Author Topic: EM Drive Developments - related to space flight applications - Thread 3  (Read 1799409 times)

Offline aero

  • Senior Member
  • *****
  • Posts: 2739
  • 92129
  • Liked: 704
  • Likes Given: 236
@Dr. Rodal

Would you like for me to change my coordinate system to make the z axis the axis of rotation? I could do that, most likely it would be quick, but I won't re-run the data that I have already uploaded so you would need to remember which runs are calculated with "image" coordinates and which runs are calculated with "physics conventions."

No thanks.  It is just that it was not clear to me what plane we were looking at.  Now I understand that we are looking at the y-z plane.  Thanks.

Everything is OK at this point.   Don't need a 12 GB file at this point :).

What I would is to get a trapezium view (instead of a circular view) numerical data for L=10.2 for one of the electric view that clearly showed 4 or 3 waves. 

It is very difficult to see visualize a circle as a square.  It will be easier looking at a trapezium view.

The circle is not a square. The circle is embedded within the square. The radius of the circle is 11.01 inches = BIG DIAMETER = 0.29235399999999995 meters and compare with the dimensions of the square computational lattice, 0.3168268537142857 x 0.3168268537142857 meters. The circle is surrounded by the cone skin,  thickness = 1/4 inch or 6.35 mm. Maybe you can find that in the data?
Retired, working interesting problems

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144
Views of NSF-1701 are up:

https://drive.google.com/folderview?id=0B1XizxEfB23tfklENXg2TWhrbUhneGxZQzJ0VVhkRFRwUUhCN2xKX24yOGM2bFQzdVV5NlE&usp=sharing

This file shows the 18 views of the E and H components from x, y and z directions.
This is from a Meep model of rfmwguy's 10.2 inch cavity, in copper, driven by a 2.45000 GHz continuous Ez source. The resonant frequency of this cavity is somewhere close to 2.45 GHz but no resonance solutions indicate exactly 2.45 GHz. The continuous source used in this model is ideal and exact. There are no "shoulders" on the input power source, all input energy is at exactly the drive frequency, 2.45 GHz. The antenna model, a dipole, is oriented perpendicular to the central axis of rotation, one-half wavelength from the small end, and is 0.058 meters long.

An important point is to note that these views use a fixed Max/Min range of power/color intensity. I'll leave it to the experts to analyse the characteristics shown.

Since we continue without having any numerical result, we have to resort to some heuristic as following to interpret these pictures:

1) Picture showing fractal are a numerical artifact due to values close to zero, hence pictures with fractals should be interpreted as low values close to zero

2) Concentrate on images showing smooth and persistent contours (persistent in several frames)

3) Use the field in the longitudinal x direction to determine mode shape

4) Conduct independent eigenvalue analysis to determine frequencies of several modes at 10.2 inches.


///////////////////

Using the above-mentioned heuristics, we determine:

1) The strong field in the longitudinal x direction is Ex, in this case Ex -y (the electric field in the longitudinal direction, seen in the x-z plane with normal y).  This field shows 4 wave patterns in the longitudinal direction.  Hence p=4.  Notice that Ex - z is zero.

2) Hx  -y and Hx -z are fractal hence interpreted as close to zero

3) Actually all magnetic fields in this case appear to be fractal, hence interpreted as very low, close to zero

4) Ez is strong both in the planes with y and z normals.  Both show p=4

5) The strong Ex points towards a TM mode.  The lack of any strong H field in the transverse direcitons points towards a TE mode

6) How can that be?  An examination of the eigenvalue problem using Mathematica and the exact solution shows:


TE114 = 2.434 GHz
TM114 = 2.479 GHz


CONCLUSION: TE114 is excited with strong participation of TM114  We see the strength of Meep: while all the other analysis (including COMSOL FEA by NASA) have been eigenvalue analyses, the Meep solution is a transient solution in time, hence it automatically incorporates a spectrum analysis: it looks at participation of nearby modes.  (This can also be done with COMSOL FEA and other FEA programs of course, but nobody else has reported transient solutions up to now).

Having two modes with equal m, n, p, nearby results in participation of both modes (having 114 mode shape).

I was thinking that because the intensity scale is fixed over each complete view data set, and there is little or no increase in intensity from beginning to end of the run, then the cavity is not resonating very well, certainly not strongly. The fractal nature of the H fields support that thought, by indicating low energy. JMO. Perhaps I will make the same run using a Gaussian source. Noise Bandwidth say, What? What is the noise bandwidth of a magnatron?

I can not belive this modes, i get TE013, TE114, TM113 near 2,45 GHz (analytical).
TM114 is located at a much higher frequency (~2.9GHz) ???
« Last Edit: 06/30/2015 08:44 PM by X_RaY »

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2326
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2956
  • Likes Given: 2588
:)

Confused as well as to what is popping out. I hope we can get it cleared up some.
Shell
« Last Edit: 06/30/2015 08:45 PM by SeeShells »

Offline ElizabethGreene

  • Member
  • Posts: 69
  • Nashville, Tennessee
  • Liked: 138
  • Likes Given: 3
I was thinking about the "Thrust greater than a photon rocket" quandry, and came across something interesting.

Source Paper: Transverse Spin and Momentum in Two-Wave
Interference  http://www.researchgate.net/publication/264276617

Quote
It is well-known, since the seminal works by J.H. Poynting [1], that light carries
momentum and angular momentum (AM) [2,3]. Typical plane-wave or Gaussian-beam
states exhibit longitudinal momentum associated with the wave vector k and also
longitudinal spin AM associated with the degree of circular polarization (helicity) σ .
Locally, optical momentum and angular-momentum densities can demonstrate unusual
features which have recently attracted considerable attention: “super-momentum” with
values higher than (hbar)k per photon [4–8],
transverse helicity-independent spin AM [9–12],
and transverse helicity-dependent momentum [10,13,14]. So far, such abnormal
momentum and spin properties have appeared only in special field configurations, namely,
evanescent waves and optical vortices. Here we find that the simplest propagating non
singular field – two interfering plane waves – also exhibits a variety of extraordinary spin
and momentum properties. Despite the seemingly planar and thoroughly-studied character
of the two-wave system, we discover that such field possesses a transverse (out-of-plane)
helicity-independent spin AM, and also a transverse polarization-dependent momentum
with unusual physical properties.

I only speak pidgin math, but I think the paper translates to say "We predict radiation pressure effects greater than hk/c perpendicular to the plane of incidence when two polarized waves cancel out".

Could I impose on someone to verify my translation?

Online Rodal

  • Senior Member
  • *****
  • Posts: 5835
  • USA
  • Liked: 5914
  • Likes Given: 5253
@Dr. Rodal

Would you like for me to change my coordinate system to make the z axis the axis of rotation? I could do that, most likely it would be quick, but I won't re-run the data that I have already uploaded so you would need to remember which runs are calculated with "image" coordinates and which runs are calculated with "physics conventions."

No thanks.  It is just that it was not clear to me what plane we were looking at.  Now I understand that we are looking at the y-z plane.  Thanks.

Everything is OK at this point.   Don't need a 12 GB file at this point :).

What I would is to get a trapezium view (instead of a circular view) numerical data for L=10.2 for one of the electric view that clearly showed 4 or 3 waves. 

It is very difficult to see visualize a circle as a square.  It will be easier looking at a trapezium view.

The circle is not a square. The circle is embedded within the square. The radius of the circle is 11.01 inches = BIG DIAMETER = 0.29235399999999995 meters and compare with the dimensions of the square computational lattice, 0.3168268537142857 x 0.3168268537142857 meters. The circle is surrounded by the cone skin,  thickness = 1/4 inch or 6.35 mm. Maybe you can find that in the data?

What is the function of the finite difference grid outside the circle?  and why does it have a similar field?

This has been useful as we have determined:

1) The field numbers are extremely small:  around 10^-8.  This may be due to the Meep units:  Meep is a code written for optical applications and you are modeling here something at microwave frequencies that are much lower.

2) Or it may be due to the fact that you gave me data for the big base:  all transverse electric fields are supposed to be exactly zero at the base.  The only non-zero field is the electric field perpendicular to the base.  But if this is an electric mode, the field in the longitudinal direction is supposed to be magnetic and the electric field perpendicular to the base is supposed to be zero.

3) I'll wait until you give us numerical data for the trapezoid sections...

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144
....

I can not belive this modes, i get TE013, TE114, TM113 near 2,45 GHz (analytical).
TM114 is located at a much higher frequency (~2.9GHz) ???

I cant't believe your numbers.  What code are you using to get such a high frequency ???

Have you ever checked your code against other solutions for a truncated cone?

COMSOL's FEA analysis gives TM212 for L=9 inches at 2.45 GHz and TM114 at a much lower frequency. 

Are you using the formula for a cylinder?  or some numerical approximation that is way too stiff?

Sounds like something is wrong with your calculation for TM114

I used the L=10.2inch=259,08 mm; Big Diameter=259,08 mm; Small Diameter =148...155mm .
I integrate over several frequencies given by a fixed length and the Diameter an any z position on this axis to get the eigenvalue of the mode (with respect to the given bessel-funktion of each mode) It's equal to integrate over several diameters like Shawyer said.
This works good enough. I work daily with this code to build conical cavities for NDT-Solutions (I calculate and build the cavities). The frequencies fits for the Eigenstates. (antenna and losses/London penetration depth gives a small frequency shift...). Got a good equipment available (Spec, VNA...!)

TE013  2,540899898GHz
TE114  2,4817950934GHz
TM113  2,5445470879GHz
« Last Edit: 06/30/2015 09:14 PM by X_RaY »

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2326
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2956
  • Likes Given: 2588
Nice find! I try to feed the Pidgins sometimes.

I find it interesting that just in normal wave cancellations they are finding this now. I've been digging through the strange  Evanescent actions of spin and momentum and will find this a good read.

Thanks for posting.

Shell

I was thinking about the "Thrust greater than a photon rocket" quandry, and came across something interesting.

Source Paper: Transverse Spin and Momentum in Two-Wave
Interference  http://www.researchgate.net/publication/264276617

Quote
It is well-known, since the seminal works by J.H. Poynting [1], that light carries
momentum and angular momentum (AM) [2,3]. Typical plane-wave or Gaussian-beam
states exhibit longitudinal momentum associated with the wave vector k and also
longitudinal spin AM associated with the degree of circular polarization (helicity) σ .
Locally, optical momentum and angular-momentum densities can demonstrate unusual
features which have recently attracted considerable attention: “super-momentum” with
values higher than (hbar)k per photon [4–8],
transverse helicity-independent spin AM [9–12],
and transverse helicity-dependent momentum [10,13,14]. So far, such abnormal
momentum and spin properties have appeared only in special field configurations, namely,
evanescent waves and optical vortices. Here we find that the simplest propagating non
singular field – two interfering plane waves – also exhibits a variety of extraordinary spin
and momentum properties. Despite the seemingly planar and thoroughly-studied character
of the two-wave system, we discover that such field possesses a transverse (out-of-plane)
helicity-independent spin AM, and also a transverse polarization-dependent momentum
with unusual physical properties.

I only speak pidgin math, but I think the paper translates to say "We predict radiation pressure effects greater than hk/c perpendicular to the plane of incidence when two polarized waves cancel out".

Could I impose on someone to verify my translation?

Online Rodal

  • Senior Member
  • *****
  • Posts: 5835
  • USA
  • Liked: 5914
  • Likes Given: 5253
....

I can not belive this modes, i get TE013, TE114, TM113 near 2,45 GHz (analytical).
TM114 is located at a much higher frequency (~2.9GHz) ???

I cant't believe your numbers.  What code are you using to get such a high frequency ???

Have you ever checked your code against other solutions for a truncated cone?

COMSOL's FEA analysis gives TM212 for L=9 inches at 2.45 GHz and TM114 at a much lower frequency. 

Are you using the formula for a cylinder?  or some numerical approximation that is way too stiff?

Sounds like something is wrong with your calculation for TM114

I used the L=10.2inch=259,08 mm; Big Diameter=259,08 mm; Small Diameter =148...155mm .
I integrate over several frequencies given by a fixed length and the Diameter an any z position on this axis to get the eigenvalue of the mode (with respect to the given bessel-funktion of each mode)
This works good enough. I work daily with this code to build conical cavities for NDT-Solutions (I calculate and build the cavities). The frequencies fits for the Eigenstates. (antenna and losses/London penetration depth gives a small frequency shift...). Got a good equipment available (Spec, VNA...!)

TE013  2,540899898GHz
TE114  2,4817950934GHz
TM113  2,5445470879GHz

Sorry, it looks like your solution does not work good enough (at least for this case):

You calculate TM113  2,5445470879GHz for L=10.2 inches, which shows that something is very wrong with your calculation.

NASA COMSOL FEA calculates for L = 9 inches that TM113 has a lower frequency:

  TM113 at 2.273 GHz for L=9 inches

the exact solution gives

  TM113 2.24832 GHz for L = 9 inches (good agreement with NASA COMSOL FEA)

Frequency goes down with length, not up  !!!  So for L=10.2 the frequency of TM113 is even lower !!!
instead of higher as you calculate (TM113 2.545 GHz for L=10.2 inches)

NOTE: I notice that you are using cylindrical Bessel functions.  The correct functions for truncated cone are the Associated Legendre function and the Spherical Bessel Functions.
« Last Edit: 06/30/2015 09:19 PM by Rodal »

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144
....

I can not belive this modes, i get TE013, TE114, TM113 near 2,45 GHz (analytical).
TM114 is located at a much higher frequency (~2.9GHz) ???

I cant't believe your numbers.  What code are you using to get such a high frequency ???

Have you ever checked your code against other solutions for a truncated cone?

COMSOL's FEA analysis gives TM212 for L=9 inches at 2.45 GHz and TM114 at a much lower frequency. 

Are you using the formula for a cylinder?  or some numerical approximation that is way too stiff?

Sounds like something is wrong with your calculation for TM114

I used the L=10.2inch=259,08 mm; Big Diameter=259,08 mm; Small Diameter =148...155mm .
I integrate over several frequencies given by a fixed length and the Diameter an any z position on this axis to get the eigenvalue of the mode (with respect to the given bessel-funktion of each mode)
This works good enough. I work daily with this code to build conical cavities for NDT-Solutions (I calculate and build the cavities). The frequencies fits for the Eigenstates. (antenna and losses/London penetration depth gives a small frequency shift...). Got a good equipment available (Spec, VNA...!)

TE013  2,540899898GHz
TE114  2,4817950934GHz
TM113  2,5445470879GHz

Sorry, it looks like your solution does not work good enough (at least for this case):

You calculate TM113  2,5445470879GHz for L=10.2 inches, which shows that something is very wrong with your calculation.

NASA COMSOL FEA calculates for L = 9 inches that TM113 has a lower frequency:

  TM113 at 2.273 GHz for L=9 inches

the exact solution gives

  TM113 2.24832 GHz for L = 9 inches (good agreement with NASA COMSOL FEA)

Frequency goes down with length, not up  !!!  So for L=10.2 the frequency of TM113 is even lower !!!
instead of higher as you calculate (2.545 GHz for L=10.2 inches)

NOTE: I notice that you are using cylindrical Bessel functions.  The correct functions for truncated cone are the Associated Legendre function and the Spherical Bessel Functions.
OK maybe (like i said that are analytical things! Maybe i have to increase the Points along the z-axis? I use 15 at the moment this could be a little bit less for modes with higher p value :-\ )

Did i used the right Diameter data?
« Last Edit: 06/30/2015 09:29 PM by X_RaY »

Online Rodal

  • Senior Member
  • *****
  • Posts: 5835
  • USA
  • Liked: 5914
  • Likes Given: 5253
...
OK maybe (like i said that are analytical things! Maybe i have to increase the Points along the z-axis?)

Did i used the right Diameter data?

Suggestion:

1) Double check all dimensions and your conversions from US to metric

Height: 9.00 inch (228.6 mm)
Top diam.: 6.25 inch (0.1588 mm)
Bottom diam.: 11.01 inch (279.7 mm)
Material: 101 Copper Alloy

2) Perform a convergence analysis by at least doubling the number of points (better in all directions, but longitudinally, specially) for L=10.2 in

3) Do a comparison for L=9 inches instead of L=10.2 and compare with the spreadsheet shown by SeeShell in her message: both vs NASA and exact for ALL the frequencies in the spreadsheet

http://forum.nasaspaceflight.com/index.php?action=dlattach;topic=37642.0;attach=1036684
« Last Edit: 06/30/2015 09:30 PM by Rodal »

Online Rodal

  • Senior Member
  • *****
  • Posts: 5835
  • USA
  • Liked: 5914
  • Likes Given: 5253

I used the L=10.2inch=259,08 mm; Big Diameter=259,08 mm; Small Diameter =148...155mm .
...
OK maybe (like i said that are analytical things! Maybe i have to increase the Points along the z-axis? I use 15 at the moment this could be a little bit less for modes with higher p value :-\ )

Did i used the right Diameter data?

It looks like your diameters are wrong:


NASA Frustum
Height: 9.00 inch (228.6 mm)
Top diam.: 6.25 inch (0.1588 mm)       
Bottom diam.: 11.01 inch (279.7 mm)
Material: 101 Copper Alloy
« Last Edit: 06/30/2015 09:34 PM by Rodal »

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144

I used the L=10.2inch=259,08 mm; Big Diameter=259,08 mm; Small Diameter =148...155mm .
...
OK maybe (like i said that are analytical things! Maybe i have to increase the Points along the z-axis? I use 15 at the moment this could be a little bit less for modes with higher p value :-\ )

Did i used the right Diameter data?

It looks like your diameters are wrong:


NASA Frustum
Height: 9.00 inch (228.6 mm)
Top diam.: 6.25 inch (0.1588 mm)       
Bottom diam.: 11.01 inch (279.7 mm)
Material: 101 Copper Alloy

 i will work on it
thanks

Edit: now it's much closer but i have used c in vacuum not in air and i will increase the number of points..
at the moment with this data of the Diameters and 10.2inch(40pionts in z-direction):
TE013=2,4481201288GHz
TE114=2,4599924415GHz
TM014=2,5574279592GHz
TM113=2,4481232842GHz
TM114=2,8893105988GHz :/ ???
« Last Edit: 06/30/2015 10:30 PM by X_RaY »

Offline zellerium

  • Full Member
  • *
  • Posts: 171
  • Pittsburgh, PA
  • Liked: 279
  • Likes Given: 400
Hey, quick question to those savvy with microwaves:

We would like to determine how well the magnetron is matched with the manufacturers microwave box so that we have a baseline for safe operating impedance of the magnetron. My idea is to put several temperature probes on the magnetron outside core and heat sink fins and log data. Our probes max out at 130 C, does anyone have any idea how quickly an uncooled magnetron will get there?

This method would allow a simple impedance measurement so that any subsequent cavity we create can be compared to the original microwave. [I'd be making the assumption that the manufacturer created a cavity that is well matched to the magnetron to minimize reflected power. Is this a valid assumption? ]

Any thoughts, concerns, suggestions?


Kurt
 

Online Rodal

  • Senior Member
  • *****
  • Posts: 5835
  • USA
  • Liked: 5914
  • Likes Given: 5253
@Dr. Rodal

Would you like for me to change my coordinate system to make the z axis the axis of rotation? I could do that, most likely it would be quick, but I won't re-run the data that I have already uploaded so you would need to remember which runs are calculated with "image" coordinates and which runs are calculated with "physics conventions."

No thanks.  It is just that it was not clear to me what plane we were looking at.  Now I understand that we are looking at the y-z plane.  Thanks.

Everything is OK at this point.   Don't need a 12 GB file at this point :).

What I would is to get a trapezium view (instead of a circular view) numerical data for L=10.2 for one of the electric view that clearly showed 4 or 3 waves. 

It is very difficult to see visualize a circle as a square.  It will be easier looking at a trapezium view.

The circle is not a square. The circle is embedded within the square. The radius of the circle is 11.01 inches = BIG DIAMETER = 0.29235399999999995 meters and compare with the dimensions of the square computational lattice, 0.3168268537142857 x 0.3168268537142857 meters. The circle is surrounded by the cone skin,  thickness = 1/4 inch or 6.35 mm. Maybe you can find that in the data?

I just found out that you have a finite difference grid outside the area where you are imposing boundary conditions and the antenna RF feed

Assuming that we are not looking at evanescent waves leaking out of the frustum at the moment (no indications of such leaking in the images and movies I have seen of the last few cases)

This is concerning because:

1) Not only you have finite difference grid taking memory space and computer time for no apparent purpose

2) but most importantly this can lead to numerical ill-conditioning of the finite-difference solution.  Those finite-difference grids outside  the boundary where boundary conditions (BC) are imposed should (if there is no evanescent wave leaking) have a zero field, so when solving for the field in the area outside the boundary where the BC are imposed those FD equations are ill-conditioned, and this numerical problem may translate artificially to the interior (depending on how the Meep authors wrote the code). 

Normally one would not include any finite difference mesh outside the boundary where the BC are imposed. 

It makes things easier for plotting the results to have the square mesh, though.

Have you double-checked that it is OK with MEEP to have a finite-difference grid outside the boundary where BC  are imposed?

///////////////////////

Another question, the last Ex x image you have is for T=325 and it looks like this:



observe that there are no field values outside the circle

I attach below the csv file plotted rotated so as to be in similar position.  It looks similar, except that

the square is mapped into the circle

So:

It looks like Meep is smart enough to realize that you have FD grids outside the BC and is ignoring them so that all that gets output is the circle FD
« Last Edit: 06/30/2015 11:37 PM by Rodal »

Offline arc

  • Member
  • Posts: 20
  • In port for a few weeks
  • Liked: 10
  • Likes Given: 32
....  The em-drive will eventually take its place next to cold fusion, polywater, and 300 MPH submarines in the Encyclopedia of Pseudoscience.

http://defensetech.org/2009/11/17/super-cavitation-and-the-truth/

Offline arc

  • Member
  • Posts: 20
  • In port for a few weeks
  • Liked: 10
  • Likes Given: 32
Weird, take a look at: http://jnaudin.free.fr/lifters/act/html/omptv1.htm

"On January 31, 2002, the NASA patent application US2002012221 " Apparatus and Method for generating a thrust using a two dimensional asymmetrical capacitor module " has been granted."

Looks alot like a shawyer frustum...

From the article:

The dielectric material of a capacitor under high voltage experiences a force. Based on the geometry of the capacitor, its material properties, and ambient conditions, the force can be predicted and utilized to move the entire capacitor and its mounting in a predictable direction.

Are there any parts of the EM drive that may act as a high voltage capacitor?

More like a high current inductor.
Warptech,  this is correct if there is full metal to metal contact at both ends.

If Shawyer ( and/or others) have used an insulating gasket between the small end and cone, or large end and cone, then there is an increased capacitive effect. the dielectric in this case is the air inside the cavity.
if the gasket is between the large end and cone and the small end is full metal contact with the cone, the small end + cone is one plate of the capacitive effect, the large end is the other plate

Offline aero

  • Senior Member
  • *****
  • Posts: 2739
  • 92129
  • Liked: 704
  • Likes Given: 236
Meep uses, in this case, a Cartesian coordinate system and computational lattice. The structure is put inside that computational lattice. I put the antenna inside the structure. The fields propagate everywhere except where they are blocked by structure. They are blocked by the material dielectric constant (?). I am running with our copper model which I have discussed before. I may not be implementing the copper permittivity model correctly. I am using 1/4 inch skin model for the frustum.  Using perfect metal and a thick enough skin on the frustum, nothing escapes the frustum. Thick enough to avoid having adjacent pixels in the time step or geometric lattice stepping across the skin. The complete Cartesian lattice is still there however.

I suggest that you may wish to look at some of the fields generated with the Gaussian source as they appear to be much stronger than those calculated using the continuous source. Csv files are available for the big end base view, and I'll see about creating csv files for the "transverse" view.

Regarding your observation, "observe that there are no field values outside the circle" I disagree. If there were no fields, the background would show as black, I think. There is some energy there, just the fields are so weak compared to the fields inside that they are not differentiated by color from some very small value.
Retired, working interesting problems

Offline zen-in

  • Full Member
  • ****
  • Posts: 531
  • California
  • Liked: 468
  • Likes Given: 365
Hey, quick question to those savvy with microwaves:

We would like to determine how well the magnetron is matched with the manufacturers microwave box so that we have a baseline for safe operating impedance of the magnetron. My idea is to put several temperature probes on the magnetron outside core and heat sink fins and log data. Our probes max out at 130 C, does anyone have any idea how quickly an uncooled magnetron will get there?

This method would allow a simple impedance measurement so that any subsequent cavity we create can be compared to the original microwave. [I'd be making the assumption that the manufacturer created a cavity that is well matched to the magnetron to minimize reflected power. Is this a valid assumption? ]

Any thoughts, concerns, suggestions?


Kurt
 
The reflected wave would likely damage the megnetron before the temperature sensors registered a high temperature.   It is better to design the feedline so that the return wave can't get back to the magneron and then match the cavity to minimize the return wave.   One photo that was posted awhile back showed a feedline with an inline waveguide circulator.   It may have been from the Chinese experiment.   It could easily be replicated by a machine shop.

Offline quixote

  • Member
  • Posts: 13
  • Liked: 0
  • Likes Given: 84
Been following this thread for a few weeks, decided to hop in to help if I could.

Aero, do you need only the most recent MEEP package?
I was going to go ahead and compile it from source for you but I found:
http://ab-initio.mit.edu/wiki/index.php/Meep_Download

According to the wiki, they have a precompiled source package available:
"apt-get install meep h5utils"

There is also a parallel source file:
"apt-get install meep-mpi"

If you need other packages compiled with it or the OpenMPI version, I will see what I can do.

I saw that one as well and was going to try it out.
If you want to help, you could try to create a Ubuntu package for Meep as aero uses Ubuntu (and so am I and a few others).  http://packaging.ubuntu.com/html/

I also do not know which compiler / optimization the source code would support, but you may want to explore recompiling the source using different compilers (gcc / LLVM / ICC / VC ...) and different optimization flags.  It can also be that the source code could enjoy being made more compiler agnostic.

There is already an Ubuntu distrubtion available.

http://ab-initio.mit.edu/wiki/index.php/Meep_download

Then look down at the "Precompiled Meep packages for Debian and Ubuntu" section. It's pretty easy. There is even an MPI enabled version.

Online Rodal

  • Senior Member
  • *****
  • Posts: 5835
  • USA
  • Liked: 5914
  • Likes Given: 5253
...
I suggest that you may wish to look at some of the fields generated with the Gaussian source as they appear to be much stronger than those calculated using the continuous source. Csv files are available for the big end base view, and I'll see about creating csv files for the "transverse" view.

Regarding your observation, "observe that there are no field values outside the circle" I disagree. If there were no fields, the background would show as black, I think. There is some energy there, just the fields are so weak compared to the fields inside that they are not differentiated by color from some very small value.

Nope.:  ALL that is shown is inside the boundary condition.  If you are inputting a finite difference mesh outside the BC,  Meep is ignoring those nodes.  Since you are only exciting the inside with RF, and you have BC, Meep has to ignore those nodes outside the BC, otherwise the solution would be ill conditioned


All the points in the csv file are inside the circle.  The circle is mapped into the square.

I choose to plot only 10 contours for clarity.


These views are rotated 90 degrees from your views.

I confirm that the fractals are a result of the very coarse mesh you are using:  the Ez and Ey fields should be zero at the base. They are not zero:  they are about 10 times smaller in magnitude only.

All numbers are very small: the highest magnitude numbers are about 10^(-13) !
« Last Edit: 07/01/2015 12:31 AM by Rodal »

Tags: