-
#1300 by RotoSequence on 10 Mar, 2015 05:03
-
Is it a good idea to suggest theories of operation at the current level of understanding of EM drive's operational parameters? I suppose a relatively straightforward experiment for testing electrogravitation would be to place the EM drive test article near a gravimeter (or other physically separated accelerometer), switch it on, and see if/how the gravimeter's readings change.
-
#1301 by Mulletron on 10 Mar, 2015 07:32
-
I haven't pointed out the reason why I'm digging into cavity QED and cavity optomechanics. What interests me is the well known strong interaction between light and matter.... the strong coupling observed within cavities is what I'm interested in.
Which also includes vacuum fluctuations and Casimir-Polder forces as indicated by other work I dug up earlier in the thread.
That momentum transfer from the qv work that was dug up can't stand on its own outside of this framework.
I really wish I could get somebody else on board with this idea because I absolutely will fail on my own. -
#1302 by frobnicat on 10 Mar, 2015 09:50
-
@Star-Drive :
I'm still asking about the huge discrepancy between the expected linear displacement of 32.3µm (assuming only flexure bearings restoring torque) in response to a calibration pulse of 29.1µm and the actual readings on chart vertical scale of a linear displacement between 1µm to 2.5µm for the calibration pulses.
There is an inconsistency of an order of magnitude between values as expected from flexure stiffness and as recorded on the display. Is it acknowledged or investigated by the team at Eagleworks ?
I'm not objecting that the thrusts measurements would not be proportional (ie for twice the deviation of cal. pulse => twice 29.1µN) but since the fixed ratio of µm displacement per µN of thrust is at the heart of the experiment, such discrepancy can only weaken the case for the charts published so far. This needs to be clarified anyhow. It could be a problem of calibration of the Philtec D63 gain, or a biased scaling factor between the analog outputs of the D63 and the final rendering of vertical scale on display...
-
#1303 by RotoSequence on 10 Mar, 2015 10:04
-
I really wish I could get somebody else on board with this idea because I absolutely will fail on my own.
I'm sorry that I can't be of much help in this area, but for what it's worth, I think your theories are worth exploring by those who have the knowledge, background, and ability to do so. -
#1304 by Notsosureofit on 10 Mar, 2015 11:20
-
@ RODAL
Just got a minute but from your p expression;
If L1/c1 = L2/c2
del f = (1/2*f)*((c1*c2)/(L1*L2))*b^2*((1/dD1^2)-(1/dD2^2))
might be a solution ??
Got to check the thinking later.
Night !
I find your previous expression
del f = ( f/(2*c^2)) * (c1^2-c2^2)
more physically appealing, since it goes to zero for equal dielectric constants, regardless or their dielectric length,
while on the other hand
del f = (1/2*f)*((c1*c2)/(L1*L2))*b^2*((1/dD1^2)-(1/dD2^2))
goes to zero for equal dielectric lengths, regardless of their dielectric constants.
The previous expression is only valid approximation for a "uniformly varying dielectric". There is no L1 and L2 in that case.
What do you think might maximize the second expression ? (valid only for L1/c1 = L2/c2 )
-
#1305 by Mulletron on 10 Mar, 2015 11:47
-
This is a link to another type of resonant cavity microwave thruster that DOES use propellant. I'm sharing it because the paper parallels what is being discussed in this thread to a great degree, such as calculating resonant modes, COMSOL analysis...etc. There is a bunch of useful stuff in there.
https://etda.libraries.psu.edu/paper/12292/7584
Edit: Unshortened the link. Google had a mile long search string to it. Some people are having ssl problems with the link, so it is attached. I'd rather save NSF the bandwidth by not downloading the large PDF from here if possible.
DESIGN AND DEVELOPMENT OF A 30-GHz MICROWAVE ELECTROTHERMAL THRUSTER
A Thesis in Aerospace Engineering
by
Erica E. Capalungan
-
#1306 by Mulletron on 10 Mar, 2015 12:06
-
http://forum.nasaspaceflight.com/index.php?topic=29276.msg1274400#msg1274400
Quoting from the Chinese paper linked to in the above post from @AeroQuoteWith the Small End of the cavity unchanged, the quality factor and thrust decrease with the increase in the Large End.
Ok, so with that in mind, why is Shawyer progressively opening up the angle of this designs? What gives?
http://forum.nasaspaceflight.com/index.php?topic=36313.msg1336811#msg1336811
-
#1307 by Star-Drive on 10 Mar, 2015 12:19
-
@Star-Drive :
I'm still asking about the huge discrepancy between the expected linear displacement of 32.3µm (assuming only flexure bearings restoring torque) in response to a calibration pulse of 29.1µm and the actual readings on chart vertical scale of a linear displacement between 1µm to 2.5µm for the calibration pulses.
There is an inconsistency of an order of magnitude between values as expected from flexure stiffness and as recorded on the display. Is it acknowledged or investigated by the team at Eagleworks ?
I'm not objecting that the thrusts measurements would not be proportional (ie for twice the deviation of cal. pulse => twice 29.1µN) but since the fixed ratio of µm displacement per µN of thrust is at the heart of the experiment, such discrepancy can only weaken the case for the charts published so far. This needs to be clarified anyhow. It could be a problem of calibration of the Philtec D63 gain, or a biased scaling factor between the analog outputs of the D63 and the final rendering of vertical scale on display...
Frobnicat & Crew:
Dr. White and his NASA interns are the folks who performed the original force calibration work on this torque pendulum, so your questions might be bettered answered by him. However, it's my understanding that the torque pendulum's actual micron displacement observed for each test run is dependent on its specific total active mass load, balance weights and all their locations on the torque pendulum arm for the test run in question. So as long as we reference the near constant calibration force from our electrostatic fin calibration system before and after each test run, and then use that specific displacement yardstick of the moment as the true measure of the test article's generated forces, it doesn't matter what the actual micron displacement turns out to be for each data run. And that has been what we've used to date report our generated forces. If there is a major problem with that approach please let us know.
Best, Paul M. -
#1308 by Rodal on 10 Mar, 2015 12:23
-
I see the wisdom in your hint that we should approach this starting from an earlier point, with the resonating field as a function of the longitudinal polar coordinate variable. In the case of only one dielectric the field is just a harmonic function of the longitudinal variable, hence your idea to perhaps do a Fourier series expansion.@ RODAL
Just got a minute but from your p expression;
If L1/c1 = L2/c2
del f = (1/2*f)*((c1*c2)/(L1*L2))*b^2*((1/dD1^2)-(1/dD2^2))
might be a solution ??
Got to check the thinking later.
Night !
I find your previous expression
del f = ( f/(2*c^2)) * (c1^2-c2^2)
more physically appealing, since it goes to zero for equal dielectric constants, regardless or their dielectric length,
while on the other hand
del f = (1/2*f)*((c1*c2)/(L1*L2))*b^2*((1/dD1^2)-(1/dD2^2))
goes to zero for equal dielectric lengths, regardless of their dielectric constants.
The previous expression is only valid approximation for a "uniformly varying dielectric". There is no L1 and L2 in that case.
What do you think might maximize the second expression ? (valid only for L1/c1 = L2/c2 )
I have to take care of some $$$ paying work first. I'll eventually be back at this in a more comprehensive way. -
#1309 by Mulletron on 10 Mar, 2015 12:37
-
http://www.emdrive.com/NWPU2010translation.pdf
In the conclusion area, last sentence in 3rd paragraph, next to last page.QuoteWith the Small End of the cavity unchanged, the quality factor and thrust decrease with the increase in the Large End.
Seems to be a summary of this on page 8:QuoteBy keep the diameter of the Small End constant, increase the large end of the cavity, in order to have the same resonant frequency, cavity height must be reduced, quality factor also reduce.
-
#1310 by Rodal on 10 Mar, 2015 13:13
-
http://www.emdrive.com/NWPU2010translation.pdf
OK, I took a look at the above quotations.
In the conclusion area, last sentence in 3rd paragraph, next to last page.QuoteWith the Small End of the cavity unchanged, the quality factor and thrust decrease with the increase in the Large End.
Seems to be a summary of this on page 8:QuoteBy keep the diameter of the Small End constant, increase the large end of the cavity, in order to have the same resonant frequency, cavity height must be reduced, quality factor also reduce.
THESE QUOTED STATEMENTS ARE NOT BASED ON ANY EXPERIMENTS
It is clear from the text that these statements are only based on their Finite Element analysis numerical solution of the eigenvalue problem to solve Maxwell's equations, from which the author's derive a thrust force. The authors then change the dimensions in their finite element model, which gives different values of the numerically calculated thrust force.
It is trivial to show that a solution that is based only on solving the eigenvalue problem of Maxwell's equations in a truncated cone cavity cannot result in a thrust force on the center of mass. Greg Egan shows this with closed-form exact equations ( http://gregegan.customer.netspace.net.au/SCIENCE/Cavity/Cavity.html ). The thrust force should be zero. This is the reason why prominent physicists and academia are so skeptical of the EM Drive, as a thrust force from such a propellant-less cavity would violate conservation momentum (unless there is coupling with an external field -external fields which the authors did not take into account in their numerical analysis-)
The fact that the authors obtain a thrust force from the solution of the eigenvalue problem for Maxwell's equations (without any consideration of external coupling fields) can be due to ill-conditioning of the eigenvalue matrix (inverted in the FEM solution) and/or problems with improper setting of boundary conditions (as their FEM analysis cannot deal with rigid body modes and they have to be suppressed to get a solution), or it could be due for example to reduced integration of the finite element model (for example spurious deformation modes like the hourglass mode of finite elements that finite element experts are familiar with, but that users of finite element packages may be unaware of).
In a few words:
*the statements are not based on experiments
*the statements are based on a numerical solution that should have given zero thrust. Since the numerical solution gave a finite thrust (with extremely small magnitude ! ), the conclusion that increasing the cone angle decreases the thrust force has to be treated as an unsupported outcome of their incorrect numerical solution.
NOTE: The quotation is still very interesting as the flawed numerical analysis by Juan Yang et.al. comes to a completely different conclusion than R. Shawyer's equation (as well as the equations of McCulloch and Notsosureofit). The fact that R. Shawyer prominently uses Juan Yang's reference in his website to validate the EM Drive, while Juan Yang's numerical results contradict Shawyer's (regarding the angle of the truncated cone) is noteworthy.
Also, if Juan Yang believes her numerical results, why didn't she decrease the angle of the truncated cone (or downright use a cylindrical cavity instead of a truncated cone in her later experiments?). If anything the angle of the truncated cone used by Juan Yang seems to have increased rather than decreased in her more recent experiments.
-
#1311 by Notsosureofit on 10 Mar, 2015 15:21
-
#1312 by frobnicat on 10 Mar, 2015 17:03
-
@Star-Drive :
I'm still asking about the huge discrepancy between the expected linear displacement of 32.3µm (assuming only flexure bearings restoring torque) in response to a calibration pulse of 29.1µm and the actual readings on chart vertical scale of a linear displacement between 1µm to 2.5µm for the calibration pulses.
There is an inconsistency of an order of magnitude between values as expected from flexure stiffness and as recorded on the display. Is it acknowledged or investigated by the team at Eagleworks ?
I'm not objecting that the thrusts measurements would not be proportional (ie for twice the deviation of cal. pulse => twice 29.1µN) but since the fixed ratio of µm displacement per µN of thrust is at the heart of the experiment, such discrepancy can only weaken the case for the charts published so far. This needs to be clarified anyhow. It could be a problem of calibration of the Philtec D63 gain, or a biased scaling factor between the analog outputs of the D63 and the final rendering of vertical scale on display...
Frobnicat & Crew:
Dr. White and his NASA interns are the folks who performed the original force calibration work on this torque pendulum, so your questions might be bettered answered by him. However, it's my understanding that the torque pendulum's actual micron displacement observed for each test run is dependent on its specific total active mass load, balance weights and all their locations on the torque pendulum arm for the test run in question. So as long as we reference the near constant calibration force from our electrostatic fin calibration system before and after each test run, and then use that specific displacement yardstick of the moment as the true measure of the test article's generated forces, it doesn't matter what the actual micron displacement turns out to be for each data run. And that has been what we've used to date report our generated forces. If there is a major problem with that approach please let us know.
Best, Paul M.
Well, the pendulum is basically a device that, ideally, converts force into displacement. What is measured is displacement D=Cst*F. The electrostatic calibration system looks like a robust way to achieve a stable reference force Fcal of 29.1µN. So there is no question that Dthrust=Cst*Fthrust together with Dcal=Cst*Fcal yields Fthrust=Fcal*Dthrust/Dcal where Cst cancels. That's a good point for the design, there is no need to know Cst precisely. But not needing to know precisely is one thing, having an order of magnitude discrepancy in the absolute value of Cst is another. Cst is at the core of the conversion of µN (what we want to know) into µm (what we measure) : at the very least there is a central aspect of the balance that seems very poorly characterised.
Add to that the fact that the whole apparatus is slightly tilted. With the actual vertical scale charts readings (that are in contradiction with both the known flexure bearing stiffness and the natural oscillation period of ~4.5s which is clearly visible on some charts) there is no way to reconstruct this tilt from the data. You seem to imply this tilt is quite low, nevertheless it does play a significant role in the rest equilibrium point of the balance since a perfectly horizontal setting wouldn't allow to have a stable rest position (which makes perfect sense since there is no angular tuning at the axis and flexure bearings rest will drift thermally and with loads). So gravity plays a role. So thermal displacements in centre of mass of a part relative to fixation point can have a significant impact on the rest position, recording as sustained displacements at the LDS. Frankly, from the shapes of the responses to the tests I don't really believe that thermal displacements alone can explain the whole response, but others may be more finicky about that, and to clearly assess those aspects needs a correct characterization of the Cst between µN thrusts and µm readings. For instance if vertical scale is showing 1µm when really it should show 10µm then one would have to explain 300g moving only 0.1mm instead of 300g moving 1mm. The correction (if it is indeed needed) would actually make it harder to explain effects from thermal displacements.
A change of stiffness at the level of flexure bearings due to axial load conditions can't explain the 4.5s period in the charts. I can't find Riverhawks charts indicating how stiffness would change with axial load, but explaining the 1µm reading for 29.1µN cal. as due to an added stiffness (30 times more) depending on axial load would be in direct contradiction with the dynamics of the chart and using moment of inertia from mass distribution (even if it's uncertain up to 50%).
This very apparent discrepancy could be due to a simple scaling factor in display, it could also indicate a problem in the operating conditions of the LDS, with an LDS working not around 500µm but nearer the peak, at reduced sensitivity and reduced linearity.
So while I wouldn't qualify the problem as major as far as Fthrust=Fcal*Dthrust/Dcal is concerned, it would still appear as a manifest and serious consistency problem for anyone looking in some depth at the data and requiring a proper characterization of the system. It could be also indicative that LDS is operating at less than ideal conditions and that alone would make it worth further inquiry.
Side note : I'm only an anonymous contributor trying to understand what is going on, I do have a background in mechanical engineering, including metrology, but can't claim this is my professional activity (neither in aerospace). I do believe my arguments are correct and worth of consideration, but this does not preclude a blunder somewhere. I understand with limited time and resources the team at Eagleworks has to weigh the priorities. Anyhow, thank you for taking time reading and answering my concerns.
-
#1313 by Rodal on 10 Mar, 2015 17:21
-
We've been laser-focused on materials with greater dielectric constants, but what about materials with similar dielectric constants, like fused quartz? Will the EM drives behave differently?
If Paul March discussed testing with dielectric materials other than Teflon and HD PE, I don't recall. It would be interesting if Paul could comment (or if Paul already discussed this, if somebody could bring the experimental results to our attention).
I understand that Roger Shawyer tested non-polymer materials as dielectrics, but the specific results and the dimensions and material properties of the dielectrics tested were not disclosed (again, if anyone has more specific details, please bring them to our attention).
Dr. Rodal:
We've only tried polyethylene, Teflon, neoprene rubber and aluminum oxide discs so far, with PE and PTFE being the most productive. However what dielectric if any will prove to be optimal in generating the most thrust in these EM-Drive like thrusters is really dependent on what physics is really driving their operation. So far the dielectrics with the largest electrostrictive coefficient combined with a largest Q-factor appear to be the winners. This implies to me that fused quartz may be a good candidate, since it has a large Q-factor with moderate electrostrictive coefficient and piezoelectric responses.
BTW, these dielectrics may prove to be the E&M/gravity field to mechanical converters needed to generate thrust. On the other hand if Shawyer and the Chinese are right in their statements that they used no dielectrics in their tens to hundreds of milli-Newton thrusters, then these dielectrics may just be means of amplifying the underlying effects that are generated just by the action of the E&M fields on the copper or silver atoms in the walls of the frustum resonant cavity. Remember that though copper and silver only have a real permittivity of 1.0, in temporal space they have a complex permittivity of greater than 100. and this is the parameter that drives E to B-field phase shifting over very small distances at microwave frequencies, (~2 microns deep at 2.0 GHz). I.e. these metallic atoms can undergo very large cyclic accelerations around their crystal lattice positions as the E&M wave fronts are dissipated in them.
Thank you, Paul. Great, very valuable information.
Having personally worked on R&D compounding of rubber and thus being very much aware of the great variety of possible properties for neoprene rubber depending on compounding, I would not draw too much general conclusions based on results for particular Neoprene rubber samples (without knowing the specific compound tested).
The information on (ceramic) aluminum oxide discs being inferior (regarding the measured thrust force) to polymers like PTFE and HD PE is very valuable, particularly in light of reports like the following that have comprehensively characterized the dielectric properties of Aluminum Oxide over a very broad frequency range utilizing multiple analytical testing techniques:
Broadband Dielectric Characterization of Aluminum Oxide (Al2O3)
Khalid Z. Rajab, Mira Naftaly, Edmund H. Linfield, Juan C. Nino, Daniel Arenas, David Tanner, Raj Mittra, and Michael Lanagan
J. Micro. and Elect. Pack. 5, 101–106 (2008)
http://www.phys.ufl.edu/~tanner/PDFS/Rajab08jmep-Al2O3.pdf
From the few reports where I saw the dielectric materials tested by R. Shawyer, what I recall is Shawyer using inorganic dielectric materials. This is important to understand the reports that Shawyer has abandoned the use of dielectric materials. Without knowing specifically what dielectric materials did Shawyer abandon, the fact that he abandoned them is not that useful. -
#1314 by Notsosureofit on 10 Mar, 2015 17:39
-
Good 10GHz data !
-
#1315 by Mulletron on 10 Mar, 2015 20:24
-
I really wish I could get somebody else on board with this idea because I absolutely will fail on my own.
I'm sorry that I can't be of much help in this area, but for what it's worth, I think your theories are worth exploring by those who have the knowledge, background, and ability to do so.
I just need to more rigorously show applicability of these concepts presented in the literature I've been posting. I'm getting into the realm where I can't find any further precedent in which I can deep dive into because the applicable work simply hasn't been written yet (or I can't find it). I'm not interested in finding anything other than established theory so this might be the wall. I'm really searching for that clincher that will be undeniable. -
#1316 by aero on 10 Mar, 2015 22:06
-
@Dr. Rodel - Did you use your magic formulas to calculate the resonance frequencies of the newest Eagleworks cavity from Paul? Including dielectric effects? If so, what were the numbers?
-
#1317 by Rodal on 10 Mar, 2015 22:21
-
@Dr. Rodel - Did you use your magic formulas to calculate the resonance frequencies of the newest Eagleworks cavity from Paul? Including dielectric effects? If so, what were the numbers?
I obtained exact solutions so far for:
1) Truncated cone with only one dielectric medium in the cavity
2) Cylindrical cavity with two dielectric coupled mediums
I haven't had the time to derive the formula for the Truncated Cone with two dielectric mediums to consider the case with vacuum and HD-PE.
I did calculate the Eagleworks cavity without the HD-PE dielectric section. -
#1318 by aero on 10 Mar, 2015 22:26
-
@Dr. Rodel - Did you use your magic formulas to calculate the resonance frequencies of the newest Eagleworks cavity from Paul? Including dielectric effects? If so, what were the numbers?
I obtained exact solutions so far for:
1) Truncated cone with only one dielectric medium in the cavity
2) Cylindrical cavity with two dielectric coupled mediums
I haven't had the time to derive the formula for the Truncated Cone with two dielectric mediums to consider the case with vacuum and HD-PE.
I did calculate the Eagleworks cavity without the HD-PE dielectric section.QuoteI did calculate the Eagleworks cavity without the HD-PE dielectric section.
And what number did you get? -
#1319 by Rodal on 10 Mar, 2015 22:39
-
....
See http://forum.nasaspaceflight.com/index.php?topic=36313.msg1339803#msg1339803QuoteI did calculate the Eagleworks cavity without the HD-PE dielectric section.
And what number did you get?
The frequency for the TM221 mode shape (without the dielectric) is 2.00709 GHz.
Is there another particular mode you are interested to know the frequency for?