, there you are. -- BB
  • Baltimore MD
  • Liked: 259
  • Likes Given: 151
While the linked article is not exactly relevant to the em nature of what we're working on, the vortex analysis could prove useful once we can lasso this thing :)

http://www.terradaily.com/reports/The_physics_of_swimming_fish_999.html
Emory Stagmer
  Executive Producer, Public Speaker UnTied Music - www.untiedmusic.com

Offline JackFlash

  • Member
  • Posts: 7
  • League City Tx
  • Liked: 1
  • Likes Given: 19
Vax,

First let me say, awesome monnicker :D

To the point, I suspect that the emdrive probably works somewhat more like a squid than a fish, but I'd bet you're on to something here nonetheless :)


Offline VAXHeadroom

  • Full Member
  • *
  • Posts: 196
  • Whereever you go, there you are. -- BB
  • Baltimore MD
  • Liked: 259
  • Likes Given: 151
Vax,

First let me say, awesome monnicker :D

To the point, I suspect that the emdrive probably works somewhat more like a squid than a fish, but I'd bet you're on to something here nonetheless :)
Thanks on both counts :)
I try to aggregate diverse, apparently unrelated information all the time.  Sooner or later it all becomes interrelated :) I said to a friend several years ago that I believed geometry is actually all important, and this investigation has only made even more convinced of that.
I'm also a serious space geek.  This was linked at www.spacedaily.com, one of about 8 space related websites I read every day.
Emory Stagmer
  Executive Producer, Public Speaker UnTied Music - www.untiedmusic.com

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2335
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2982
  • Likes Given: 2601
Sigh... This is like waiting for the 4th of July fireworks.

Shell

Offline rfmwguy

  • EmDrive Builder (retired)
  • Senior Member
  • *****
  • Posts: 2165
  • Liked: 2681
  • Likes Given: 1124
This is awesome collaboration on software modeling, guys and gals. Congratulations, its not easy to lose me, technically, but you've all done a fine job. One small favor to help me keep track, can you label any frustum analysis pics or movies by a common name? For example, if its the 11.0x6.25x10.2L (mine), use the label NSF-1701. If its 9.0L, use Brady/ShawyerJulian or whatever everyone thinks is appropriate. Think Shell will need a nickname on her newfangled frustum as well (get busy shell).

Thanks for the consideration and happy 4th to all...

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
Dr. Rodal,

I have generated csv data sets for time slice 8, 9, 10, 11, 12, and 13 and uploaded them to the csv folder. Time slice 13 is the same data set as you have been working with, but now they are all named consistently and in the same place.

It only takes a few seconds for h5totxt to run, and about 2 minutes to upload each time slice data set so I spend more time fooling around structuring the folders and moving files than actually running the computer.

Enjoy.

Added: As you can see from my time numbers, I could generate and upload time slices 0 thru 7 without to much trouble. It gets easier as I do it repeatedly.

OK I got some more interesting results.  In order to avoid people getting too excited, we should double check this to make sure, and get at least full cycle before posting the results.  (right now we have  60% of a cycle)

Could you please add 4 or 5 more immediately preceding time steps so that we have at least a total of 10 or 11 time steps so that we have a FULL cycle to show please?

TS03
TS04  (this will give us a full cycle)
TS05
TS06
TS07

__________

PS1: I fully agree with RFMWGUY: we should have a Naming convention for the files, containing the dimensions of the bases, the length, and the excitation frequency at a minimum for each file

PS2: this is an amazing demonstration of what is possible to accomplish nowadays with personal computing and cooperation through the Internet.  Nobody to my knowledge had performed this deep and complete an analysis of the EM Drive,  nobody had performed a transient analysis, and or showed the Poynting vector field with the RF feed on and losses modeled in the copper.  Certainly not Shawyer or Yang or NASA.   Greg Egan just modeled the standing waves, frozen in space, without taking into account the RF feed.  Many "theories" that have been proposed so far are like "sketches" on a napkin: with drastic simplifications of what is going on.

« Last Edit: 07/03/2015 02:27 PM by Rodal »

Offline aero

  • Senior Member
  • *****
  • Posts: 2785
  • 92129
  • Liked: 724
  • Likes Given: 249
Dr. Rodal,

I have generated csv data sets for time slice 8, 9, 10, 11, 12, and 13 and uploaded them to the csv folder. Time slice 13 is the same data set as you have been working with, but now they are all named consistently and in the same place.

It only takes a few seconds for h5totxt to run, and about 2 minutes to upload each time slice data set so I spend more time fooling around structuring the folders and moving files than actually running the computer.

Enjoy.

Added: As you can see from my time numbers, I could generate and upload time slices 0 thru 7 without to much trouble. It gets easier as I do it repeatedly.

aero if I am not mistaking you were previously searching for an MPI enabled version of Meep.  Do you have access to a cluster ? Otherwise, the "serial" version of Meep appears to be multi-threaded and will run as fast if not fasted on multi-core machines. 
See here for example: http://biowulf.nih.gov/apps/meep.html

Quote from: unkwown
Direction is so much more important than speed. Many are going nowhere fast.

No, I hope I typed MPB enabled version. I have been running meep-mpi since almost day one. Multi processor interface programs are not to difficult.

MIT Photonic-Bands package, MPB, is their eigensolver which can be called from Meep when properly configured. That would allow me to get away from such heavy reliance on Harminv. Here:
http://ab-initio.mit.edu/wiki/index.php/MPB_User_Reference
and here:http://ab-initio.mit.edu/wiki/index.php/MIT_Photonic_Bands
« Last Edit: 07/03/2015 02:40 PM by aero »
Retired, working interesting problems

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2335
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2982
  • Likes Given: 2601
This is awesome collaboration on software modeling, guys and gals. Congratulations, its not easy to lose me, technically, but you've all done a fine job. One small favor to help me keep track, can you label any frustum analysis pics or movies by a common name? For example, if its the 11.0x6.25x10.2L (mine), use the label NSF-1701. If its 9.0L, use Brady/ShawyerJulian or whatever everyone thinks is appropriate. Think Shell will need a nickname on her newfangled frustum as well (get busy shell).

Thanks for the consideration and happy 4th to all...
I'm being busy and it's going well. Been doing the extra drawings I need and getting all the hardware, wires, SS cables, lasers, even graph paper, ordered the Perforated copper, finished the design for 2 antennas, looking into parts and layout to control the frequency stability of the magnetron (if needed), got the composite beam done and may redo it... I've a friend who teaches woodwork and has a great shop offer to build me another much lighter and stronger oh and the last thing I think is I found a magnetron but still looking for a spare.

You think Warp-Shell is a little much for a name? The tests on the cavity in meep for my design have been done in basic copper but what would make me feel better is seeing the helical 2.45Ghz antenna in a hexagon perforated copper shape and meep as it sits can't do it. Trust me I understand Aero you have done a great job, more than great but no problemo, I have confidence.


Shell

Offline aero

  • Senior Member
  • *****
  • Posts: 2785
  • 92129
  • Liked: 724
  • Likes Given: 249
This is awesome collaboration on software modeling, guys and gals. Congratulations, its not easy to lose me, technically, but you've all done a fine job. One small favor to help me keep track, can you label any frustum analysis pics or movies by a common name? For example, if its the 11.0x6.25x10.2L (mine), use the label NSF-1701. If its 9.0L, use Brady/ShawyerJulian or whatever everyone thinks is appropriate. Think Shell will need a nickname on her newfangled frustum as well (get busy shell).

Thanks for the consideration and happy 4th to all...
I'm being busy and it's going well. Been doing the extra drawings I need and getting all the hardware, wires, SS cables, lasers, even graph paper, ordered the Perforated copper, finished the design for 2 antennas, looking into parts and layout to control the frequency stability of the magnetron (if needed), got the composite beam done and may redo it... I've a friend who teaches woodwork and has a great shop offer to build me another much lighter and stronger oh and the last thing I think is I found a magnetron but still looking for a spare.

You think Warp-Shell is a little much for a name? The tests on the cavity in meep for my design have been done in basic copper but what would make me feel better is seeing the helical 2.45Ghz antenna in a hexagon perforated copper shape and meep as it sits can't do it. Trust me I understand Aero you have done a great job, more than great but no problemo, I have confidence.


Shell

"Meep Can't" is a little strong.

Right, those are not Meep default options but the cavity could be pieced together, holes and all, and I'd bet that a helical antenna could be modelled as a helix of point sources around a dielectric cylinder, but getting the phasing right between point sources is the part that's tricky.
Retired, working interesting problems

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2335
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2982
  • Likes Given: 2601
This is awesome collaboration on software modeling, guys and gals. Congratulations, its not easy to lose me, technically, but you've all done a fine job. One small favor to help me keep track, can you label any frustum analysis pics or movies by a common name? For example, if its the 11.0x6.25x10.2L (mine), use the label NSF-1701. If its 9.0L, use Brady/ShawyerJulian or whatever everyone thinks is appropriate. Think Shell will need a nickname on her newfangled frustum as well (get busy shell).

Thanks for the consideration and happy 4th to all...
I'm being busy and it's going well. Been doing the extra drawings I need and getting all the hardware, wires, SS cables, lasers, even graph paper, ordered the Perforated copper, finished the design for 2 antennas, looking into parts and layout to control the frequency stability of the magnetron (if needed), got the composite beam done and may redo it... I've a friend who teaches woodwork and has a great shop offer to build me another much lighter and stronger oh and the last thing I think is I found a magnetron but still looking for a spare.

You think Warp-Shell is a little much for a name? The tests on the cavity in meep for my design have been done in basic copper but what would make me feel better is seeing the helical 2.45Ghz antenna in a hexagon perforated copper shape and meep as it sits can't do it. Trust me I understand Aero you have done a great job, more than great but no problemo, I have confidence.


Shell

"Meep Can't" is a little strong.

Right, those are not Meep default options but the cavity could be pieced together, holes and all, and I'd bet that a helical antenna could be modelled as a helix of point sources around a dielectric cylinder, but getting the phasing right between point sources is the part that's tricky.

Sorry, it was too strong, of course it can be. 

It seems like so much work and right now Dr. Rodal has you hopping and that's a huge priority for all of us here.

Also after much deliberation and burning the midnight oil I'm going with these dims on the extended split shape. I was going to go with RS's and TT's but the cone angle isn't right for what I need to do and what I expect the cavity to be doing either.

Warp-shell
Length         9.4488
Large Diameter 7.9134
Small Diameter 5.874
All in inches
2.45 GHz magnetron

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
...The tests on the cavity in meep for my design have been done in basic copper but what would make me feel better is seeing the helical 2.45Ghz antenna in a hexagon perforated copper shape and meep as it sits can't do it. Trust me I understand Aero you have done a great job, more than great but no problemo, I have confidence.


Shell
Modeling the hexagonal shape of the copper boundaries is much, less, less, less important than modeling the geometry of the antenna.   One can show that the fields want to be spherical, the field will accommodate to an equivalent circle that is half-way in between the inscribed and circumscribed circles of the hexagon cross-section.

The difference between the inscribed and circumscribed circles is small, and therefore one can show that the effect of the hexagon is small and negligible.

On the other hand the RF feed has a paramount effect, as it is the only thing that can give thrust. Experimentally we know that with the RF feed OFF there is no thrust.  Theoretically we know that with the RF feed OFF all you have are standing waves in a cavity, with a Poyinting vector averaging zero over integer number of periods.

The antenna shape and position is most important.  If Meep modeling time would go into this should be in modeling a helical antenna inside a circular frustum, as the hexagonal shape has, in comparison, an insignificant effect.  :)

Whether the helical antenna shape is an improvement or not over the dipole antenna remains to be modeled and shown (*), but prior to that we need to arrive at a way to verify thrust (or something "pointing towards" it -pun intended-) in modeling with Meep:

1) by calculating the Poynting vector (being done)

2) by calculating the Meep force at each time step (remains to be done).  Also, whether the force calculation is applicable, is debatable, as it assumes conservation of photons.  We may need to calculate the force from integration from Maxwell's stress tensor instead.

___________

(*) The dipole antenna is producing almost a flat excitation while the helical should produce a 3D helix, which should be superior -in theory- but is it too long? how should it be optimized?
« Last Edit: 07/03/2015 03:20 PM by Rodal »

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2335
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2982
  • Likes Given: 2601
...The tests on the cavity in meep for my design have been done in basic copper but what would make me feel better is seeing the helical 2.45Ghz antenna in a hexagon perforated copper shape and meep as it sits can't do it. Trust me I understand Aero you have done a great job, more than great but no problemo, I have confidence.


Shell
Modeling the hexagonal shape of the copper boundaries is much, less, less, less important than modeling the geometry of the antenna.   One can show that the fields want to be spherical, the field will accommodate to an equivalent circle that is half-way in between the inscribed and circumscribed circles of the hexagon cross-section.

The difference between the inscribed and circumscribed circles is small, and therefore one can show that the effect of the hexagon is small and negligible.

On the other hand the RF feed has a paramount effect, as it is the only thing that can give thrust. Experimentally we know that with the RF feed OFF there is no thrust.  Theoretically we know that with the RF feed OFF all you have are standing waves in a cavity, with a Poyinting vector averaging zero over integer number of periods.

The antenna shape and position is most important.  If Meep modeling time would go into this should be in modeling a helical antenna inside a circular frustum, as the hexagonal shape has, in comparison, an insignificant effect.  :)

Whether the helical antenna shape is an improvement or not over the dipole antenna remains to be modeled and shown (*), but prior to that we need to arrive at a way to verify thrust (or something "pointing towards" it -pun intended-) in modeling with Meep:

1) by calculating the Poynting vector (being done)

2) by calculating the Meep force at each time step (remains to be done).  Also, whether the force calculation is applicable, is debatable, as it assumes conservation of photons.  We may need to calculate the force from integration from Maxwell's stress tensor instead.

___________

(*) The dipole antenna is producing almost a flat excitation while the helical should produce a 3D helix, which should be superior -in theory- but is it too long? how should it be optimized?

It is the 3D helix symmetrical pattern with ~ 8db of forward gain with a wide frequency response. That is the shape that's needed and the lobes are very minimal on the back side of the pattern. I was going to place this on the small plate but I'm flip flopping until I see the poynting vectors you publish and it may go into the large end plate to use as a ground plane.

A quick calculator for size and care must be taken in the wire size as it's quite important.
http://jcoppens.com/ant/qfh/calc.en.php

I don't think it's too long at just under 2 inches or 48.4 mm 2450 MHz as it will be pointing down the length long axis between the plates.

Edit: added a little clarity.
« Last Edit: 07/03/2015 03:39 PM by SeeShells »

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
...
A quick calculator for size and care must be taken in the wire size as it's quite important.
http://jcoppens.com/ant/qfh/calc.en.php

I don't think it's too long at just under 2 inches or 48.4 mm 2450 MHz as it will be pointing down the length long axis between the plates.

Edit: added a little clarity.
I admit I thought it was longer than 2 inches based on (admittedly bad) guesstimating from the picture.  I think what matters is not the free space wavelength but the length of a wave-pattern inside the frustum, where the antenna is located:  what matters is the following ratio
 
(Length of Antenna)/(Length of wave-pattern inside the frustum where antenna is located)

The length of wave-pattern inside the frustum where antenna is located depends on the shape of the frustum and the "p" of the mode shape TEmnp or TMmnp, and the position of the antenna, as the wave-pattern length usually (not always) gets longer towards the apex of the cone.
« Last Edit: 07/03/2015 03:56 PM by Rodal »

Offline ZuluMoon99

  • Member
  • Posts: 11
  • UK
  • Liked: 7
  • Likes Given: 35


Dear 99th ZuluMoon,

If you have some time it would be interesting to try recompiling the latest sources of Meep, document the process or otherwise automate it, then create Ubuntu packages with the binaries. I tried this for the past couple of days but I am stuck with the HDF5 dependency, because the HDF5 libraries have not been properly designed on the one hand, and because the dev team changed APIs in recent versions on the other hand, which makes Meep API incompatible with new HDF5 libraries, which forces it to use a workaround.  It would be good to upodate the Mee p sources to use the new HDF5 Api.

Only when we can re-compile Meep from its sources can we start looking at improving it (ie speed optimizations, bug fixes etc).

Only when we have clean, "click to install" packages can we expect more people to start modelling and simulating things.

Right now, because of the complexity of the setup, all of the above is a bit difficult.

Quote from: Confucius
Life is really simple, but we insist on making it complicated.
[/quote]

Hi deuteragenie

Will look into this and give it a try - not promising anything but will definitely give it a bash.

Will let the forum know how I get on when I make some kind of progress and will document as requested.

Kind Regards

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2335
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2982
  • Likes Given: 2601
...
A quick calculator for size and care must be taken in the wire size as it's quite important.
http://jcoppens.com/ant/qfh/calc.en.php

I don't think it's too long at just under 2 inches or 48.4 mm 2450 MHz as it will be pointing down the length long axis between the plates.

Edit: added a little clarity.
I admit I thought it was longer than 2 inches based on (admittedly bad) guesstimating from the picture.  I think what matters is not the free space wavelength but the length of a wave-pattern inside the frustum, where the antenna is located:  what matters is the following ratio
 
(Length of Antenna)/(Length of wave-pattern inside the frustum where antenna is located)

The length of wave-pattern inside the frustum where antenna is located depends on the shape of the frustum and the "p" of the mode shape TEmnp or TMmnp, and the position of the antenna, as the wave-pattern length usually (not always) gets longer towards the apex of the cone.
Considering it will be TE012 (see poynting vector images) from COMSOL I went from the small endplate to seriously considering the larger but i wanted to verify that move with what you will see with the poynting data from meep. I would like for them to agree. Then running in meep both ways will solidify placement.

Offline lmbfan

  • Member
  • Posts: 55
  • Phoenix
  • Liked: 42
  • Likes Given: 42
This is awesome collaboration on software modeling, guys and gals. Congratulations, its not easy to lose me, technically, but you've all done a fine job. One small favor to help me keep track, can you label any frustum analysis pics or movies by a common name? For example, if its the 11.0x6.25x10.2L (mine), use the label NSF-1701. If its 9.0L, use Brady/ShawyerJulian or whatever everyone thinks is appropriate. Think Shell will need a nickname on her newfangled frustum as well (get busy shell).

Thanks for the consideration and happy 4th to all...
I'm being busy and it's going well. Been doing the extra drawings I need and getting all the hardware, wires, SS cables, lasers, even graph paper, ordered the Perforated copper, finished the design for 2 antennas, looking into parts and layout to control the frequency stability of the magnetron (if needed), got the composite beam done and may redo it... I've a friend who teaches woodwork and has a great shop offer to build me another much lighter and stronger oh and the last thing I think is I found a magnetron but still looking for a spare.

You think Warp-Shell is a little much for a name? The tests on the cavity in meep for my design have been done in basic copper but what would make me feel better is seeing the helical 2.45Ghz antenna in a hexagon perforated copper shape and meep as it sits can't do it. Trust me I understand Aero you have done a great job, more than great but no problemo, I have confidence.


Shell

"Meep Can't" is a little strong.

Right, those are not Meep default options but the cavity could be pieced together, holes and all, and I'd bet that a helical antenna could be modelled as a helix of point sources around a dielectric cylinder, but getting the phasing right between point sources is the part that's tricky.

Sorry, it was too strong, of course it can be. 

It seems like so much work and right now Dr. Rodal has you hopping and that's a huge priority for all of us here.

Also after much deliberation and burning the midnight oil I'm going with these dims on the extended split shape. I was going to go with RS's and TT's but the cone angle isn't right for what I need to do and what I expect the cavity to be doing either.

Warp-shell
Length         9.4488
Large Diameter 7.9134
Small Diameter 5.874
All in inches
2.45 GHz magnetron

Just last night I got a hexagonal frustum modeled to my satisfaction.  I have plugged in these numbers, plus .125" for the wall thickness and attached a .ctl file as a .txt file.  The frequency, material, and scale are completely arbitrary at the moment, just something to test the geometry.  It also lacks end caps, which should be relatively easy to add on.  I added some comments and removed some dead code just now (I was planning on posting tomorrow, but carpe diem), and since I didn't test it, be prepared to have at least one syntax error (with my luck).

No perforations are modeled, in theory they should not affect the analysis, correct?

-=EDIT=-

Modeling the hexagonal shape of the copper boundaries is much, less, less, less important than modeling the geometry of the antenna.   One can show that the fields want to be spherical, the field will accommodate to an equivalent circle that is half-way in between the inscribed and circumscribed circles of the hexagon cross-section.

The difference between the inscribed and circumscribed circles is small, and therefore one can show that the effect of the hexagon is small and negligible.

On the other hand the RF feed has a paramount effect, as it is the only thing that can give thrust. Experimentally we know that with the RF feed OFF there is no thrust.  Theoretically we know that with the RF feed OFF all you have are standing waves in a cavity, with a Poyinting vector averaging zero over integer number of periods.

The antenna shape and position is most important.  If Meep modeling time would go into this should be in modeling a helical antenna inside a circular frustum, as the hexagonal shape has, in comparison, an insignificant effect.  :)

NOW you tell me.

Meep does not support arbitrary geometry for sources (antennas) as far as I can determine, just a point/line/plane/cube.  Does anyone have an intuition as to what would be a good way to model the helix?  I was thinking of short line segments, but I don't know if the line sources in Meep can be skewed or slanted.  The custom-src function looks promising.  I may have time this weekend to look into that.

-=EDIT 2=-

Ugh, I fat fingered the dimensions.  Lines 8-11 should read:

(define hex-small 5.874) ; radius of the small hex end of the cavity (circumscribed)
(define hex-large 7.9134) ; radius of the large hex end of the cavity
(define hex-h 9.4488) ; height of the hex cavity
(define hex-wall .125) ; thickness of the cavity wall.  This is currently limited to a maximum of (hex-large - hex-small)
« Last Edit: 07/03/2015 04:49 PM by lmbfan »

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5895
  • USA
  • Liked: 6045
  • Likes Given: 5325
...
A quick calculator for size and care must be taken in the wire size as it's quite important.
http://jcoppens.com/ant/qfh/calc.en.php

I don't think it's too long at just under 2 inches or 48.4 mm 2450 MHz as it will be pointing down the length long axis between the plates.

Edit: added a little clarity.
I admit I thought it was longer than 2 inches based on (admittedly bad) guesstimating from the picture.  I think what matters is not the free space wavelength but the length of a wave-pattern inside the frustum, where the antenna is located:  what matters is the following ratio
 
(Length of Antenna)/(Length of wave-pattern inside the frustum where antenna is located)

The length of wave-pattern inside the frustum where antenna is located depends on the shape of the frustum and the "p" of the mode shape TEmnp or TMmnp, and the position of the antenna, as the wave-pattern length usually (not always) gets longer towards the apex of the cone.
Considering it will be TE012 (see poynting vector images) from COMSOL I went from the small endplate to seriously considering the larger but i wanted to verify that move with what you will see with the poynting data from meep. I would like for them to agree. Then running in meep both ways will solidify placement.

Hi Shell,

You are amazing.  How did you get those images?

They are Poynting vector field images I calculated based on a program I wrote using Mathematica, an exact solution using Associated Legendre functions and Spherical Bessel Functions.  They are not calculated using COMSOL FEA.

I also wrote the computer program to display the Poynting vector field.

Since they are based on an exact solution for standing waves inside the frustum, the Poynting vector averages zero over an integer number of periods.  (It fluctuates like a sine, pointing in one direction over 1/2 a period, and pointing in the opposite direction over the next 1/2 period).

I can already tell you that the Poynting vector field using the exact solution based on standing waves does NOT agree with Meep, because Meep takes into account the RF feed antenna, which the exact solution does NOT.  The antenna disturbs the Poynting vector field:

1) It changes its direction around the antenna, this impacts the whole field

2) It changes its time average, which is not zero anymore.
« Last Edit: 07/03/2015 04:35 PM by Rodal »

Tags: