Author Topic: EM Drive Developments - related to space flight applications - Thread 2  (Read 2105348 times)

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260
@Mulletron

I made some cylindrical cavity runs.....
Are you modeling the geometrical shape now with a 3-D (three-dimensional) MEEP mesh or is your present MEEP model a flat plane with a rectangular boundary (2-D two-dimensional model) ?

If it is a 3-D (three-dimensional) mesh, can you also show the solution plots for the circular cross-section of the cylindrical cavity ?
« Last Edit: 03/02/2015 01:03 AM by Rodal »

Offline aero

  • Senior Member
  • *****
  • Posts: 2743
  • 92129
  • Liked: 704
  • Likes Given: 239
No, its still a 2D model. The wife hasn't bought me a new computer yet. This model is smaller than the Copper Kettle thruster so I could likely make one good 3D run in a day's time. Like to know what you'd like to see, re-drive frequency, dielectric constant and so forth, before I do that. And what Mulletron has to say.

Still having trouble calculating the resonant frequency, too. I've picked the frequency off of the F/P curves and run the field patterns from that. It looks to be close to resonance but the frequencies I used for the field patterns are clearly not in resonance. I save a frame every 1/6 th of the period of the drive frequency (Using vacuum light speed for the Period calculation) so I can tell almost when a cycle completes. But the pattern is never quite the same six frames later.

Oh, I used my old stand-by dielectric constant of 1.76 in these runs, for what its worth.
Retired, working interesting problems

Offline Notsosureofit

  • Full Member
  • ****
  • Posts: 656
  • Liked: 704
  • Likes Given: 1364
Thanks guys !

So I have data for everything except the mode(s), and temp size for the flight cone.

I need to use X numbers of 21, 26, and 77, respectively, for the proto, demo and flight cones to get those numbers. (w/o dielectric that is)

What that means is TBD of course.  There may be another variable involved, in or out of favor of a real effect.

Realized this morning I'd been missing something after rereading Shawer's flight paper and the Magnetron "lock" time.

Kicking myself for not recognizing a "Q multiplier circuit" (I've used enough of them)  The long lock time (time constant) is the tip-off if this is what he's doing and the results are real.  The trade off in that case is spectral purity vs the frequency stability of the oscillator since you track the cavity.

Needs only a modest *10 for the proto and demo cases.  The flight system needs *100 which is not out of the question at all.

Again, this is pure speculation on my part w/o confirmation from Shawyer, but it puts the mode back in a reasonable range.



Notsosureofit:

Could you describe to me how one goes about building a "Q-multiplier Circuit" for a 1.90 GHz RF amplifier circuit?  I've used the old Heathkit QF-1 Q-Multiplier for my old shortwave radio receiver back in high school, see: http://tubularelectronics.com/Heath_Manual_Collection/Heath_Manuals_O-RX/QF-1/QF-1.pdf ,  but I've never thought to use one to enhance the Q-Factor of a microwave frustum cavity before...

Best, Paul M.

"Q-multiplier" is the Heath term for adding feedback, short of oscillation, to an IF amplifier to narrow the bandwidth
and enhance its "Q" as a filter.  (had one on an HRO receiver in the 50's.  HRO long gone but the Q-Multiplier is still in the pile somewhere)

In the case of greater feed back you get an oscillator. (had a 220MHz re-entrant cavity oscillator at that time)

The right feedback loop will improve the phase coherence of the oscillation and the "circuit Q".  You pay for this with an increased time constant and are limited by the (thermal ?) drift rate of the cavity.  The "flight" cavity might be heavily built for that reason ?

Hopefully there is a radar guy (like Shawyer) on tap that could give a better explanation. (I'm pulling this out of memories of my misspent youth...)

I was trying to remember something about radar systems (Russian ?) that had a dielectric resonator suspended in a microwave cavity.............

The "Q multiplier" effect comes from the calculation of the loaded resonator Q, as f/2 times the loop gain phase slope.

"Loaded-Q represents the width of the resonance curve, or phase slope, including the
effects of external components. In this case the Q is determined mostly by the external
components."

In typical references that I found, ie.low noise frequency sources, the loaded Q is between 20% to 50% of the unloaded Q.

So for the "Shawyer" type application, to get a maximized loaded Q in a cavity oscillator, one should maximize the loop gain phase slope.  (use the highest gain-bandwidth amp you can find ?)

One question that comes to mind at the moment is;  What is the loaded Q of the present (non oscillator) experiments since that is presumably the relevant parameter.

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260
DIYFAN:

Once the test series we are working on is finished, I will suggest to Dr. White that we try the use of the more readily available NiFeCo mu-metal from McMaster-Carr (See: http://www.mcmaster.com/#mu-metal-foil/=w4hfa3 ) for such a test.  However I think we will have to copper plate the side of the mu-metal facing the interior of the cavity with about 10 microns of copper or silver to keep this large OD end-cap from greatly reducing the Q-Factor of the copper frustum.  Mu-metal resistivity is much higher than copper...

Best,  Paul M.
I couldn't find magnetic permeability values for mu metal close to the GHz range, except the frequent warning (also included in the Wikipedia article) <<The high permeability makes mu-metal useful for shielding against static or low-frequency magnetic fields>> (bold added for emphasis).

I wonder what is the relative magnetic permeability of mu metal in the GHz range, and how effective it will be for the purposes described by Aquino.

Here is an example of a Korean high permeability material (not mu metal) with an initial relative permeability of 15000.  The report shows that the relative permeability drops precipitously at frequencies over 100 KHz and goes to zero at frequencies above 1 MHz


http://www.samwha.co.kr/electronics/product/material/Material%20characteristics_SM-150.pdf

Offline Star-Drive

  • Member
  • Full Member
  • ****
  • Posts: 825
  • TX/USA
  • Liked: 866
  • Likes Given: 9
Dr. Rodal:

Duly noted on mu-metal's drastic falloff in permeability with frequency especially above 1.0 MHz.  Looks Like I will have to continue the search if I perform this Aquino experiment...

Best, Paul M.
Star-Drive

Offline aero

  • Senior Member
  • *****
  • Posts: 2743
  • 92129
  • Liked: 704
  • Likes Given: 239
@Rodal, @notsosureofit, and everyone else.
I've been wondering how to treat the dielectric to compute the resonant frequency of the cylindrical cavity containing a dielectric. After some Google time I find that the velocity factor, VF, of a material is given by the inverse square root of relative permittivity. That's the dielectric constant! And I found that VF for a solid polyethylene core transmission cable is 0.66. That's right for the dielectric constant = 2.3.

So with this, the velocity of the RF waves in the dielectric is V = c *VF, or V = 0.66 c (c - speed of light). Isn't that the same as saying the equivalent vacuum distance within the dielectric, D = 1.515 d, where d is the geometry of the dielectric disk? So a cylinder of 1 unit high and 1 unit diameter containing a dielectric disk of 1/2 unit high and 1 unit diameter would respond to RF as though it were (1/2 +1.515/2) units high and 1.515 units diameter at the end containing the dielectric and 1 unit in diameter at the empty end. So we now have a frustum 1.515 x 1 units diameters x 1.251 high. Is that right? If so, we can use Dr. Rodal's formula to calculate the resonant frequency.

Maybe Dr. Rodal will tell me what the resonant frequency should be for the cylindrical cavity with dielectric that I posted data for earlier today?
Retired, working interesting problems

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260


Maybe Dr. Rodal will tell me what the resonant frequency should be for the cylindrical cavity with dielectric that I posted data for earlier today?
Please fill-in the following data (question marks ? below) for the cylindrical cavity with a dielectric section having the same diameter as the cavity's ID and located at one end of the cavity:

GEOMETRY

Inner Diameter of cylindrical cavity =   ? meters (cavity has a constant, same diameter throughout)
Total Inner Length of cavity            =   ? meters (Length including dielectric length)
Length of dielectric section              =   ? meters

CONSTITUTIVE PROPERTIES

Relative electric permittivity of dielectric =      ?     (dimensionless)  (for HD PE it is reported as 2.3)
Relative magnetic permeability of dielectric =      ?     (dimensionless)  (1 ?)
Relative electric permittivity of empty section =   ?   (dimensionless)  (air or vacuum ?)
Relative magnetic permeability of empty section =   ?   (dimensionless)  (air or vacuum ?)

FREQUENCY of interest

There are an infinite number of resonant frequencies for a resonant cylindrical cavity. 
What resonant frequency are you referring to?     ?  (The lowest natural frequency? )
« Last Edit: 03/02/2015 02:10 PM by Rodal »

Offline Notsosureofit

  • Full Member
  • ****
  • Posts: 656
  • Liked: 704
  • Likes Given: 1364
@ Star-Drive


Realized this morning I'd been missing something after rereading Shawer's flight paper and the Magnetron "lock" time.

Kicking myself for not recognizing a "Q multiplier circuit" (I've used enough of them)  The long lock time (time constant) is the tip-off if this is what he's doing and the results are real.  The trade off in that case is spectral purity vs the frequency stability of the oscillator since you track the cavity.

Needs only a modest *10 for the proto and demo cases.  The flight system needs *100 which is not out of the question at all.

Again, this is pure speculation on my part w/o confirmation from Shawyer, but it puts the mode back in a reasonable range.


Notsosureofit:

Could you describe to me how one goes about building a "Q-multiplier Circuit" for a 1.90 GHz RF amplifier circuit?  I've used the old Heathkit QF-1 Q-Multiplier for my old shortwave radio receiver back in high school, see: http://tubularelectronics.com/Heath_Manual_Collection/Heath_Manuals_O-RX/QF-1/QF-1.pdf ,  but I've never thought to use one to enhance the Q-Factor of a microwave frustum cavity before...

Best, Paul M.

"Q-multiplier" is the Heath term for adding feedback, short of oscillation, to an IF amplifier to narrow the bandwidth
and enhance its "Q" as a filter. In the case of greater feed back you get an oscillator. The right feedback loop will improve the phase coherence of the oscillation and the "circuit Q".  You pay for this with an increased time constant and are limited by the (thermal ?) drift rate of the cavity.  The "flight" cavity might be heavily built for that reason ?

The "Q multiplier" effect comes from the calculation of the loaded resonator Q, as f/2 times the loop gain phase slope.

"Loaded-Q represents the width of the resonance curve, or phase slope, including the effects of external components. In this case the Q is determined mostly by the external
components."

In typical references that I found, ie. low noise frequency sources, the loaded Q is between 20% to 50% of the unloaded Q.

So for the "Shawyer" type application, to get a maximized loaded Q in a cavity oscillator, one should maximize the loop gain phase slope.  (use the highest gain-bandwidth amp you can find ?)


The oscillator model is the same as the tuned amp model w/ the cavity used as a filter in the feedback loop.  In a quick search, I was only able to find one worked out example of using feedback to increase the loaded Q of an oscillator over that of the unloaded Q.

  www.zen22142.zen.co.uk/Design/vcqmswo.pdf

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260
....
The oscillator model is the same as the tuned amp model w/ the cavity used as a filter in the feedback loop.  In a quick search, I was only able to find one worked out example of using feedback to increase the loaded Q of an oscillator over that of the unloaded Q.

  www.zen22142.zen.co.uk/Design/vcqmswo.pdf
Very interesting that this is the only example readily available in a quick search.  The example comes from Lima, Peru.  One cannot help but wonder whether the worked out analysis in the example is flawless and whether it can work in practice to increase the loaded Q over that of the unloaded Q of the EM Drive.

Have the loaded Q's always been lower than the unloaded Q's in all the EM Drive experiments performed in the UK, US and China?  There seems to be a paucity of data on this, is my memory is correct...
« Last Edit: 03/02/2015 03:02 PM by Rodal »

Offline aero

  • Senior Member
  • *****
  • Posts: 2743
  • 92129
  • Liked: 704
  • Likes Given: 239


Maybe Dr. Rodal will tell me what the resonant frequency should be for the cylindrical cavity with dielectric that I posted data for earlier today?
Please fill-in the following data (question marks ? below) for the cylindrical cavity with a dielectric section having the same diameter as the cavity's ID and located at one end of the cavity:

GEOMETRY

Inner Diameter of cylindrical cavity =   ? meters (cavity has a constant, same diameter throughout)
Total Inner Length of cavity            =   ? meters (Length including dielectric length)
Length of dielectric section              =   ? meters

CONSTITUTIVE PROPERTIES

Relative electric permittivity of dielectric =      ?     (dimensionless)  (for HD PE it is reported as 2.3)
Relative magnetic permeability of dielectric =      ?     (dimensionless)  (1 ?)
Relative electric permittivity of empty section =   ?   (dimensionless)  (air or vacuum ?)
Relative magnetic permeability of empty section =   ?   (dimensionless)  (air or vacuum ?)

FREQUENCY of interest

There are an infinite number of resonant frequencies for a resonant cylindrical cavity. 
What resonant frequency are you referring to?     ?  (The lowest natural frequency? )

I just printed these numbers out from my program so this is what I used when generating the posted data.
Inner diameter of cylindrical cavity, 0.08278945,m
 total inner length of cavity, 0.1224489,m
 Length of dielectric section 0.027282494103102, m

 Relative electric permittivity of dielectric =1.76


Relative magnetic permeability of dielectric =      1     
Relative electric permittivity of empty section =   vacuum   (a meep program option)
Relative magnetic permeability of empty section = vacuum

I know that the dielectric constant of hdpe is 2.3. I will make some runs later using that value but for now I am using 1.76.

The resonances I'm looking for are those at the peaks of the force/power curves above, near 1.8 Ghz, 1.9 GHz and 2.4 GHz using an electric source, and about 2.3 GHz with the magnetic source.

It could be that there are no resonances in those frequency ranges but everything we know right now says that there will be.

And thanks.
Retired, working interesting problems

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260


Maybe Dr. Rodal will tell me what the resonant frequency should be for the cylindrical cavity with dielectric that I posted data for earlier today?
Please fill-in the following data (question marks ? below) for the cylindrical cavity with a dielectric section having the same diameter as the cavity's ID and located at one end of the cavity:

GEOMETRY

Inner Diameter of cylindrical cavity =   ? meters (cavity has a constant, same diameter throughout)
Total Inner Length of cavity            =   ? meters (Length including dielectric length)
Length of dielectric section              =   ? meters

CONSTITUTIVE PROPERTIES

Relative electric permittivity of dielectric =      ?     (dimensionless)  (for HD PE it is reported as 2.3)
Relative magnetic permeability of dielectric =      ?     (dimensionless)  (1 ?)
Relative electric permittivity of empty section =   ?   (dimensionless)  (air or vacuum ?)
Relative magnetic permeability of empty section =   ?   (dimensionless)  (air or vacuum ?)

FREQUENCY of interest

There are an infinite number of resonant frequencies for a resonant cylindrical cavity. 
What resonant frequency are you referring to?     ?  (The lowest natural frequency? )

I just printed these numbers out from my program so this is what I used when generating the posted data.
Inner diameter of cylindrical cavity, 0.08278945,m
 total inner length of cavity, 0.1224489,m
 Length of dielectric section 0.027282494103102, m

 Relative electric permittivity of dielectric =1.76


Relative magnetic permeability of dielectric =      1     
Relative electric permittivity of empty section =   vacuum   (a meep program option)
Relative magnetic permeability of empty section = vacuum

I know that the dielectric constant of hdpe is 2.3. I will make some runs later using that value but for now I am using 1.76.

The resonances I'm looking for are those at the peaks of the force/power curves above, near 1.8 Ghz, 1.9 GHz and 2.4 GHz using an electric source, and about 2.3 GHz with the magnetic source.

It could be that there are no resonances in those frequency ranges but everything we know right now says that there will be.

And thanks.

GEOMETRICAL INPUT for cylindrical cavity:

bigDiameter = 8.278945 centimeter;
smallDiameter = bigDiameter;
length = 12.24489 centimeter;
dielectricThickness = 2.7282494103102 centimeter;



You can verify this case directly from the equation in Wikipedia http://en.wikipedia.org/wiki/Microwave_cavity#Cylindrical_cavity

First four mode shapes and frequencies for Relative electric permittivity of dielectric =1 (Same thing as no dielectric)

{{"TE", 1, 1, 0}, 2.12223*10^9},
{{"TE", 1, 1, 1}, 2.44998*10^9}, {{"TM", 0, 1, 0}, 2.77191*10^9}, {{"TM", 0, 1, 1}, 3.03019*10^9}



First three mode shapes and frequencies for Relative electric permittivity of dielectric =1.76; dielectricThickness = 2.7282494103102 centimeter

{{"TE", 1, 1, 1}, 2.31958*10^9}, {{"TE", 1, 1, 2}, 3.03479*10^9}, {{"TM", 0, 1, 2}, 3.47425*10^9}



First three mode shapes and frequencies for Relative electric permittivity of dielectric =2.3; dielectricThickness = 2.7282494103102 centimeter

{{"TE", 1, 1, 1}, 2.26774*10^9}, {{"TE", 1, 1, 2}, 2.93557*10^9}, {{"TM", 0, 1, 2}, 3.37114*10^9}


« Last Edit: 03/02/2015 07:23 PM by Rodal »

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260
 Perhaps you are getting lower frequencies because you are using Maxwell's equations in 2-D ?, and you are modeling the cavity as a flat plane bounded by a rectangle instead of 3-D cylindrical cavity under 3-D Maxwell's equations

Intuition would say that 2-D would lead to higher frequencies (the opposite of what you get), because there is less possible motion (everything is restricted to the plane).  If natural frequencies are higher in 3-D than 2_D that means that there are more constraints in Maxwell's eqns. in 3-D than in 2-D.  ....Mmmmm....I would need to know more about the 2-Dimensional model, and how it attempts to replicate a 3-Dimensional reality.

This brings back memories from a thought-provoking book, (but it didn't deal with Maxwell's equations in 2-D and 3-D, the people in Flatland are indeed more constrained and everything is stiffer there   :)  ) :

http://www.amazon.com/Flatland-Romance-Dimensions-Thrift-Editions/dp/048627263X/ref=sr_1_1?ie=UTF8&qid=1425320304&sr=8-1&keywords=flatland+abbott
« Last Edit: 03/02/2015 11:24 PM by Rodal »

Offline aero

  • Senior Member
  • *****
  • Posts: 2743
  • 92129
  • Liked: 704
  • Likes Given: 239


Maybe Dr. Rodal will tell me what the resonant frequency should be for the cylindrical cavity with dielectric that I posted data for earlier today?
                                  ... snip ...

                       ... snip ...

GEOMETRICAL INPUT for cylindrical cavity:

bigDiameter = 8.278945 centimeter;
smallDiameter = bigDiameter;
length = 12.24489 centimeter;
dielectricThickness = 2.7282494103102 centimeter;



You can verify this case directly from equation in Wikipedia http://en.wikipedia.org/wiki/Microwave_cavity#Cylindrical_cavity

First four mode shapes and frequencies for Relative electric permittivity of dielectric =1 (No dielectric)

{{"TE", 1, 1, 0}, 2.12223*10^9},
{{"TE", 1, 1, 1}, 2.44998*10^9}, {{"TM", 0, 1, 0}, 2.77191*10^9}, {{"TM", 0, 1, 1}, 3.03019*10^9}



First three mode shapes and frequencies for Relative electric permittivity of dielectric =1.76; dielectricThickness = 2.7282494103102 centimeter

{{"TE", 1, 1, 1}, 2.31958*10^9}, {{"TE", 1, 1, 2}, 3.03479*10^9}, {{"TM", 0, 1, 2}, 3.47425*10^9}



First three mode shapes and frequencies for Relative electric permittivity of dielectric =2.3; dielectricThickness = 2.7282494103102 centimeter

{{"TE", 1, 1, 1}, 2.26774*10^9}, {{"TE", 1, 1, 2}, 2.93557*10^9}, {{"TM", 0, 1, 2}, 3.37114*10^9}


Perhaps you are getting much lower frequencies because you are using Maxwell's equations in 2-D ?, and you are modeling the cavity as a flat plane bounded by a rectangle instead of 3-D cylindrical cavity under 3-D Maxwell's equations

Thank you for running those numbers. You are likely correct in that 2D and 3D do seem to give different answers, although I find the 2D results indicative of what the 3D will be. I will need to run a few 3D samples to see what happens in detail though.
Retired, working interesting problems

Offline frobnicat

  • Full Member
  • ****
  • Posts: 518
  • Liked: 500
  • Likes Given: 151
Dr. March,

I'm trying to model various aspects of the whole system to put upper bounds on thermal effects, and possibly also reconstruct the thrust(t) original signal from the distance(t) given in the charts. It would be a nice boost to this (amateur level) effort if you could confirm either :
- That the flexure bearings have a stiffness of 0.007 in-Lb/deg ? Each ? Both together ? Do you know the exact model reference ?
- That the vertical scale in the charts (indicated in m, around 500) are relevant or not relevant.

I ask this question because I find a contradiction between the stiffness around the vertical axis and the recorded deviation from the 30N calibration pulses (at .007 in-Lb/deg the deviation of the linear displacement sensor would be above 40m, at .014 in-Lb/deg still above 20m). The readings amount for between 1 to 2.5 m for the 30N calibration pulses. So I'm stuck.

While I'm at it : is the plane in which the arm rotates kept as horizontal as possible (ie the axis of rotation as vertical as possible) or is there a small slope voluntarily introduced leading to some pendulum effect against g (for stabilisation or tuning purpose) ? That could explain the varying deviation (in m) for the same calibration pulses thrusts. Also wondered if this is what was implied in this post :
Quote from: Star-Drive
...
These thermally induced actions to the left requires the torque pendulum's arm to move to the right to maintain the balance of the torque pendulum's arm in the lab's 1.0 gee gravity field, since we also use the Earth's g-field to help null the pendulum's movements.
...

Thanks


Frobnicat:

To answer your question:

" - That the flexure bearings have a stiffness of 0.007 in-Lb/deg ? Each ? Both together ? Do you know the exact model reference?"

The two torsion bearings used in or torque pendulum are supposed to have a stiffness of 0.007 in-Lb/deg, +/-10% and is made by the Riverhawk Co. in New York USA.  As to their model number find the data sheet for same attached. 

"- That the vertical scale in the charts (indicated in m, around 500) are relevant or not relevant."

The Philtec D63 fiber-optic displacement sensor measures distance from its target mirror in microns, so the numbers on the left hand side of the force plots measure the distance from the end of the fiber-optic laser head to its mirror target mounted on the torque pendulum arm.  The data sheet for same is attached.

"While I'm at it : is the plane in which the arm rotates kept as horizontal as possible (ie the axis of rotation as vertical as possible) or is there a small slope voluntarily introduced leading to some pendulum effect against g (for stabilization or tuning purpose)?"

The design of our Torque pendulum follows what JPL and Busek Co did at their respective facility, see attached report from Busek.  We found that if we tried to keep the arm completely horizontal though that the pendulum's neutral point would wonder erratically and make alignments near impossible.  So yes I balance the pendulum arm so there is always a slight tilt in it, however this tilt angle magnitude is not controlled as well as it probably should.

Best, Paul M.

Thank you very much for those precious informations. The tilt angle magnitude can probably be inferred from the deviation against the calibration pulses, if we can model the gravitational pendulum component on top of the flexure bearing restoring torque component.

For that we need to know :

Mass of :
  frustum, without dielectric : 1.606 kg
  microwave power amplifier : below 8kg ? 
  faztek horizontal beam : 2.18 Lb (from 1.09Lb/Ft) ?
  Ideally, Total mass with a rough estimate of position of each part...
 
Distances along the arm from vertical axis of rotation to the centre of :
  Long end of arm (frustum side) : 15.5''
  Short end of arm (amplifier side) : 8.5''
  Frustum : 15.5 - 4 = 11.5'' ?
  Electrostatic Fins Calibration System : 15.5-4 = 11.5'' ?
  Linear Displacement Sensor : 15.5-1 = 14.5'' ?
  microwave power amplifier : between 4.25'' and 8.5'' ? 

Stiffness of flexure bearings : .014 in-Lb/deg total  (2 times .007 each)

In short : what is the total mass of the whole rotating assembly, where is the centre of mass of the whole rotating assembly relative to axis of rotation, and what is the moment of inertia around the (almost) vertical axis of rotation (for the later, to assess the dynamics and not just the equilibrium).

green : explicitly provided value
orange : inferred from pictures or derived by me from faztek sellers, to be confirmed
red : not found, do we have better than bounds for those ?


« Last Edit: 03/03/2015 08:37 AM by frobnicat »

Offline zen-in

  • Full Member
  • ****
  • Posts: 531
  • California
  • Liked: 468
  • Likes Given: 365



"- That the vertical scale in the charts (indicated in m, around 500) are relevant or not relevant."

The Philtec D63 fiber-optic displacement sensor measures distance from its target mirror in microns, so the numbers on the left hand side of the force plots measure the distance from the end of the fiber-optic laser head to its mirror target mounted on the torque pendulum arm.  The data sheet for same is attached.

"While I'm at it : is the plane in which the arm rotates kept as horizontal as possible (ie the axis of rotation as vertical as possible) or is there a small slope voluntarily introduced leading to some pendulum effect against g (for stabilization or tuning purpose)?"

The design of our Torque pendulum follows what JPL and Busek Co did at their respective facility, see attached report from Busek.  We found that if we tried to keep the arm completely horizontal though that the pendulum's neutral point would wonder erratically and make alignments near impossible.  So yes I balance the pendulum arm so there is always a slight tilt in it, however this tilt angle magnitude is not controlled as well as it probably should.

Best, Paul M.

The torque pendulum arm had a slight tilt so that alignment was easier.  The Philtec D63 fiber-optic displacement sensor measures distance from its target mirror by measuring the intensity of the reflected light.   If the change in the center of mass reduces the pendulum arm tilt, the light intensity may increase.  This would explain the negative slope of the baseline waveform (apparent movement closer) after the RF is switched off.   It would be interesting to see what the thrust waveform looks like when the cavity is turned around.
« Last Edit: 03/03/2015 06:12 AM by zen-in »

Offline frobnicat

  • Full Member
  • ****
  • Posts: 518
  • Liked: 500
  • Likes Given: 151

"- That the vertical scale in the charts (indicated in m, around 500) are relevant or not relevant."

The Philtec D63 fiber-optic displacement sensor measures distance from its target mirror in microns, so the numbers on the left hand side of the force plots measure the distance from the end of the fiber-optic laser head to its mirror target mounted on the torque pendulum arm.  The data sheet for same is attached.

"While I'm at it : is the plane in which the arm rotates kept as horizontal as possible (ie the axis of rotation as vertical as possible) or is there a small slope voluntarily introduced leading to some pendulum effect against g (for stabilization or tuning purpose)?"

The design of our Torque pendulum follows what JPL and Busek Co did at their respective facility, see attached report from Busek.  We found that if we tried to keep the arm completely horizontal though that the pendulum's neutral point would wonder erratically and make alignments near impossible.  So yes I balance the pendulum arm so there is always a slight tilt in it, however this tilt angle magnitude is not controlled as well as it probably should.

Best, Paul M.

The torque pendulum arm had a slight tilt so that alignment was easier.  The Philtec D63 fiber-optic displacement sensor measures distance from its target mirror by measuring the intensity of the reflected light.   If the change in the center of mass reduces the pendulum arm tilt, the light intensity may increase.  This would explain the negative slope of the baseline waveform (apparent movement closer) after the RF is switched off.   It would be interesting to see what the thrust waveform looks like when the cavity is turned around.

What I understand is that the main vertical axis of rotation Z of balance arm is tilted so that the pendulum gains in stability : if the CoM of the rotating assembly is in front of the axis (frustum side, X+) then the tilt would be to the front, the arm rotates in a plane XY with a slight downward slope toward the frustum (X+). If the CoM of the rotating assembly is back of the axis (amplifier side, X-) then the tilt would be to the back, the arm rotates in a plane XY with a slight downward slope toward the back (X-).



I wouldn't say that a shift in centre of mass would change (significantly) the tilt (ie XY plane's slope) as the flexure bearing have high stiffness around Y and X, but my concern specifically is that a shift in centre of mass (tangential relative to arm rotation, along Y) would change significantly the equilibrium point within the sloped XY plane. It's well known that for movements constrained in a plane with slope Alpha relative to horizontal (ie air cushion inclined table) there is an acceleration equivalent to a reduced gravity of magnitude  g*sin(Alpha).

We know from this post of one chart with the frustum mounted in reverse (small end toward the right, Y+) that after power-on (short and of low magnitude in this case) following drift is upward:


While in the "forward" configuration ( small end of frustum toward the left, Y-) the baseline drift after power-on is generally downward. See for instance from this post


This would imply that this likely thermal drift may not or no longer entirely be explained by a heating of flexure bearings by IR of amplifier : wouldn't direction of such effect don't depend on orientation of frustum ?
Or maybe not, if mounting frustum in reverse alters torque around X... needs coupling analysis, the flexure bearings are not spiral springs...

Anyhow, considerable effort has gone into isolating the flexure bearings from amplifier IR heating, as seen in picture attached with this post :


Such effort were already relatively successful before such thermal shrinkwrapping if we are to judge from the dummy load test  taken before from same post


BTW in this last dummy load test, we don't see the stray 9.6N : "average null force is 9.6 micronewtons due to 5.6A DC current in power cable" figure 20 page 16 in Brady et al report (anomalous...)
Was that cancelled somehow ? Are we supposed non longer to subtract 9.6N to equilibrium thrusts ?

Offline lele

  • Member
  • Posts: 56
  • France
  • Liked: 29
  • Likes Given: 65
What do you think will be needed to prove there is (or not) a real effect, or at least to get enough scientists interested so that a proof or a debunk of the effect is done relatively quickly?
Can something simpler than an on-orbit demonstration be enough?

It's a quite general question which can be asked for any scientific controversy, but since the existence of an EM Drive would be at odd with what 99% of scientists hold to be true, it seems to me they don't want to be caugh anywhere near what seems to be a crackpot theory. Even if building and testing an EM Drive seems relatively cheap and fast*, especially when compared to various fusion initiatives for example.
Actually the fact that it's cheap may not help: it looks like a relatively mundane way to reach fundamentals things of the universe, maybe it would be better accepted if it was a billion-dollars initiative...

I feel a little silly posting that because I'm not adding any concrete value to the thread  :) In any case, I want to thank the people here for working quite hard on this problem and telling us what they're finding, since it's the best way to reach a definitive answer.


*But still an important effort for one person

Offline Stormbringer

  • Full Member
  • ****
  • Posts: 1285
  • Liked: 222
  • Likes Given: 80
well; one of the slides that Mr March provided showed 98 micronewtons. He has also said that due to sensitivity thresholds of the balance equipment at the Glenn Research Center facilities the Eagle Works team need consistent thrust signals of 100 micronewtons before they can send it out GRC for independant replication. So while on the face of it they should be pretty close if the 98 micronewton signal from that slide is consistently obtained; I don't think they have consistent output at that level though.  So they are doing everything they can to improve thier device but there is a deadline of EOM march. Not much time.

In my opinion such a low tempo low priority side project that could probably be fully funded from the couch change from break room lounge chairs should not be micromanaged like it's a multi-billion dollar resource intensive space probe project and put under deadline pressure like that.


« Last Edit: 03/03/2015 12:17 PM by Stormbringer »
When antigravity is outlawed only outlaws will have antigravity.

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260
Having successfully compared our exact solution with both the COMSOL Finite Element solution and the experimental thermal IR camera results presented by Paul March ( http://forum.nasaspaceflight.com/index.php?topic=36313.msg1339803#msg1339803), we now compare our exact solution with Greg Egan's exact solution results ( http://gregegan.customer.netspace.net.au/SCIENCE/Cavity/Cavity.html ), as follows:


Numerical data for example TM modes, r1=2.5 cm, r2=10 cm, θw = 20

GREG EGAN            RODAL

TM011
n = 6.38323            6.38323         
k = 0.861947 cm1         0.861947 cm1
freq = 4.12 GHz         4.11265 GHz

TM012
n = 6.38323            6.38323
k = 1.2847 cm1         1.28475 cm1
freq = 6.13 GHz         6.12998 GHz

TM013
n = 6.38323            6.38323
k = 1.6409 cm1         1.64095 cm1
freq = 7.83 GHz         7.82954 GHz


Numerical data for example TE modes, r1=2.5 cm, r2=10 cm, θw = 20

GREG EGAN            RODAL

TE011
n = 10.4885            10.4885
k = 1.55771 cm1         1.55771 cm1   
freq = 7.438 GHz         7.43236 GHz   

TE012
n = 10.4885            10.4885
k = 1.96024 cm1         1.96024 cm1
freq = 9.359 GHz         9.35299 GHz

TE013
n = 10.4885            10.4885
k = 2.326 cm1         2.32619 cm1
freq = 11.10 GHz         11.099 GHz


The comparison is excellent.  Below is the comparison of the contour plots for the TM (transverse magnetic) modes:



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

Online Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5260
Here is the comparison of the contour plots for the TE (transverse electric) modes



« Last Edit: 03/03/2015 02:00 PM by Rodal »

Tags: