-
#3260 by deuteragenie on 13 May, 2015 20:57
-
Any resemblance with existing or future device(s) is purely coincidental.
Made with MEEP. Does not use the correct frequencies, material, etc. but looks cute.
-
#3261 by deltaMass on 13 May, 2015 21:08
-
An HT Mettler balance is a work of art. Fully mechanical and gets you ten micrograms resolution which is 0.1 uN. Can't be fooled by e/m.
-
#3262 by Rodal on 13 May, 2015 21:41
-
Any resemblance with existing or future device(s) is purely coincidental.
Made with MEEP. Does not use the correct frequencies, material, etc. but looks cute.
Is that a 2-Dimensional model with MEEP?
Maxwell's equations in a flat 2-D surface?
modeling the truncated cone as a FLAT trapezium ?
Is the magnetic field (for TM modes) a point scalar (only able to have + or - sign but the direction is always perpendicular to the surface) instead of being a vector in the azimuthal (circumferential) direction ?
What are we seeing out of the EM Drive? evanescent wave field?
If the answers are yes, do you have enough memory to run a 3D model instead?
Thanks
-
#3263 by aero on 13 May, 2015 22:00
-
Any resemblance with existing or future device(s) is purely coincidental.
Made with MEEP. Does not use the correct frequencies, material, etc. but looks cute.
You might want to increase your skin thickness or Meep resolution. Looks kind of like a Gaussian source with some of the shorter wavelengths bouncing around and longer wavelengths stepping over the boundary (numerically)?
If you are using perfect metal for the skin, then the thickness won't change the result but will help to avoid the model numerically spanning the skin. If you are modelling copper in the GHz range, please tell me about your model as I have struggled for 6 months trying to find a Drude model for copper at 2 GHz. -
#3264 by frobnicat on 13 May, 2015 23:04
-
...
Mmm, yes, for having "probed" that specific aspect of the pendulum system used at EagleWorks, their design is quite stiff : produces small deviation wrt forces. The forces are small and this relatively high stiffness (due to flexure bearings as pivots) doesn't help. Rambling again : the exact apparent stiffness (in µN/µm at end of arm) at EW is a poorly characterised aspect of the experiments, taking the calibration pulses as a reliable starting point gives between 9µN/µm to 40µN/µm across the various released charts, and all are in contradiction with both flexure bearings ratings and harmonic oscillation period (when visible, on underdamped plots). Appears the measures on the vertical scale of plots show much higher stiffness than it should be (too low displacement readings ?).
Anyway, there is no reason not to go with a much lower stiffness system, or no stiffness at all (no position restoring force) and record mm or cm displacements (linear or angular) as thrust accelerates the mass (as in Shawyer).
What measured force at NASA Eagleworks would be high enough to give confidence in the force, in your opinion?
Or there would still be a problem in your opinion with higher measured forces and higher measured displacements, as long as the stiffness remains in that range?
For example would a measured force of 200 µN and a measured displacement of 20 µm giving 10 µN/µm stiffness still lack confidence in your opinion ?
(200 µN is twice the minimum of NASA Glenn's threshold force 100 µN stipulated for measuring the EM Drive)
Concerning the stiffness specifically, my remark that a relatively high stiffness (like the one at EW) "doesn't help" was for measuring displacements in the context of DIY replication attempts. Less stiffness, i.e. more displacement for a given thrust, seems more easy to measure (but at the price of longer time constants).
This is all going a frantic pace... not sure my opinion will matter in a few month but as you can guess me and a lot of people sceptics of the results are waiting for a better characterised/uncharacterised ratio rather than necessarily reaching a given absolute thrust magnitude or displacement threshold. For instance, if experiments with huge particle colliders can reliably claim fundamental discoveries from signals that are sometimes deviations less than 1% of background is because the Standard Model background is so well understood and precisely characterised (by so many people) in the context of the beams and detectors. That and a lot of data and exhaustively recorded and documented parameters to dig and slice into.
Obviously, having an experiment with a better SN ratio from the ground up is always better than recording at bad SN and then subtracting the noise or background effects, because in the latter case validity of discovery rests on the proper characterisation of such background.
Come to think of it, in the unlikely (ahem) event that there is no EM drive effect after all. Then where all those (false) positives would come from ? Necessarily from a methodology that might try to diminish backgrounds that clearly don't look like what is expected (we are trying to reduce the drift thermal effect...) but that will keep the backgrounds that look like what is expected and confounding them with signal : because removing the dielectrics slabs make thrust disappear we keep running with dielectric slabs. And those slabs are fixed with nylons screws that sometimes melt but that appear to do the job otherwise. In the hypothesis that there is no EM drive effect, experiments showing positives will always depend on strange recipes, dark corners, and overall lack of consistency and reliability. Like one shot performances that are too hard to reproduce and experiment with, or thrust that wont show same magnitude when turning test article 180°...
What I fear is that the new experiment at kW range at EW makes more thrust but no better characterised/uncharacterised ratio, that the inconsistencies or poorly characterised aspects of the experiments at 16W and 50W get forgotten or never quite elucidated, and the new experiment introduces new poorly characterised aspects, and so on.
So, what would count as a true positive IMO ? Absence of dark corners, either much better SN ratio (preferably) or much much better justified and recorded and tested background, stability against a range of parameters and phenomenological predictability wrt such parameters, no man in the feedback loop, 100s plots with more parameter studies (like the one on power), stationary or pulsed stationary operation long enough to reach new thermal equilibrium. This takes time, patience, rigoristic methodology, and enthusiasm to discover nothing if there is nothing. Not necessarily millions of $ or kW.
-
#3265 by TheTraveller on 13 May, 2015 23:13
-
After going over Shawyers equations and referencing several microwave info sites, I have come up with what I think is the Shawyer Df equation as attached.
It appears to be different to Dr. Rodal's Df equation as there are no references, that I know of, to cavity length and focuses on the wavelengths (Lambda g1 & Lambda g2) at the 2 end plates.
While Shawyer doesn't give equations for lambda g1 or Lambda g2 as used in his Df equation, I did found the attached reference equations which uses the circular waveguide cutoff frequency to calc end plate wavelength and the other equation to calc cutoff frequency from the circular diameter at each end plate.
Here I note Shawyer says the small end (g2) should operate just above cutoff and the big end (g1) as close to Lambda 0 as possible, while having the thinness cone angle as possible.
Trust this makes some sense. My bed time. Please throw small rocks so I can sleep and dream of microwaves flowing through tapered circular waveguides.
-
#3266 by deltaMass on 13 May, 2015 23:47
-
Quote from: TheTraveller
It appears to be different to Dr. Rodal's Df equation as there are no references, that I know of, to cavity length and focuses on the wavelengths (Lambda g1 & Lambda g2) at the 2 end plates.
How's that again? There is only one wavelength in the cavity - that of the RF. -
#3267 by frobnicat on 13 May, 2015 23:50
-
http://www.radioeng.cz/fulltexts/2011/11_02_472_478.pdf
Attenuation in Rectangular Waveguides with Finite Conductivity Walls
Kim Ho YEAP, Choy Yoong THAM, Ghassan YASSIN, Kee Choon YEONG
RADIOENGINEERING, VOL. 20, NO. 2, JUNE 2011Quote from: Kim Ho YEAP, Choy Yoong THAM, Ghassan YASSIN, Kee Choon YEONGAn important consequence of this
work is the demonstration that the loss computed for degenerate
modes propagating simultaneously is not simply
additive. In other words, the combined loss of two co-existing
modes is higher than adding the losses of two modes
propagating independently. This can be explained by the
mode coupling effects, which is significant when the phase
constants of two propagating modes are different yet very
close.
Nice gold nugget. Not quite understanding yet what mechanism is at the core of the non linearity... is it the ohmic losses in the skin effect ?
Could it allow a net DC current component in the walls from a mix of AC waves ? I'm thinking of a possible coupling with exterior B field (geomagnetic field, or nearby powerful magnetic damping system at EW) : one sine skin current would always average to 0 net Lorentz force, but could sum of sine excitations get a non null net Lorentz force, locally, from a static B field present in vicinity ?
-
#3268 by TheTraveller on 13 May, 2015 23:53
-
Cavity wavelength varies depending on the cavity dimensions.Quote from: TheTravellerIt appears to be different to Dr. Rodal's Df equation as there are no references, that I know of, to cavity length and focuses on the wavelengths (Lambda g1 & Lambda g2) at the 2 end plates.
How's that again? There is only one wavelength in the cavity - that of the RF.
http://www.microwaves101.com/encyclopedias/waveguide-mathematics
-
#3269 by flux_capacitor on 13 May, 2015 23:56
-
Cavity wavelength varies depending on the cavity dimensions.Quote from: TheTravellerIt appears to be different to Dr. Rodal's Df equation as there are no references, that I know of, to cavity length and focuses on the wavelengths (Lambda g1 & Lambda g2) at the 2 end plates.
How's that again? There is only one wavelength in the cavity - that of the RF.
http://www.microwaves101.com/encyclopedias/waveguide-mathematics
Indeed. The increasing confinement of a narrowing waveguide (convergent) produces a widening wavelength and a decrease of the group velocity. Conversely, a widening waveguide (divergent) produces a narrowing wavelength and an increase of the group velocity. -
#3270 by TheTraveller on 14 May, 2015 00:01
-
Cavity wavelength varies depending on the cavity dimensions.Quote from: TheTravellerIt appears to be different to Dr. Rodal's Df equation as there are no references, that I know of, to cavity length and focuses on the wavelengths (Lambda g1 & Lambda g2) at the 2 end plates.
How's that again? There is only one wavelength in the cavity - that of the RF.
http://www.microwaves101.com/encyclopedias/waveguide-mathematics
Indeed. The increasing confinement of a narrowing waveguide (convergent) produces a widening wavelength and a decrease of the group velocity. Conversely, a widening waveguide (divergent) produces a narrowing wavelength and an increase of the group velocity.
The different wavelengths at the big & small ends, due to the different diameters, are what drives Shawyers Df equation.
In fact they are what cause the end plate force differentials that drive the EM Drive.
-
#3271 by TheTraveller on 14 May, 2015 00:30
-
Cavity wavelength varies depending on the cavity dimensions.Quote from: TheTravellerIt appears to be different to Dr. Rodal's Df equation as there are no references, that I know of, to cavity length and focuses on the wavelengths (Lambda g1 & Lambda g2) at the 2 end plates.
How's that again? There is only one wavelength in the cavity - that of the RF.
http://www.microwaves101.com/encyclopedias/waveguide-mathematics
Indeed. The increasing confinement of a narrowing waveguide (convergent) produces a widening wavelength and a decrease of the group velocity. Conversely, a widening waveguide (divergent) produces a narrowing wavelength and an increase of the group velocity.
The different wavelengths at the big & small ends, due to the different diameters, are what drives Shawyers Df equation.
In fact they are what cause the end plate force differentials that drive the EM Drive.
Notice that the expression you have posted above, as defined by Shawyer blows up (goes to infinity) for the denominator going to zero. This occurs for
lambda0 = Sqrt[lambdag1*lambdag2]
or equivalently
cutOffWavelength = Sqrt[lambdag1*lambdag2]
same condition I have above in my message.
_________________________________________
Disclaimer: I don't agree with the description above (since Shawyer's uses a lot of unstated engineering approximations whose validity neither he or anyone else has proven) . I would instead write: according to Shawyer "they are what cause the end plate force differentials that drive the EM Drive."
Will ask Shawyer how SPR calculates big and small end wavelengths Lambda g1 and g2. If he shares that info, then we will have eliminated one unknown. I dislike guessing and dislike even more to reinvent the wheel especially when it involves making actual hardware. Playing mind games or with excel is different. There we can play a bit. But building hardware is serious business. I will do what ever is necessary to reduce uncertainty about operational parameters before the cavity build starts.
Don't understand your Lambda0 equation, which I use as c wavelength in reference medium, our case air.
Each end has it's own Cutoff frequency / wavelength as per these 2 industry equations.
-
#3272 by Rodal on 14 May, 2015 00:40
-
...
Writing technical papers is just as much serious business.
Will ask Shawyer how SPR calculates big and small end wavelengths Lambda g1 and g2. If he shares that info, then we will have eliminated one unknown. I dislike guessing and dislike even more to reinvent the wheel especially when it involves making actual hardware. Playing mind games or with excel is different. There we can play a bit. But building hardware is serious business. I will do what ever is necessary to reduce uncertainty about operational parameters before the cavity build starts.
I also dislike guessing. Readers shouldn't have to guess what an author means when the author submits a "theory paper."
If you have to ask Shawyer how he defined something in his papers is admitting that in your view his papers are insufficiently clear for you.
One thing for sure: Shawyer never stated in his papers how he defines the cut-off frequency. If the cut-off frequency is to be based on the truncated cone cut-off frequency (and not an approximation) one then has to run a a numerical solution (an eigenvalue problem) to obtain the cut-off frequency.
That's a small part of the reason why the engineering/scientific community has issues with his "theory paper". The biggest part is that he makes a large range of assumptions that are not clearly stated and are not clearly supported (certainly his reference to Cullen's paper does not support his Design Factor, Cullen's paper has nothing to do with cavities having different diameter ends).
-
#3273 by TheTraveller on 14 May, 2015 00:50
-
...
Writing technical papers is just as much serious business.
Will ask Shawyer how SPR calculates big and small end wavelengths Lambda g1 and g2. If he shares that info, then we will have eliminated one unknown. I dislike guessing and dislike even more to reinvent the wheel especially when it involves making actual hardware. Playing mind games or with excel is different. There we can play a bit. But building hardware is serious business. I will do what ever is necessary to reduce uncertainty about operational parameters before the cavity build starts.
I also dislike guessing. Readers shouldn't have to guess what an author means when the author submits a "theory paper."
If you have to ask Shawyer how he defined something in his papers is admitting that in your view his papers are insufficiently clear for you.
One thing for sure: Shawyer never stated in his papers how he defines the cut-off frequency. If the cut-off frequency is to be based on the truncated cone cut-off frequency (and not an approximation) one then has to run a a numerical solution (an eigenvalue problem) to obtain the cut-off frequency.
That's a small part of the reason why the engineering/scientific community has issues with his "theory paper". The biggest part is that he makes a large range of assumptions that are not clearly stated and are not clearly supported (certainly his reference to Cullen's paper does not support his Design Factor).
Shawyer pointed me to a reference book that he said had the equations I needed. I Google to find them. The microwave industry has those equations and I suggest Shawyer wrote his theory paper based on someone skilled in the microwave industry knowing how to use microwave industry standard equations to obtain Lambda g1 & g2. Didnt take me long to find them, so not secret squirrel info.
Anyway my bed time. -
#3274 by aero on 14 May, 2015 01:15
-
Doesn't that mean that it blows up for a cylindrical cavity? Yes. And for what other conditions does it blow up?
Lambda02 = Lambdag1 * Lambdag2 : I forget which one is the big end but it blows up whenever
Lambdag-small = Lambda02/Lambdag-big
I agree that the papers would be far more enlightening if the likelihood of such an occurrence were explained. And so would this forum for that matter. -
#3275 by deltaMass on 14 May, 2015 01:20
-
Can we at least assume that lambda1 <= lambda0 <= lambda2 ?
-
#3276 by deltaMass on 14 May, 2015 01:28
-
was that a "yes"?
-
#3277 by TheTraveller on 14 May, 2015 02:05
-
Doesn't that mean that it blows up for a cylindrical cavity? Yes. And for what other conditions does it blow up?
Lambda02 = Lambdag1 * Lambdag2 : I forget which one is the big end but it blows up whenever
Lambdag-small = Lambda02/Lambdag-big
I agree that the papers would be far more enlightening if the likelihood of such an occurrence were explained. And so would this forum for that matter.
Lambda0 = free air
Lambda1 = big end
Lambda2 = small end.
As defined in the attachment dominant circular waveguide cutoff wavelength is 1.7 x diameter. So Lambda1 (big end) is > Lambda2 (small end).
-
#3278 by deltaMass on 14 May, 2015 02:15
-
First people are saying that the big end has a smaller lambda than the small end, and then other people say the opposite. Can't be bothered assembling the quotes.
In any case, if lambda1 < lambda2 (whichever ends they represent), I am still asking if the following is true
lambda1 <= lambda0 <= lambda2
-
#3279 by Rodal on 14 May, 2015 02:20
-
First people are saying that the big end has a smaller lambda than the small end, and then other people say the opposite. Can't be bothered assembling the quotes.
If Shawyer's papers would clearly define the variables, you wouldn't need to ask, would you
In any case, if lambda1 < lambda2 (whichever ends they represent), I am still asking if the following is true
lambda1 <= lambda0 <= lambda2
:
Shawyer's technical paper (IAC- 08 – C4.4.7):
http://www.emdrive.com/IAC-08-C4-4-7.pdf
Shawyer's theory paper:
http://www.emdrive.com/theorypaper9-4.pdf
: