-
#1460 by zellerium on 08 Jan, 2016 04:04
-
Roger shared some advise with us about real world effects of trying to get good thrust.
1) don't use a dielectric in the frustum.
2) avoid making the small end too narrow such that a constant diameter waveguide of that diameter, excited in the mode of choice, at the frequency of choice, would be at or below cutoff using standard microwave industry equations. His advise is that ideally the small end diameter should be just a bit bigger diameter that cutoff diameter.
You are here building EmDrives because of his work. Your choice to take his advise notice or not.Quote from: Roger Shawyer emailHi Phil
Your proposals sound fine to me.
Note that the Q you achieve will also be dependent on how well you tune and match the impedance of the input antenna. We have used probe, loop and waveguide iris plates as input circuits. All have their own problems, but you should first calculate the wave impedance of the cavity at the input position. Standard text book equations work, as they always do. You can then design your chosen input circuit to match the wave impedance at the cavity resonant frequency.
All successful EmDrive thrusters that I know of have incorporated a tuning element of some sort at the input.
Also no successful design used COMSOL without correction, as the software does not seem to cope with conditions close to cut-off, as NASA should have realised.
Best regards
Roger
I agree with 1) but what is the evidence that leads to 2)?
Why does Shawyer suggest to keep the small end above cutoff? -
#1461 by mwvp on 08 Jan, 2016 08:01
-
What happens when you're past cutoff?
"Electromagnetic fields and transmission properties in tapered hollow metallic waveguides Xiahui Zeng* and Dianyuan Fan" (bottom) http://emdrive.wiki/Evanescent_waves
Has propagation constant curves, from which you can calculate the dispertion from the slope of beta/k.
This tells you on each reverberation of an impulse, what the attenuation and phase shift will be. And group delay, Vg.
Speaking of dispertion
What is dispersion? From Bradshaw:Quotedispersive means:
the refractive index varies with frequency
group velocity differs from phase velocity
group index differs from refractive index
...
k Angular wave number (ωn(ω)/c)
n Real refractive index (Re[√er μr])
ng Group refractive index or just group index (c/vg )
vg Group velocity (dω/dk)
...
So, we must understand group velocity and group index
Writing k = nω/c and V = c/n, vg = dω/dk
ng = c/vg = c * (1/vg) = c * dk/dω
group velocity is the ratio of infinitesimal changes in temporal periodicity to infinitesimal changes in spatial periodicity
I don't like dealing with refractive index (n). I prefer velocity factor (1/n).
How does dispersion work in the frustrum?
Thrust is generated by the large radiation pressure in the frustrum becoming unbalanced at each end with unbalanced dissipation. Thermal radiation pressure is miniscule and diffused.
Frustrum acceleration is required to create the doppler spreading.
Cavity Q and dispersion determines the least acceleration that produces the greatest frequency separation of the doppler-spread microwaves.
Crudely speaking with a poor metaphor, the dispersion is like a nozzle; the doppler-spread sidebands thus separate through the dispersion/frequency gradient.
Q determines how much pressure can build up in the "combustion chamber".
Since power, Q, dispersion, and acceleration determine how much unbalanced radiation pressure there is, it could be very easy to make a dud engine.
If my conjecture is right, it is predictable by a model that accounts for physical acceleration of the frustrum. I can think of a crude way to do that with spice, using a lumped-element model of the waveguide, with mixers at each end to make the doppler shift. That will produce the waveguide signal, and parameters, but the pointing vector/radiation pressure would be crudely inferred by voltage and currents on the L-R-C elements.
Should be simply tested through a two or even one D Meep simulation, but there's the acceleration to deal with. Perhaps making 3 copies, past-present-future of the mesh, and compressing the blue end and expanding the red end and swapping the matrix?
Oops, I forgot to say, notice the steepness of the beta/k curves in Zeng & Fan (above)? You want to be there because you get the biggest phase shift and doppler spreading. The equation for doppler spreading could be from https://en.wikipedia.org/wiki/Fizeau_experiment
see https://upload.wikimedia.org/math/b/6/8/b688f7f7b2c8e7643fcc94870371cb5e.png
and notice the term for dispersion. I could no doubt do better by digging through Bradshaw but its late, sorry.
-
#1462 by Mulletron on 08 Jan, 2016 08:06
-
Want to make sure guidance from Shawyer (via Aero) isn't lost. Here's a blast from the past:
http://forum.nasaspaceflight.com/index.php?topic=29276.msg1275094#msg1275094QuoteShawyer's statement, "The small end diameters are set just above the cut-off diameter corresponding to the mode and frequency of the design."
-
#1463 by TheTraveller on 08 Jan, 2016 08:45
-
I agree with 1) but what is the evidence that leads to 2)?
Why does Shawyer suggest to keep the small end above cutoff?
It is stated in Roger's thrust equation:
Thrust = (2 * Qu * Df * Power) / c
Where Df is as attached and is where the small end cutoff comes into the equation. With the small end in cutoff, there is no thrust.
Please refer to the 2nd and 3rd attachments as well
-
#1464 by Mulletron on 08 Jan, 2016 09:46
-
COMSOL vs measurements from my copy of Eaglework's frustum.
-
#1465 by TheTraveller on 08 Jan, 2016 11:02
-
COMSOL vs measurements from my copy of Eaglework's frustum.
Interesting work. Doesn't look like a S11 scan. Is this S21 using separate excitation and sample ports? Looks like the coupling fell apart from 1.9 GHz and above? What are your couplers like and where are they positioned? Got a photo? -
#1466 by Mulletron on 08 Jan, 2016 11:12
-
COMSOL vs measurements from my copy of Eaglework's frustum.
Interesting work. Doesn't look like a S11 scan. Is this S21 using separate excitation and sample ports? Looks like the coupling fell apart from 1.9 GHz and above? What are your couplers like and where are they positioned? Got a photo?
That's taken from the sample port. Excitation and sample ports are 31mm E-field probes just for the measurement (location and size not optimized for every frequency combination). Later the sample port was changed to just a solder cup. -
#1467 by TheTraveller on 08 Jan, 2016 12:12
-
COMSOL vs measurements from my copy of Eaglework's frustum.
Interesting work. Doesn't look like a S11 scan. Is this S21 using separate excitation and sample ports? Looks like the coupling fell apart from 1.9 GHz and above? What are your couplers like and where are they positioned? Got a photo?
That's taken from the sample port. Excitation and sample ports are 31mm E-field probes just for the measurement (location and size not optimized for every frequency combination). Later the sample port was changed to just a solder cup.
Hi Mulletron,
Good to have you back posting.
Can you please confirm EW's TE012 resonance at 2.168 GHz as attached? From your scan, it doesn't appear to be there.
I plan to build their frustum & to try to find a resonant freq & mode that will generate thrust. Should then be simply for them to verify.
-
#1468 by Mulletron on 08 Jan, 2016 12:22
-
There's a peak in the ball park of where 2168mhz is. Loook closer. If it's really really really important to you, I can haul it back in to work and focus narrowly on just that mode. I have confidence that Eagleworks knows what they're doing and did in fact hit a resonance at 2168mhz. Rodal's calcs * were almost spot on with Eagleworks too.
* http://forum.nasaspaceflight.com/index.php?topic=39214.msg1470390#msg1470390 -
#1469 by TheTraveller on 08 Jan, 2016 12:51
-
There's a peak in the ball park of where 2168mhz is. Loook closer. If it's really really really important to you, I can haul it back in to work and focus narrowly on just that mode. I have confidence that Eagleworks knows what they're doing and did in fact hit a resonance at 2168mhz. Rodal's calcs * were almost spot on with Eagleworks too.
* http://forum.nasaspaceflight.com/index.php?topic=39214.msg1470390#msg1470390
Yes I would appreciate you doing a S11 scan from 2.0 to 2.5 GHz. It will give me good data to match to my research on that frustum build. Would also need to know your coupling type and location. My 1st frustum builds will end up with lots of holes as I investigate stub & loop coupling options at many locations.
Hopefully can develop a coupler & location that you and EW's can use to generate enough thrust to be measured on a 0.01g resolution scale, using low cost WiFi amps & simple no loss reflected power meter to indicate optimal excitation freq generated by a low cost adjustable Rf generator.
I really want to create a very low cost system that just about anyone can build and measure thrust. Your assistance with your EW clone frustum and test equipment would be most useful and appreciated.
As for the existing work, had Rogers Force equation, with the Df calc had been done, the lack of thrust would not have come as a surprise. At the excited resonant freq, the calculated Df must be above 0 and ideally as close to 1 as you can get it otherwise there will be no thrust generated.
Just because there is a S11 or S21 resonance indication, does not mean that thrust can be produced from that resonance.
When excited in TE012 mode and at 2.168 GHz, the EW frustum's small end is in cutoff and as such there will be no thrust generated. See attached. It is simple to verify that calc.
-
#1470 by Rodal on 08 Jan, 2016 13:06
-
What happens when you're past cutoff?
WarpTech and I went over Zeng and Fan's equations in excruciating detail. These discussions took place in prior threads. WarpTech had a mathematical theory for the EM Drive related to this.
"Electromagnetic fields and transmission properties in tapered hollow metallic waveguides Xiahui Zeng* and Dianyuan Fan" (bottom) http://emdrive.wiki/Evanescent_waves
Has propagation constant curves, from which you can calculate the dispertion from the slope of beta/k.
This tells you on each reverberation of an impulse, what the attenuation and phase shift will be. And group delay, Vg.
Speaking of dispertion
What is dispersion? From Bradshaw:Quotedispersive means:
the refractive index varies with frequency
group velocity differs from phase velocity
group index differs from refractive index
...
k Angular wave number (ωn(ω)/c)
n Real refractive index (Re[√er μr])
ng Group refractive index or just group index (c/vg )
vg Group velocity (dω/dk)
...
So, we must understand group velocity and group index
Writing k = nω/c and V = c/n, vg = dω/dk
ng = c/vg = c * (1/vg) = c * dk/dω
group velocity is the ratio of infinitesimal changes in temporal periodicity to infinitesimal changes in spatial periodicity
I don't like dealing with refractive index (n). I prefer velocity factor (1/n).
How does dispersion work in the frustrum?
Thrust is generated by the large radiation pressure in the frustrum becoming unbalanced at each end with unbalanced dissipation. Thermal radiation pressure is miniscule and diffused.
Frustrum acceleration is required to create the doppler spreading.
Cavity Q and dispersion determines the least acceleration that produces the greatest frequency separation of the doppler-spread microwaves.
Crudely speaking with a poor metaphor, the dispersion is like a nozzle; the doppler-spread sidebands thus separate through the dispersion/frequency gradient.
Q determines how much pressure can build up in the "combustion chamber".
Since power, Q, dispersion, and acceleration determine how much unbalanced radiation pressure there is, it could be very easy to make a dud engine.
If my conjecture is right, it is predictable by a model that accounts for physical acceleration of the frustrum. I can think of a crude way to do that with spice, using a lumped-element model of the waveguide, with mixers at each end to make the doppler shift. That will produce the waveguide signal, and parameters, but the pointing vector/radiation pressure would be crudely inferred by voltage and currents on the L-R-C elements.
Should be simply tested through a two or even one D Meep simulation, but there's the acceleration to deal with. Perhaps making 3 copies, past-present-future of the mesh, and compressing the blue end and expanding the red end and swapping the matrix?
Oops, I forgot to say, notice the steepness of the beta/k curves in Zeng & Fan (above)? You want to be there because you get the biggest phase shift and doppler spreading. The equation for doppler spreading could be from https://en.wikipedia.org/wiki/Fizeau_experiment
see https://upload.wikimedia.org/math/b/6/8/b688f7f7b2c8e7643fcc94870371cb5e.png
and notice the term for dispersion. I could no doubt do better by digging through Bradshaw but its late, sorry.
To make the long story short:
1) Zeng and Fan's paper is for an open waveguide having open end. By contrast, the EM Drive as conceived by Shawyer is a completely closed cavity with closed ends. Cut-off conditions apply to open waveguides, and not to closed cavities.
2) I found an important mathematical mistake in Zeng and Fan's paper that affects their results.
3) I calculated Zeng and Fan's expressions for the actual EM Drive experiments with and without the correction needed to address their mathematical mistake.
4) The calculations based on Zeng and Fan's paper don't support the "anomalous" thrust claimed in EM Drive experiments.
One of several papers that appeared very interesting at first, but that upon working out the numbers, unfortunately could not numerically explain the claimed "anomalous" thrust.
______
PS: Concerning <<Thrust is generated by the large radiation pressure in the frustrum becoming unbalanced at each end>>, the radiation pressure in the EM Drive can be shown to be perfectly balanced (hence no net force) if you take into account the radiation pressure on the lateral conical walls, and include all terms, including the time rate of the Poynting vector, in addition to the gradient of the stress, when relying on Maxwell's equations.
Balance equation, including all terms:
where E is the electric field and B is the magnetic field; ρ is the charge density (charge per unit volume) (*), J is the current density corresponding to the motion of the charge and where the electromagnetic momentum density pem is due to the Poynting vector field S:
The Poynting vector S:
Maxwell stress terms:
and what is known as "the Lorentz body force" (force per unit volume) are the first two terms in the balance equation:
so, the balance equation can also be expressed as:
The above balance equation can be simplified, for finding out the net forces on a closed resonant cavity, by applying the divergence theorem of vector calculus ( https://en.wikipedia.org/wiki/Divergence_theorem ) .
The divergence theorem states that the outward flux of a vector field through a closed surface is equal to the volume integral of the divergence over the region inside the surface. It states that the sum of all sources minus the sum of all sinks gives the net flow out of a region.
Since the EM Drive is a closed surface, supposedly not interacting with external fields, and without anything going in or out of it (according to Shawyer), the divergence theorem can be applied to the above balance equation.
According to Maxwell's equations, since the EM Drive is a closed surface, it should not self-accelerate as a result of any electromagnetic action going on inside it. To self-accelerate you need any of the following:
1) Not being really "closed" and hence interacting with external fields in a way that its claimed performance (greater than a photon rocket) and conservation of energy issues can be satisfactorily addressed.
We know something that escapes the EM Drive: heat. As the EM Drive magnetic field inductively heats the metal surfaces, this heat can escape the EM Drive. Hence the emphasis on thermal effects that can explain the "anomalous" forces being simply due to thermal effects.
2) New physics that can explain its self-acceleration without contradicting any of the huge amount of experimental observations we have accumulated of the Cosmos.
____________________
(*) Because the electric field in transverse magnetic (TM) modes has components perpendicular to the metal walls, this electric field component (perpendicular to the wall) in TM modes does not vanish as it approaches the cavity wall, and therefore it will induce a charge distribution on the inner surface of the metal wall, in TM modes.
Also dielectric breakdown will occur in air at an electric field strength of about Emax = 3 × 10^6 V/m, when the charge buildup exceeds the electrical limit or dielectric strength of air. When air molecules become ionized at E >= 3 × 10^6 V/m, the air changes from an insulator to a conductor. Sparks occur because of the recombination of electrons and ions. In lightning, ionized air becomes a good conductor and provides a path whereby charges can flow from clouds to ground. -
#1471 by rfmwguy on 08 Jan, 2016 15:30
-
-
#1472 by VAXHeadroom on 08 Jan, 2016 16:30
-
Just VMS for me. Came to it in about 1987. Every other operating system I've worked on seems like a step backwards (including VxWorks which I use for satellite control systems). I in fact have a MicroVAX 3100-90 on my desk (currently powered off
Let's see what we can work up!
(edit: should really have said "algorithm" instead of "equations" above
)
I've always wondered about the VAX in your moniker. Dare I ask if you know RT-8, RT-11, TSX, RSX-11, RSTS? I suspect VMS is on the list too?
)
-
#1473 by chavv on 08 Jan, 2016 17:04
-
Maybe related:
http://arxiv.org/abs/1504.00333QuoteHow Current Loops and Solenoids Curve Space-time
Andre Fuzfa
Namur Center for Complex systems (naXys),
University of Namur, Belgium
(Dated: December 15, 2015) -
#1474 by DnA915 on 08 Jan, 2016 17:06
-
I am just curious, how useful would Eagle Works QVP simulator be to this group? I know Nasa frequently releases their software to U.S. citizens although it took quite a long time last time I requested one.
- David -
#1475 by Rodal on 08 Jan, 2016 17:10
-
I am just curious, how useful would Eagle Works QVP simulator be to this group? I know Nasa frequently releases their software to U.S. citizens although it took quite a long time last time I requested one.
It would be really great if that software would be made available to the public, for multiple reasons.
- David
First of all, the actual computer code would precisely reveal what are the specific equations being used by NASA to try to explain the anomalous force claimed in the experiments.
The software results could be compared vs. actual experimental results at NASA and at various other centers.
The software could be run to model "what if" scenarios to address a huge number of ideas being discussed: is there a cut-off condition in the NASA model ? (I don't think so), the influence of different mode shapes (NASA has relied on TM mode 212 instead of the TE012 and TE013 modes used by Shawyer and Yang), different geometries, different end plates (the idea of De Aquino to use a ferromagnetic instead of a diamagnetic end, etc etc) -
#1476 by glennfish on 08 Jan, 2016 17:12
-
Just VMS for me. Came to it in about 1987. Every other operating system I've worked on seems like a step backwards (including VxWorks which I use for satellite control systems). I in fact have a MicroVAX 3100-90 on my desk (currently powered off )
You missed the golden days when we moved from 64kb RAM to more, but you got to play with the first 1 MIPS mini.
So, thoughts on a crowdsourced emulator and development framework? -
#1477 by Vesc on 08 Jan, 2016 17:47
-
Just VMS for me. Came to it in about 1987. Every other operating system I've worked on seems like a step backwards (including VxWorks which I use for satellite control systems). I in fact have a MicroVAX 3100-90 on my desk (currently powered off
)
I hear that a lot from former co-workers who have or who are nearing retirement these days...
The VAX-11/750 and I were well acquainted, you might say...
-
#1478 by X_RaY on 08 Jan, 2016 17:54
-
That you can't solve this set of equations in the under cut off volume doesn't mean there can't be thrust at all.I agree with 1) but what is the evidence that leads to 2)?
Why does Shawyer suggest to keep the small end above cutoff?
It is stated in Roger's thrust equation:
Thrust = (2 * Qu * Df * Power) / c
Where Df is as attached and is where the small end cutoff comes into the equation. With the small end in cutoff, there is no thrust.
Please refer to the 2nd and 3rd attachments as well
(If there was ever thrust like discussed and not other effects like Lorentz force or whatever)
As far as I know Shawyer's Df and thrust equations are not proven till now.
All discussed thrust equations for conical cavities leads to larger thrust the closer the narrow end of the cavity is in relation to it's apex (mathematically singularities at null distance).
Some threads ago we had a huge discussion especially of the effects of evanescent waves.
Question: Was this been tested by Shawyer and his company or is it his intuitive opinion that it would not work with an undersized end of the cone, just below cutoff diameter? -
#1479 by Mulletron on 08 Jan, 2016 17:55
-
Rfmwguy, what is that dandy hand tool he's using to cut those templates out?
Edit: Oh, it's electric tin snips!
)
)