-
#520 by aero on 14 Aug, 2015 20:56
-
Quote
Source excitation magnetic component Hz. Yes
Run length - 64 full cycles Yes
Time slices output for each field component:
every 1/10-th cycle from 62.7 cycles to 64 cycles (end of run) Yes
run 0 finished at t = 13.054 (6527 timesteps) - No -
run time set to 26.104377295238095 meep time set
run 0 finished at t = 26.106 (13053 timesteps) Yes -
#521 by Rodal on 14 Aug, 2015 21:25
-
don't know how helpful this is but i do search at arxiv.org for "tapered waveguide" ...
find this for exampe:
http://arxiv.org/pdf/1001.1254v1.pdf
and many many many more...
Only the whole community is able to review all of this
Thank you. I had not seen this paper.
This paper is very, very important as it answers a question that has repeatedly been asked in this forum and in Reddit:
what is the best function to use for the taper?
Shawyer used a linear taper (conical) but lacked the resources to explore other geometries (either numerically or by experiment)
Other researchers have not explored other taper geometries either
Musical people usually come with this question as to why don't people use musical horn shapes for the EM Drive. Invariably their question is answered by saying that the frequencies of the EM drive are much higher than the acoustical frequencies of French Horns, but the question of optimal shape is not answered.
The authors of this paper answer by saying:QuoteBest tradeoff is obtained with the exponential formulation with a slow variation of the
material parameters
that the best taper is exponential instead of linear (conical). However they do not use the same formulation of the problem as they don't of course deal with "thrust" (if there is such a thing) and they consider a simultaneous variation of material properties. -
#522 by rfmwguy on 14 Aug, 2015 21:43
-
Doc, have a bit of news for you here. Because of the flexibility of the copper mesh, I can "compress" it with a band to form an exponential taper. Not a lot, but some.don't know how helpful this is but i do search at arxiv.org for "tapered waveguide" ...
find this for exampe:
http://arxiv.org/pdf/1001.1254v1.pdf
and many many many more...
Only the whole community is able to review all of this
Thank you. I had not seen this paper.
This paper is very, very important as it answers a question that has repeatedly been asked in this forum and in Reddit:
what is the best function to use for the taper?
Shawyer used a linear taper (conical) but lacked the resources to explore other geometries (either numerically or by experiment)
Other researchers have not explored other taper geometries either
Musical people usually come with this question as to why don't people use musical horn shapes for the EM Drive. Invariably their question is answered by saying that the frequencies of the EM drive are much higher than the acoustical frequencies of French Horns, but the question of optimal shape is not answered.
The authors of this paper answer by saying:QuoteBest tradeoff is obtained with the exponential formulation with a slow variation of the
material parameters
that the best taper is exponential instead of linear (conical). However they do not use the same formulation of the problem as they don't of course deal with "thrust" (if there is such a thing) and they consider a simultaneous variation of material properties.
Now, if I just knew how much and where
-
#523 by deltaMass on 14 Aug, 2015 22:06
-
Hmm - Adjustapalooza. 6 bands at 4 adjusting screws per band plus 3 screws for impedance matching plus end plate screw. 28 screws and you're set
In fact, compared with the typical complexity of an optical bench for quantum experiments, this is child's play. -
#524 by RotoSequence on 14 Aug, 2015 22:16
-
Doc, have a bit of news for you here. Because of the flexibility of the copper mesh, I can "compress" it with a band to form an exponential taper. Not a lot, but some.
Now, if I just knew how much and where
Assuming all of this is real, it's starting to sound like the optimal design is a convergence of Shawyer's EM drive and the Cannae device. What are the odds?
-
#525 by Josave on 14 Aug, 2015 22:19
-
Another geometry is proposed in http://arxiv.org/pdf/math-ph/0408055v1.pdf . Picture is attached.
This spheroid must be moving to create wavelets, that unlike sinusoids, have much more frequency components. Maybe this is a way to allow in McCulloch terms "the number of allowed Unruh waves increases..." in the "...wider side of the cavity".
These are my speculations, but the key in this article is that a moving spheroid geometry can lead to interesting field distributions.
-
#526 by WarpTech on 14 Aug, 2015 22:33
-
...Sully positioned his more near the small end. So at full small-end extension, the waveguide will be roughly in the middle of the two end plates....
From what we are learning from Meep, the best place to place the RF feed is near the big base, not at the middle and not at the small end. I initially thought it was better to have the antenna near the small end (based on several arguments). I have changed my mind based on the facts presented by multiple Meep runs.
Why? Do you believe it is better for resonance, or better for thrust? It seems counter intuitive because the big end will be reflecting more, not less.
Todd
-
#527 by deltaMass on 14 Aug, 2015 22:51
-
The configuration space is massive. Take, for example, a tuba mounted on a blender
-
#528 by TheTraveller on 14 Aug, 2015 23:03
-
...Sully positioned his more near the small end. So at full small-end extension, the waveguide will be roughly in the middle of the two end plates....
From what we are learning from Meep, the best place to place the RF feed is near the big base, not at the middle and not at the small end. I initially thought it was better to have the antenna near the small end (based on several arguments). I have changed my mind based on the facts presented by multiple Meep runs.
Why? Do you believe it is better for resonance, or better for thrust? It seems counter intuitive because the big end will be reflecting more, not less.
Todd
Assuming the antenna is 1/4 guide wavelength away from the end plate, the reflected EM wave is 180 deg out of phase with the radiating antenna. The big end plate becomes a reflector element in a 2 element array. This will cause more of the antenna's radiated energy to be directed at the small end than at the big end as the antenna is now directional.
http://www.ph.surrey.ac.uk/satellites/main/assets/schoolzone/project1/reflectors_directors.htm
Suspect this may also cause an out of phase shadow zone on the big end plate, reducing bounce Force inside the shadow zone due to phase distortion between the radiating antenna and the EM wave propagating from the small end.. -
#529 by TheTraveller on 14 Aug, 2015 23:18
-
Interesting Reddit post.
micronewton electromagnetic thruster
https://www.reddit.com/r/EmDrive/comments/3h15k6/a_couple_of_interesting_links_not_emdrive_related/ -
#530 by Rodal on 14 Aug, 2015 23:48
-
Excellent analysis...Sully positioned his more near the small end. So at full small-end extension, the waveguide will be roughly in the middle of the two end plates....
From what we are learning from Meep, the best place to place the RF feed is near the big base, not at the middle and not at the small end. I initially thought it was better to have the antenna near the small end (based on several arguments). I have changed my mind based on the facts presented by multiple Meep runs.
Why? Do you believe it is better for resonance, or better for thrust? It seems counter intuitive because the big end will be reflecting more, not less.
Todd
Assuming the antenna is 1/4 guide wavelength away from the end plate, the reflected EM wave is 180 deg out of phase with the radiating antenna. The big end plate becomes a reflector element in a 2 element array. This will cause more of the antenna's radiated energy to be directed at the small end than at the big end as the antenna is now directional.
http://www.ph.surrey.ac.uk/satellites/main/assets/schoolzone/project1/reflectors_directors.htm
Suspect this may also cause an out of phase shadow zone on the big end plate, reducing bounce Force inside the shadow zone due to phase distortion between the radiating antenna and the EM wave propagating from the small end.. -
#531 by lmbfan on 15 Aug, 2015 04:56
-
Attached is the update to the python script that generates the csv and png files. The current iteration runs the h5totxt command and captures the output. The output is parsed, and the minimum and maximum values for that field and axis are stored. After all csv files are created, the stored values are consolidated into the fields, E and H, to produce a final min and max value for each field. These values are then used to generate the png files. The end result is that all E fields will have the same min and max values, as will all the H fields.
The min and max values are collected in each direction specified. Therefore, if only z slices are output, only the min and max for those slices will be considered, and the min and max may be different if x or y slices are also included in the run. The upshot is, each set of slices should be re-run to get the full range. The values are collected over time slices, so the min and max output at t=0 will be the same as at t=13.
Using the -h flag will list all the other flags that can be used. If no x, y, or z flags are used, nothing will be output. The special small and big end flags (-s/--z-small-end, -b/--z-big-end) will use the output directories Small-end-z-views and Big-end-z-views for the png files. The x, y, and z flags put the png in x-views, y-views, z-views. The special flags -x0, -y0, and -z0 will put them in Center-x-views, etc. These directories must be pre-created. The required flag -d/--csv-directory is the directory the csv files will be created in, and must also be pre-created. Having these directories created if they do not already exist is the next improvement on the list.
-
#532 by X_RaY on 15 Aug, 2015 05:22
-
Be careful with that, the eigenmodes in your frustum will get a shift to higher frequencies if you compress the side walls to a lower diameter.
Doc, have a bit of news for you here. Because of the flexibility of the copper mesh, I can "compress" it with a band to form an exponential taper. Not a lot, but some.don't know how helpful this is but i do search at arxiv.org for "tapered waveguide" ...
find this for exampe:
http://arxiv.org/pdf/1001.1254v1.pdf
and many many many more...
Only the whole community is able to review all of this
Thank you. I had not seen this paper.
This paper is very, very important as it answers a question that has repeatedly been asked in this forum and in Reddit:
what is the best function to use for the taper?
Shawyer used a linear taper (conical) but lacked the resources to explore other geometries (either numerically or by experiment)
Other researchers have not explored other taper geometries either
Musical people usually come with this question as to why don't people use musical horn shapes for the EM Drive. Invariably their question is answered by saying that the frequencies of the EM drive are much higher than the acoustical frequencies of French Horns, but the question of optimal shape is not answered.
The authors of this paper answer by saying:QuoteBest tradeoff is obtained with the exponential formulation with a slow variation of the
material parameters
that the best taper is exponential instead of linear (conical). However they do not use the same formulation of the problem as they don't of course deal with "thrust" (if there is such a thing) and they consider a simultaneous variation of material properties.
Now, if I just knew how much and where -
#533 by SteveD on 15 Aug, 2015 05:29
-
I keep thinking about the idea I had a few days ago and pushed it further. I thought about Rodal and WarpTech's advice of increasing cone angle and keeping the apex as near as possible of the small end. I also kept Shawyer's high Df and concentric spherical ends to maintain a high Q, and I've ended up with this wide and shallow resonant design, with incredible large spherical ends, making the frustum almost a half-sphere:
Db = 600 mm
Ds = 150 mm
L = 51.20 mm
r1 = 76.92 mm
r2 = 307.67 mm
r2-r1 = 230.75 mm
θ = 77.18°
Resonance at 2.45 GHz in TE013 mode, Df = 0.96
Would be a challenge to build but nevertheless interesting to test. What could be the Q of such a cavity?
I might be able to model it. Can you export an STL of the part?
Although SketchUp is not good at all with circles (it draws them as a series of segments) I updated the number of segments per circle from 24 to 240 in order to improve precision.
I don't know if you need the 2D plan or the 3D modeled object, so I created both versions and exported them in STL format. You can find them zipped below.
This is a very interesting shape to test the outer limits of the theories and methods involved.
Due to the extreme spherical conical shape of this cavity, the limitations of the spreadsheet approach (that in a kludgy way intends to model a spherical cone as a large series of cylindrical waveguides) is more crudely exposed:
the natural frequency of mode TE013 is 2.132 GHz (instead of 2.45 GHz), a difference of 15% in frequency (for cone angles of 15 degrees the spreadsheet is 1 to 2% different from the exact solution)
It does resonate, and it resonates well:
theoretical Q = 94,254
using a resistivity = 1.678*10^(-8)(*copper*) (Q will go down with increasing resitivity materials and geometrical imperfections)
although this is not much more than the Q calculated for the 30 degrees cavity, so it looks like there are diminishing returns after 30 degrees
I attach below the contour plots for
1) the magnetic field in the spherical radial direction
2) the electric field in the azimuthal circumferential direction
Note how distorted is the magnetic field in the spherical radial direction
Um doc, didn't the South African science fair experiment show greater thrust by increasing the size of the cavity? Seems like that would indicate that volume of the cavity and not just Q is a factor. -
#534 by Rodal on 15 Aug, 2015 12:02
-
...Um doc, didn't the South African science fair experiment show greater thrust by increasing the size of the cavity? Seems like that would indicate that volume of the cavity and not just Q is a factor.
1) Volume of the cavity was taken into account in the calculations: one cannot calculate the natural frequency and mode shape without knowing the dimensions. To calculate the quality factor Q one needs to perform an integration of the electromagnetic fields over the whole volume. Thus the Q is very much related to the volume.
2) Very little is known about the South African experiment. Important variables were not measured for the South African experiment, for example the Q was not measured or reported. We don't know whether the cavity was at a higher amplitude resonance by extending the cavity length and hence whether the Q was higher by extending the cavity length, purely due to resonance. -
#535 by Rodal on 15 Aug, 2015 16:21
-
Here are the force calculations corresponding to the stresses shown in http://forum.nasaspaceflight.com/index.php?topic=38203.msg1416281#msg1416281, for a total run time of 0.026 microseconds with the RF feed on.
The net force (shown in green) is the vector addition of the forces on both ends.
Some noteworthy comments:
1) The force magnitude is a whooping 10,000 times higher than for previous cases. At this point we don't know how much of this greater magnitude is due to the fact that this computer run is for twice as long a time as previous runs (with stresses that are increasing with time) and how much is due to the fact that this force is due to a transverse electric (TE) mode shape while the other ones were for transverse magnetic (TM) modes.
2) Due to the fact that the stress is tensile at the small base and compressive at the big base, both forces, at the small base and at the big base point in the same direction, from the small base towards the big base. This is the first run where we encounter both forces pointing in the same direction. This is only possible for TE modes because they have a magnetic axial field and the magnetic field is able to impart either a tensile or a compressive force on a surface (while the TM modes have electric axial fields that can only impart a compressive force on surfaces). From the geometry, the force on the lateral conical surfaces due to the electric field must be in the opposite direction, from the big base towards the small base, countering the net force on the bases.
3) This computer run had the antenna positioned near the small end. It will be very interesting to see whether having the antenna at the big end results in a net force in the opposite direction (as previously observed in previous runs with the antenna near the big end for TM modes) which cannot be countered by the lateral conical surfaces.
-
#536 by WarpTech on 15 Aug, 2015 16:48
-
don't know how helpful this is but i do search at arxiv.org for "tapered waveguide" ...
find this for exampe:
http://arxiv.org/pdf/1001.1254v1.pdf
and many many many more...
Only the whole community is able to review all of this
Thank you. I had not seen this paper.
This paper is very, very important as it answers a question that has repeatedly been asked in this forum and in Reddit:
what is the best function to use for the taper?
Shawyer used a linear taper (conical) but lacked the resources to explore other geometries (either numerically or by experiment)
Other researchers have not explored other taper geometries either
Musical people usually come with this question as to why don't people use musical horn shapes for the EM Drive. Invariably their question is answered by saying that the frequencies of the EM drive are much higher than the acoustical frequencies of French Horns, but the question of optimal shape is not answered.
The authors of this paper answer by saying:QuoteBest tradeoff is obtained with the exponential formulation with a slow variation of the
material parameters
that the best taper is exponential instead of linear (conical). However they do not use the same formulation of the problem as they don't of course deal with "thrust" (if there is such a thing) and they consider a simultaneous variation of material properties.
"Like a trombone" I said...
Todd
-
#537 by SeeShells on 15 Aug, 2015 16:59
-
Here are the force calculations corresponding to the stresses shown in http://forum.nasaspaceflight.com/index.php?topic=38203.msg1416281#msg1416281
Some noteworthy comments:
1) The force magnitude is a whooping 10,000 times higher than for previous cases. At this point we don't know how much of this greater magnitude is due to the fact that this computer run is for twice as long a time as previous runs (with stresses that are increasing with time) and how much is due to the fact that this force is due to a transverse electric (TE) mode shape while the other ones were for transverse magnetic (TM) modes.
2) Due to the fact that the stress is tensile at the small base and compressive at the big base, both forces, at the small base and at the big base point in the same direction, from the small base towards the big base. This is the first run where we encounter both forces pointing in the same direction. This is only possible for TE modes because they have a magnetic axial field and the magnetic field is able to impart either a tensile or a compressive force on a surface (while the TM modes have electric axial fields that can only impart a compressive force on surfaces). From the geometry, the force on the lateral conical surfaces due to the electric field must be in the opposite direction, from the big base towards the small base, countering the net force on the bases.
3) This computer run had the antenna positioned near the small end. It will be very interesting to see whether having the antenna at the big end results in a net force in the opposite direction (as previously observed in previous runs with the antenna near the big end for TM modes) which cannot be countered by the lateral conical surfaces.
A very nice piece of work Dr. Rodal! I was expecting an increase in stress forces for the TE mode although not nearly this magnitude. And what is interesting the antenna is out of phase creating a shadow zone in the small cavity of decaying waveforms. Interesting.
I'm waiting for the big end run to finalize the second generation placement of the dual waveguide insertion into the cavity from a single magnetron source.
I fussed over how to make the small plate adjustable with a "Rube Goldberg" contraption on the outside of the frustum to the bottom secured plat to allow for thermal expansion of the walls and hated each iteration. I flashed on an idea of using a quartz rod which is very transparent to microwaves down through the very center, attaching the large base to it and the letting the top small base slide inside of the extended cavity. Still need to capture the small plate with a hollow threaded rod on the outside of it and a captured nut extended to the sidewalls of the extended cavity. This is where I stopped and still fleshing the little details out. The waveguides are just for looks and not representative of where the final fleshed out design will be.
Busy day today, off to lurking in the shop.
Shell
-
#538 by X_RaY on 15 Aug, 2015 17:21
-
Hi Shell.Here are the force calculations corresponding to the stresses shown in http://forum.nasaspaceflight.com/index.php?topic=38203.msg1416281#msg1416281
Some noteworthy comments:
1) The force magnitude is a whooping 10,000 times higher than for previous cases. At this point we don't know how much of this greater magnitude is due to the fact that this computer run is for twice as long a time as previous runs (with stresses that are increasing with time) and how much is due to the fact that this force is due to a transverse electric (TE) mode shape while the other ones were for transverse magnetic (TM) modes.
2) Due to the fact that the stress is tensile at the small base and compressive at the big base, both forces, at the small base and at the big base point in the same direction, from the small base towards the big base. This is the first run where we encounter both forces pointing in the same direction. This is only possible for TE modes because they have a magnetic axial field and the magnetic field is able to impart either a tensile or a compressive force on a surface (while the TM modes have electric axial fields that can only impart a compressive force on surfaces). From the geometry, the force on the lateral conical surfaces due to the electric field must be in the opposite direction, from the big base towards the small base, countering the net force on the bases.
3) This computer run had the antenna positioned near the small end. It will be very interesting to see whether having the antenna at the big end results in a net force in the opposite direction (as previously observed in previous runs with the antenna near the big end for TM modes) which cannot be countered by the lateral conical surfaces.
A very nice piece of work Dr. Rodal! I was expecting an increase in stress forces for the TE mode although not nearly this magnitude. And what is interesting the antenna is out of phase creating a shadow zone in the small cavity of decaying waveforms. Interesting.
I'm waiting for the big end run to finalize the second generation placement of the dual waveguide insertion into the cavity from a single magnetron source.
I fussed over how to make the small plate adjustable with a "Rube Goldberg" contraption on the outside of the frustum to the bottom secured plat to allow for thermal expansion of the walls and hated each iteration. I flashed on an idea of using a quartz rod which is very transparent to microwaves down through the very center, attaching the large base to it and the letting the top small base slide inside of the extended cavity. Still need to capture the small plate with a hollow threaded rod on the outside of it and a captured nut extended to the sidewalls of the extended cavity. This is where I stopped and still fleshing the little details out. The waveguides are just for looks and not representative of where the final fleshed out design will be.
Busy day today, off to lurking in the shop.
Shell
At the first look the orientation of your rectangular waveguide is not the best to excite TE012 (for the other possible modes i have to think about later this day...).
For this type of waveguide the E field is in b-direction which is the shorter side of the rectangle.. H is in a-z direction(z means propagation direction in the waveguide). For TE01 i would rotate both antennas 90deg. But of course its your turn.
https://en.wikipedia.org/wiki/Waveguide_(electromagnetism) -
#539 by WarpTech on 15 Aug, 2015 17:29
-
Hi Shell.Here are the force calculations corresponding to the stresses shown in http://forum.nasaspaceflight.com/index.php?topic=38203.msg1416281#msg1416281
Some noteworthy comments:
1) The force magnitude is a whooping 10,000 times higher than for previous cases. At this point we don't know how much of this greater magnitude is due to the fact that this computer run is for twice as long a time as previous runs (with stresses that are increasing with time) and how much is due to the fact that this force is due to a transverse electric (TE) mode shape while the other ones were for transverse magnetic (TM) modes.
2) Due to the fact that the stress is tensile at the small base and compressive at the big base, both forces, at the small base and at the big base point in the same direction, from the small base towards the big base. This is the first run where we encounter both forces pointing in the same direction. This is only possible for TE modes because they have a magnetic axial field and the magnetic field is able to impart either a tensile or a compressive force on a surface (while the TM modes have electric axial fields that can only impart a compressive force on surfaces). From the geometry, the force on the lateral conical surfaces due to the electric field must be in the opposite direction, from the big base towards the small base, countering the net force on the bases.
3) This computer run had the antenna positioned near the small end. It will be very interesting to see whether having the antenna at the big end results in a net force in the opposite direction (as previously observed in previous runs with the antenna near the big end for TM modes) which cannot be countered by the lateral conical surfaces.
A very nice piece of work Dr. Rodal! I was expecting an increase in stress forces for the TE mode although not nearly this magnitude. And what is interesting the antenna is out of phase creating a shadow zone in the small cavity of decaying waveforms. Interesting.
I'm waiting for the big end run to finalize the second generation placement of the dual waveguide insertion into the cavity from a single magnetron source.
I fussed over how to make the small plate adjustable with a "Rube Goldberg" contraption on the outside of the frustum to the bottom secured plat to allow for thermal expansion of the walls and hated each iteration. I flashed on an idea of using a quartz rod which is very transparent to microwaves down through the very center, attaching the large base to it and the letting the top small base slide inside of the extended cavity. Still need to capture the small plate with a hollow threaded rod on the outside of it and a captured nut extended to the sidewalls of the extended cavity. This is where I stopped and still fleshing the little details out. The waveguides are just for looks and not representative of where the final fleshed out design will be.
Busy day today, off to lurking in the shop.
Shell
At the first look the orientation of your rectangular waveguide is not the best to excite TE012 (for the other possible modes i have to think about later this day...).
For this type of waveguide the E field is in b-direction which is the shorter side of the rectangle.. H is in a-z direction(z means propagation direction in the waveguide). For TE01 i would rotate both antennas 90deg. But of course its your turn.
Picture source:wikipedia
It is a hexagon. She made the sides flat to make it easier to attach 2 magnetrons opposite each other, so they will couple and synchronize. Now you're saying she should mount them to the opposing 120-deg corners? If what you say is correct, then a rectangle or octagon would've been better than a hexagon.
Todd
