Author Topic: EM Drive Developments - related to space flight applications - Thread 3  (Read 1803785 times)

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2326
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 2956
  • Likes Given: 2589
Yes and as much of early thread 2 will attest, using methods for calculating cylinders will get you close to frustums but no cigar. Also the accuracy of Eagleworks Comsol simulations has been proven accurate using physical measurement (thermal camera for example).

Back then a proper Df equation did not exist nor an understanding how the SPR method of Df > length resonance > thrust (at selected mode) method hangs together.

With respect to COMSOL, so far all it has been able to do is to predict thrust with dielectrics. From my reading of the past comments, COMSOL never predicted any thrust without a dielectric. So if it can't predict thrust without a dielectric, as SPR can, why is it being used to try to model what is happening inside a cavity that has no dielectric?
All this testo engineer argument stuff is not productive.
Now this old gal engineer states that over 40 years in engineering I've seen  design calculations ( I don't care what or who does them) fall a little short time and time again. Bottom line, build the test bed, but be a smart engineer knowing we can't know all and design the flexibility to fine tune it. Simple. This is what I'm doing.
We're dealing with something here that nobody quite knows how it works.

Offline aero

  • Senior Member
  • *****
  • Posts: 2744
  • 92129
  • Liked: 704
  • Likes Given: 239
Yes and as much of early thread 2 will attest, using methods for calculating cylinders will get you close to frustums but no cigar. Also the accuracy of Eagleworks Comsol simulations has been proven accurate using physical measurement (thermal camera for example).

Back then a proper Df equation did not exist nor an understanding how the SPR method of Df > length resonance > thrust (at selected mode) method hangs together.

With respect to COMSOL, so far all it has been able to do is to predict thrust with dielectrics. From my reading of the past comments, COMSOL never predicted any thrust without a dielectric. So if it can't predict thrust without a dielectric, as SPR can, why is it being used to try to model what is happening inside a cavity that has no dielectric?

This line of questioning will not lead anywhere. We use the tools we have and know. If they are "good enough" then our engineering result will be "in the right direction." If others get better engineering results, then maybe their tools are better, maybe their materials are better, maybe their technique is better, only with hindsight will we be able to know.

The only answer is to run experiments and see who gets the better results. Rest assured that as soon as a viable engine with repeatible results comes to the fore, everyone will scrutinize all aspects of the effort, the tools, system and everything imaginable. Continue your experimental efforts following the best path that you know. And if your tools are "good enough" then your results will tell us so. And if your tools are the best that can be, then many folks will be suprised because first generation tools and devices are rarely the best that can be.
Retired, working interesting problems

Offline TheTraveller

...

With respect to COMSOL, so far all it has been able to do is to predict thrust with dielectrics. From my reading of the past comments, COMSOL never predicted any thrust without a dielectric. So if it can't predict thrust without a dielectric, as SPR can, why is it being used to try to model what is happening inside a cavity that has no dielectric?

1) The NASA Eagleworks COMSOL FEA solutions being discussed in this thread are not at all thrust predictions.
They are eigensolutions to the eigenvalue problem: they give the mode shape electromagnetic field distributions, the natural frequencies and the predicted Q.  They are used at CERN and at major companies and academic institutions to predict natural frequencies, mode shapes and cut-off.

2) The approximate cut-off equation you are using from a handbook also does not predict thrust

3) The  approximate natural frequency formula you are using also does not predict thrust.

None of the above solutions deal (solely by themselves) with thrust predictions.

Even if one were to use, for example, MEEP to try to calculate thrust based on evanescent waves, for example, the analyst must keep a clear distinction between on what basis are mode shapes, frequencies and cut-off being predicted, and what is involved in any thrust modeling calculation.   Commingling these solutions together only results in confusion.

Simple question:

What cutoff wavelength, guide wavelength & group velocity does COMSOL, MEEP or some other tool, predict for the 2 ends of the EW frustum when excited in TE012 mode at 2.168GHz?
« Last Edit: 05/31/2015 05:39 PM by TheTraveller »
"As for me, I am tormented with an everlasting itch for things remote. I love to sail forbidden seas.
Herman Melville, Moby Dick

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144
Oh Baby, here is a test stand floating on air, ready for baby EM drive   :)
...

Is this "baby" EM Drive supposed to have lower thrust according to any of the theories? Or is thrust completely based on resonance?


Some theories have thrust inversely proportional to frequency (hence this one at 24 GHz should have ~10 times less thrust than the ones at 2.4 GHz so far tested, based on inverse of linear proportionality alone)

McCulloch's formula  F = PQl/c * (1/w_small - 1/w_big) where l is the cavity length is independent of frequency.  But it is still proportional to power input.

Most theories have thrust proportional to PowerInput  This just has a little battery, so also have to factor out less thrust due to the lower Power Input

So, yes, substantially less thrust, according to those theories.

Higher frequency also means more geometrical attenuation, perhaps that's good, if it also has higher Q to go with it

And you can put a lot of these ones together, and it looks much neater and Hi-Tech  :)

If it works, it can go right away into a CubeSat

Notsosureofit's expression is also independent of frequency when the diameter and the length of the cavity are both scaled to decrease inversely proportional to increasing frequency, in order to maintain the same mode shape.

When the frequency increases by a factor of 10 (24 GHz = 10 * 2.4 GHz), then the diameter needs to be decreased by a factor of 10 and the length needs to be decreased by a factor of 10, in order to keep the same mode shape and thrust, as per Notsosureofit's formula.

When the diameter and the length of the cavity are both scaled  to decrease inversely proportional to increasing frequency, McCulloch's thrust expression also stays invariant.

So, Baby EM Drive by the guy in Aachen, Germany, at 24 GHz will be an extremely interesting test to find out whether McCulloch's and Notsosureofit equations are correct.

IMHO this Baby EM Drive test is the most interesting EM Drive test !!!

The ammonia molecule readily undergoes nitrogen inversion at room temperature. The resonance frequency is 23.79 GHz, corresponding to microwave radiation of a wavelength of 1.260 cm. The absorption at this frequency was the first microwave spectrum to be observed.  Ammonia has been used for Masers at 24 GHz for these reasons.

When using ammonia, safety precautions should be followed:
https://www.health.ny.gov/environmental/emergency/chemical_terrorism/ammonia_tech.htm



It looks like a Innosent CW-Radar modul @24.125GHz:

https://hackaday.io/project/5596-em-drive/log/18379-24-ghz-source-working
http://www.produktinfo.conrad.com/datenblaetter/500000-524999/502371-da-01-en-RADAR_BEWEGUNGSMELDER_RSM_1700.pdf

"Some theories have thrust inversely proportional to frequency (hence this one at 24 GHz should have ~10 times less thrust than the ones at 2.4 GHz so far tested, based on inverse of linear proportionality alone)"

Yes may be 10 times lower or 10 times higher, dont know, but what i know is that these radars have an output power of
typ. +10...20dBm (10...100mW). That is many times lower than in the magnetron case.

If the SMA- connector and cable produce only -3dB insert loss, they have approximately 50mW (in the +20dBm case). And if the resonator inclusive antenna have a other resonance frequence like 24.125GHz the HF-power would be reflected at the antenna.
I am not sure if these modules have got a isolator inside(Can check this tomorrow, i have such module available... ), if not it can be a problem caused by the reflected power  going back to the source. That is a free running oscillator inside! The in and outgoing waves will be mixed and  produce interferences inside the oscillator, it can destabilize the 24GHz output frequency.
I cant see on the fotos if there is some load, hybrid, circulator or circuit elements like that to terminate the HF-Power otherwise.

However good luck with this experiment :)

Offline VAXHeadroom

  • Full Member
  • *
  • Posts: 194
  • Whereever you go, there you are. -- BB
  • Baltimore MD
  • Liked: 255
  • Likes Given: 151
Yes and as much of early thread 2 will attest, using methods for calculating cylinders will get you close to frustums but no cigar. Also the accuracy of Eagleworks Comsol simulations has been proven accurate using physical measurement (thermal camera for example).

Back then a proper Df equation did not exist nor an understanding how the SPR method of Df > length resonance > thrust (at selected mode) method hangs together.

With respect to COMSOL, so far all it has been able to do is to predict thrust with dielectrics. From my reading of the past comments, COMSOL never predicted any thrust without a dielectric. So if it can't predict thrust without a dielectric, as SPR can, why is it being used to try to model what is happening inside a cavity that has no dielectric?
All this testo engineer argument stuff is not productive.
Now this old gal engineer states that over 40 years in engineering I've seen  design calculations ( I don't care what or who does them) fall a little short time and time again. Bottom line, build the test bed, but be a smart engineer knowing we can't know all and design the flexibility to fine tune it. Simple. This is what I'm doing.
We're dealing with something here that nobody quite knows how it works.

We really do need both.

10 years before the Wright brothers flew the first airplane, there were tables compiled (I forget by whom) that showed lift vs drag coefs for a variety of wing shapes.  People had been trying for years to build airplanes from those tables.

Those tables were wrong.

The genius of the Wrights was not really in building the first airplane, it was in building the first wind tunnel so they could test wing shapes and figure out how this stuff REALLY worked.

I feel like that's where we are.  And we need both physical and virtual models so we figure out where the equations are 'Wright' and wrong by building physical devices to test the theories.

This work would progress WAY faster if we had access to a 3D metal printer - I'm guessing bronze would work almost the same as pure copper and you could build a new device in about a day.  Imagine if you could iterate a device every 2-3 days!!!
Emory Stagmer
  Executive Producer, Public Speaker UnTied Music - www.untiedmusic.com

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5261
...

We really do need both.

10 years before the Wright brothers flew the first airplane, there were tables compiled (I forget by whom) that showed lift vs drag coefs for a variety of wing shapes.  People had been trying for years to build airplanes from those tables.

Those tables were wrong.

The genius of the Wrights was not really in building the first airplane, it was in building the first wind tunnel so they could test wing shapes and figure out how this stuff REALLY worked.

I feel like that's where we are.  And we need both physical and virtual models so we figure out where the equations are 'Wright' and wrong by building physical devices to test the theories.

This work would progress WAY faster if we had access to a 3D metal printer - I'm guessing bronze would work almost the same as pure copper and you could build a new device in about a day.  Imagine if you could iterate a device every 2-3 days!!!

To resolve this technical argument all we need is the S21 plot, that's actual experimental data giving the actual natural frequency, instead of model predictions.

Unfortunately the Brady et.al. paper (http://www.libertariannews.org/wp-content/uploads/2014/07/AnomalousThrustProductionFromanRFTestDevice-BradyEtAl.pdf) reported it only up to 2 GHZ, so there is no data to show where the >2GHz natural frequencies occur.

One's expectation is that NASA Eagleworks when testing the TE012 had a S21 plot and thus could determine whether they were testing at a resonance or not and also could determine the loaded Q.

To state that NASA Eagleworks were not testing at a resonance at  2.168GHz (and that they should have tested instead at 2.315 GHz based on handbook-based simple formula spreadsheet calculations http://forum.nasaspaceflight.com/index.php?topic=37642.msg1382477#msg1382477 ) would imply that:

1) NASA "was flying blind". NASA Eagleworks did not have a S21 plot available and Eagleworks didn't know whether they were at resonance at that frequency, or what the loaded Q was at that frequency.

2) that a handbook-based simple formula spreadsheet calculation predicting 2.315 GHz is more reliable as to whether Eagleworks was testing resonance at  2.168GHz than Eagleworks reported experience.


I don't think that's probable.  I would think that NASA had a S21 plot (they had S21 plots for the other modes), and that they could tell what the Q was and that they were at resonance at  2.168GHz.

As to the Wright Brothers example, it would be like saying: the Wright Brothers reported experimental measurement is wrong, "they were flying blind", because this spreadsheet calculation (based on simple formulas from a handbook) says so.
« Last Edit: 05/31/2015 08:39 PM by Rodal »

Offline TheTraveller

A simple but very important question:

What cutoff wavelength, guide wavelength & group velocity does COMSOL, MEEP or some other tool, predict for the 2 ends of the EW frustum when excited in TE012 mode at 2.168GHz?
« Last Edit: 05/31/2015 05:58 PM by TheTraveller »
"As for me, I am tormented with an everlasting itch for things remote. I love to sail forbidden seas.
Herman Melville, Moby Dick

Offline TheTraveller

An equally simple and related question: what Shawyer Design Factor does COMSOL, MEEP or some other tool predict?

I think what you are saying is COMSOL can't predict the small and big end cutoff wavelength, guide wavelength and group velocity values?

If it can't, then for sure it is a nice tool but virtually useless in predicting EMDrive cavity dynamics and related thrust.
"As for me, I am tormented with an everlasting itch for things remote. I love to sail forbidden seas.
Herman Melville, Moby Dick

Offline aero

  • Senior Member
  • *****
  • Posts: 2744
  • 92129
  • Liked: 704
  • Likes Given: 239
An equally simple and related question: what Shawyer Design Factor does COMSOL, MEEP or some other tool predict?

I think what you are saying is COMSOL can't predict the small and big end cutoff wavelength, guide wavelength and group velocity values?

If it can't, then for sure it is a nice tool but virtually useless in predicting EMDrive cavity dynamics and related thrust.

@Traveler - Why don't you just run with that and get on with your build?
Retired, working interesting problems

Offline Mulletron

  • Full Member
  • ****
  • Posts: 1113
  • Liked: 776
  • Likes Given: 1013
« Last Edit: 05/31/2015 09:26 PM by Mulletron »
Challenge your preconceptions, or they will challenge you. - Velik

Offline ThinkerX

  • Full Member
  • ***
  • Posts: 303
  • Alaska
  • Liked: 113
  • Likes Given: 59
Doctor McCulloch is rethinking his concept as to how the EM Drive works within the context of his theory:

http://physicsfromtheedge.blogspot.com/

Beyond me, but some of what he's considering looks comparable to some of what's been posted here. 

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144
Okay, seriously can we get past this back and forth arguing between the traveler and rodal.  Its giving me a headache and its deluding the conversation for us all.

I think it is important to know why my EM Drive Calculator, which is based on microwave industry & SPR equations for Df, shows the EW frustum was in cutoff and not capable of generating thrust, while COMSOL says it was not in cutoff and should. Note here the actual result was no thrust as predicted by the Calculator.

I'm an engineer, I need the numbers to stack up and it they don't, I need to know why. If my Calculator is in error, then I need to fix it. But so far it is predicting what SPR is measuring.

Please understand the small end being in cutoff to the guide wavelength only involves TE01 mode. It does not involve TE012 as that is about 2 x 1/2 waves fitting in between the end plate spacing. So the length mode is not involved in small end cutoff. Only diameter, BesselJ cutoff function at the TE01 excitation mode and external Rf frequency are involved.

Here is the test:

Show where a 0.1588m diameter circular waveguide can propagate a 2.168GHz signal at TE01 mode. All the microwave industry equations say it can't, so why does COMSOL say it can??

If I have this wrong, then the basis of the Df equation is wrong.

Cutoff and guide wavelengths from here:
http://www.tuks.nl/pdf/Reference_Material/Circular_Waveguides.pdf



Cutoff frequency is usable if you want to transmit TEM waves along a waveguide, at cutoff f the strength of the field decrease exponentially along the z axis. In the case of a resonator who is one side is under the cutoff diameter, the sidewall will reflect the wave. I have test this fact experimental one or two years ago. Under this conditions the resonator are still in resonance. My test equipment was an open ended resonator, clear under cutoff diameter at the small end, connected to a networkanalyzer. The Q factor of the this resonance (TE01)was higher if the diameter at this side was smaller! The sidewall play the role of the endplate, its topologically allmost  the same like a endplate in such a situation (with respect to the angle of frustrum sidewalls) !  If the cone is spitz at the end it acts as a reflector and it is still resonant..

edit:
On the other hand, all waves with eigenvalue index(Jnm)p=0 do not travel along the z axis. I think thats why only lower and bigger diameter is necessary not the length in some equations.
« Last Edit: 06/01/2015 05:35 AM by X_RaY »

Offline aero

  • Senior Member
  • *****
  • Posts: 2744
  • 92129
  • Liked: 704
  • Likes Given: 239
I found something similar with Meep images. Of course Meep doesn't calculate why it works, just that it does.
Retired, working interesting problems

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5838
  • USA
  • Liked: 5919
  • Likes Given: 5261
...I think it is important to know why my EM Drive Calculator, which is based on microwave industry & SPR equations for Df, shows the EW frustum was in cutoff and not capable of generating thrust, while COMSOL says it was not in cutoff and should. Note here the actual result was no thrust as predicted by the Calculator.

I'm an engineer, I need the numbers to stack up and it they don't, I need to know why. If my Calculator is in error, then I need to fix it. But so far it is predicting what SPR is measuring.

Please understand the small end being in cutoff to the guide wavelength only involves TE01 mode. It does not involve TE012 as that is about 2 x 1/2 waves fitting in between the end plate spacing. So the length mode is not involved in small end cutoff. Only diameter, BesselJ cutoff function at the TE01 excitation mode and external Rf frequency are involved.

Here is the test:

Show where a 0.1588m diameter circular waveguide can propagate a 2.168GHz signal at TE01 mode. All the microwave industry equations say it can't, so why does COMSOL say it can??

If I have this wrong, then the basis of the Df equation is wrong.

Cutoff and guide wavelengths from here:
http://www.tuks.nl/pdf/Reference_Material/Circular_Waveguides.pdf

TheTraveller, I completely agree with you that it is important to understand where this difference between your calculations and the Finite Element and the exact solutions lies, and you are owed an answer (anyone that gets a headache from reading this  :) doesn't need to read this.  This is what the EM Drives threads are all about: to discuss technical issues).  In these exchanges, we are all pushed to understand what is the accuracy of mathematical models to model the real world.  Once we are done with the discussion, we are both better off because we have understood the mathematical models and the real world a little better. 

I thank you for displaying the equation you are using for cut-off frequency because it shows the problem:



the equation you are using is based on the solution for a cylindrical waveguide It is based on the zeros Xmn of the cylindrical Bessel and the zeros X'mn of the derivative of the cylindrical Bessel functions.

The EM Drive is not a cylinder, it is a truncated cone.,  that is the reason why the equation you are using is inexact.

The cut-off frequency for a truncated cone is not based on cylindrical Bessel functions (that you are using).
The cut-off frequency for a truncated cone is based on spherical Bessel functions and associated Legendre functions.  To find out what modes are cut-off at what frequency for a truncated cone, you have to solve two eigenvalue problems, you cannot just get it from a table of values (like you do when you solve for a cylinder).

The only way you have to improve this would be for you to find out the equation for the cut-off frequency for a truncated cone: which I don't think you are going to find, because there is no closed-form equation for the cut-off frequency for a truncated cone.

Anyone using that cut-off frequency equation (based on a cylinder) is going to make an error when modeling a cone.  COMSOL FEA, MEEP and the exact solution do not make that error because they don't model the truncated cone as a cylinder.


Therefore, I think that the evidence points towards the fact that NASA Eagleworks conducted the TE012 test probably at the correct frequency:

1) Your equation is based on the cut-off frequency for a cylinder, instead of the cut-off frequency for a truncated cone.  Your equation is off by only 6% (it is really not that bad, when considering you are modeling a different geometry).  Your formula fits on a postage stamp, the exact solution takes several lines of Mathematica code.  The FEA solution comprises millions of computer instructions and inverting a huge matrix.

2) NASA would not be "flying blind". NASA Eagleworks probably had a S21 plot available and was able to see the resonance peak and calculated the loaded Q at the frequency they tested, so they knew they were at resonance.
« Last Edit: 05/31/2015 09:23 PM by Rodal »

Offline X_RaY

  • Full Member
  • ****
  • Posts: 774
  • Germany
  • Liked: 993
  • Likes Given: 2144
I got some points to discuss:

1. I think the equation vec_k = vec_p/h_quer   
is the key if we look at the impuls at the endplates, k is clear different in both directions, h_quer is a constant, vec_p has to be different. http://en.wikipedia.org/wiki/Dispersion_relation
1.1. Also interresting are the ideas of some guys here are looking at the poynting vector.. but also interesting is the poynting vector at the sidewall, there is a force vec_r x vec_z if one think geometric! At every reflection there is a z component in just one direction, forward the small diameter.
2. An own idea: If we set the E x H field equal to an ideal gas, it would be compacted in one and expanded in the other direction, in both there is a backreaction always in one z direction, also forward the small diameter. http://en.wikipedia.org/wiki/Ideal_gas

Got somebody ideas to that? ::)
« Last Edit: 06/01/2015 07:42 PM by X_RaY »

Offline TheTraveller

This gold plated Emdrive was a bust:

http://forum.nasaspaceflight.com/index.php?topic=36313.msg1338971#msg1338971

I wonder how the baby Emdrive will do, is it made of steel or aluminum???

http://forum.nasaspaceflight.com/index.php?topic=37642.msg1382129#msg1382129

Edit:
What in the world was the inside of this one made of, and did it have a dielectric insert?
http://forum.nasaspaceflight.com/index.php?topic=37642.msg1380043#msg1380043


Aluminum?????Steel?
And why I'm wondering...
http://forum.nasaspaceflight.com/index.php?topic=29276.msg1275985#msg1275985
http://forum.nasaspaceflight.com/index.php?topic=36313.msg1369555#msg1369555

Only the 1st Experimental EM Drive used a dielectric. The later Demonstrator and Flight Thruster (pictured) did not use a dielectric.

I was advised copper is fine, no coating needed, just have a bright shiny polished finish inside.
"As for me, I am tormented with an everlasting itch for things remote. I love to sail forbidden seas.
Herman Melville, Moby Dick

Offline dustinthewind

  • Full Member
  • ****
  • Posts: 587
  • U.S. of A.
  • Liked: 228
  • Likes Given: 259
...
If I have this wrong, then the basis of the Df equation is wrong.
...


...
The EM Drive is not a cylinder, it is a truncated cone.,  that is the reason why the equation you are using is inexact.
...

1) Your equation is based on the cut-off frequency for a cylinder, instead of the cut-off frequency for a truncated cone.  Your equation is off by only 6% (it is really not that bad, when considering you are modeling a different geometry)....
.

Wasn't Shawyer's equation some how based on cylinders also?  Could it be that the best force is some how slightly off resonance?  Do we have plots of force as frequency is changed from peak resonance to slightly off?  Could throwing in the dielectric slightly throw it off resonance?
« Last Edit: 05/31/2015 10:24 PM by dustinthewind »

Offline RERT

Frobnicat - sorry, my post on page 31 wasn't ignoring your earlier post, I hadn't seen it. I'm afraid that posting from my phone at a children's birthday party precluded scrolling through everything first.

Yes, I quickly realised that Integral(J^B) = (Integral J)^B if B is the constant Earth magnetic field, and so net force wasn't possible. Hence concentrating on the possibility of a couple in the later post.

Further on the same topic, I can estimate the circulating frustrum current in two ways: one is that required to produce the same scale of force measured, and the other is to assume that the input power is all ultimately dissipated in resistance in the copper. Both ways seem to point to a similar magnitude of current, but unfortunately it's in the range of thousands of amps according to my guestimates.

My main thought now is to try and figure out if there is a near-resonant mode for the frustrum which can support that kind of circulating DC current.

I don't really buy KML's idea that dependence on geographical orientation would have been noticed unless someone was looking for it. It would just add to the variability of measured thrust between runs.


Offline sneekmatrix

Bright shiny finish sounds like titanium oxide

Offline TheTraveller

...I think it is important to know why my EM Drive Calculator, which is based on microwave industry & SPR equations for Df, shows the EW frustum was in cutoff and not capable of generating thrust, while COMSOL says it was not in cutoff and should. Note here the actual result was no thrust as predicted by the Calculator.

I'm an engineer, I need the numbers to stack up and it they don't, I need to know why. If my Calculator is in error, then I need to fix it. But so far it is predicting what SPR is measuring.

Please understand the small end being in cutoff to the guide wavelength only involves TE01 mode. It does not involve TE012 as that is about 2 x 1/2 waves fitting in between the end plate spacing. So the length mode is not involved in small end cutoff. Only diameter, BesselJ cutoff function at the TE01 excitation mode and external Rf frequency are involved.

Here is the test:

Show where a 0.1588m diameter circular waveguide can propagate a 2.168GHz signal at TE01 mode. All the microwave industry equations say it can't, so why does COMSOL say it can??

If I have this wrong, then the basis of the Df equation is wrong.

Cutoff and guide wavelengths from here:
http://www.tuks.nl/pdf/Reference_Material/Circular_Waveguides.pdf

TheTraveller, I completely agree with you that it is important to understand where this difference between your calculations and the Finite Element and the exact solutions lies, and you are owed an answer (anyone that gets a headache from reading this  :) doesn't need to read this.  This is what the EM Drives threads are all about: to discuss technical issues).  In these exchanges, we are all pushed to understand what is the accuracy of mathematical models to model the real world.  Once we are done with the discussion, we are both better off because we have understood the mathematical models and the real world a little better. 

I thank you for displaying the equation you are using for cut-off frequency because it shows the problem:



the equation you are using is based on the solution for a cylindrical waveguide It is based on the zeros Xmn of the cylindrical Bessel and the zeros X'mn of the derivative of the cylindrical Bessel functions.

The EM Drive is not a cylinder, it is a truncated cone.,  that is the reason why the equation you are using is inexact.

The cut-off frequency for a truncated cone is not based on cylindrical Bessel functions (that you are using).
The cut-off frequency for a truncated cone is based on spherical Bessel functions and associated Legendre functions.  To find out what modes are cut-off at what frequency for a truncated cone, you have to solve two eigenvalue problems, you cannot just get it from a table of values (like you do when you solve for a cylinder).

The only way you have to improve this would be for you to find out the equation for the cut-off frequency for a truncated cone: which I don't think you are going to find, because there is no closed-form equation for the cut-off frequency for a truncated cone.

Anyone using that cut-off frequency equation (based on a cylinder) is going to make an error when modeling a cone.  COMSOL FEA, MEEP and the exact solution do not make that error because they don't model the truncated cone as a cylinder.


Therefore, I think that the evidence points towards the fact that NASA Eagleworks conducted the TE012 test probably at the correct frequency:

1) Your equation is based on the cut-off frequency for a cylinder, instead of the cut-off frequency for a truncated cone.  Your equation is off by only 6% (it is really not that bad, when considering you are modeling a different geometry).  Your formula fits on a postage stamp, the exact solution takes several lines of Mathematica code.  The FEA solution comprises millions of computer instructions and inverting a huge matrix.

2) NASA would not be "flying blind". NASA Eagleworks probably had a S21 plot available and was able to see the resonance peak and calculated the loaded Q at the frequency they tested, so they knew they were at resonance.

The way my EM Drive Calculator calculates the effective guide wavelength is as per SPR directions as how their in-house software works. Was told to calculate the cutoff wavelength and the resultant guide wavelength for say 1,000 point along the frustum side wall and then numerically integrate them into the effective guide wavelength.

The SPR Df is based on using the industry standard cylindrical waveguide cutoff and guide equations I have provided, at each point of diameter. From the replicant Flight Thruster example I provided to SPR, their Df and mine are the same (at the same frequency) to at least 4 decimal points. Our resonance frequency (end plate spacing) is out by 0.5%, which I consider close enough.

While SPR may be using a different approach that you would have followed, it is how SPR do their calcs and get their Df, effective guide wavelength and external Rf frequency (end plate spacing) needed to get resonance at the selected excitation mode and external Rf frequency.

Quote
Therefore, I think that the evidence points towards the fact that NASA Eagleworks conducted the TE012 test probably at the correct frequency

As far as I know, EW did not use the SPR design solution to determine the best frequency for their frustum. Shawyer has said he will assist them to get this right. From the example SPR shared with me, that can now be done with the EM Drive Calculator as it matches the SPR inhouse Df and external Rf frequency versus end plate spacing and mode selection.
« Last Edit: 05/31/2015 11:23 PM by TheTraveller »
"As for me, I am tormented with an everlasting itch for things remote. I love to sail forbidden seas.
Herman Melville, Moby Dick

Tags: