-
#1540 by Notsosureofit on 10 Jan, 2016 01:49
-
Acceleration does give a calculable frequency shift (Need to calculate it...)
Note: Then you have to compensate for velocity relative to your reference frame.
For an acceleration (a) of 1m/s^2 and a cavity length (L) of 1m, df/f is ~ 10^-17, in the frame of the cavity. (using the simple linear dispersion model)
So the change in energy is ~ PQLa/2pifc^2. -
#1541 by Rodal on 10 Jan, 2016 02:25
-
Acceleration does give a calculable frequency shift (Need to calculate it...)
Note: Then you have to compensate for velocity relative to your reference frame.
For an acceleration (a) of 1m/s^2 and a cavity length (L) of 1m, df/f is ~ 10^-17, in the frame of the cavity. (using the simple linear dispersion model)
So the change in energy is ~ PQLa/2pifc^2.
The frequency shift associated with an acceleration of of 1m/s^2 is really tiny indeed (df/f is ~ 10^-17) -
#1542 by Notsosureofit on 10 Jan, 2016 02:32
-
Acceleration does give a calculable frequency shift (Need to calculate it...)
Note: Then you have to compensate for velocity relative to your reference frame.
For an acceleration (a) of 1m/s^2 and a cavity length (L) of 1m, df/f is ~ 10^-17, in the frame of the cavity. (using the simple linear dispersion model)
So the change in energy is ~ PQLa/2pifc^2.
The frequency shift associated with an acceleration of of 1m/s^2 is really tiny indeed (df/f is ~ 10^-17)
This is only first order around static..probably drops faster than linear. Maby get some time tomorrow. -
#1543 by aero on 10 Jan, 2016 02:57
-
For information, if anyone wants to learn about Harminv and how it works, see attached.
And here is the Harminv Programmer's take on negative Q.
https://www.mail-archive.com/meep-discuss%40ab-initio.mit.edu/msg00161.html
one cause of a badly calculated Q: Meep simulations have been too short (0.01 microseconds), the Meep discuss page says:Quoterun the simulation longer
Running Meep for too short a time leads to malformed modes and travelling waves that may not be anywhere close to what steady state looks like, and incorrect Q calculations
QUESTION: have the cases for which Q was negative, been run for twice as long or longer?, if yes, did that bring Q into a more reasonable value ?
As far as I can tell, running the simulation longer is the same as narrowing the bandwidth. There isn't a direct run length control as I use-
(run-sources+ (* gc T_meep)
(after-sources (harminv Ex (vector3 dlx dly dlz) fmeep BW 5))
-for Harminv runs ... Unless he is referring to the "idle
time" (* gc T_meep) between source cut-off and Harminv start. Narrowing the bandwidth gets very expensive very quickly in terms of run time but increasing the "idle time" is no big deal.
Maybe Meeper,ThereIWas3, had some success doing this, at least he did mention something along these lines.
Mathematically narrowing the bandwidth is physically identical to increasing the Q.
I should have mentioned that when using the command "run sources+" meep runs the sources for a longer time than I normally do simply to generate images.
For example, "run sources+" propagates the fields for 164.26 meep time units, that's a little over 0.164 μ s simulated time before idling the specified time interval, waiting to initiate the Harminv calculations. That's with resolution set to 134. The simulated time by this command goes up rapidly with increased resolution, as does the wall clock run time of course.
This is a case where more computer, or longer running times could be helpful. Perhaps Tajmar's cavity model will give an opportunity to test higher resolution meep/Harminv runs compared to published data. -
#1544 by VAXHeadroom on 10 Jan, 2016 03:32
-
...
I have a machine at work dedicated to running big/long engineering simulations - a quad core XEON with 16GB RAM. I just installed VirtualBox on it and have the Ubuntu VM with meep on an SD card to take in and install on it. It can run as long as we like - I don't care if it runs a meep run for a few weeks, and nobody but me uses it
This is a case where more computer, or longer running times could be helpful. Perhaps Tajmar's cavity model will give an opportunity to test higher resolution meep/Harminv runs compared to published data.
. Let me know what you'd like to run on it.
-
#1545 by oliverio on 10 Jan, 2016 06:48
-
...
I have a machine at work dedicated to running big/long engineering simulations - a quad core XEON with 16GB RAM. I just installed VirtualBox on it and have the Ubuntu VM with meep on an SD card to take in and install on it. It can run as long as we like - I don't care if it runs a meep run for a few weeks, and nobody but me uses it
This is a case where more computer, or longer running times could be helpful. Perhaps Tajmar's cavity model will give an opportunity to test higher resolution meep/Harminv runs compared to published data. . Let me know what you'd like to run on it.
I am not sure if this is correct... but I don't think you're ever going to get great cpu performance out of virtualbox. As I understand virtualbox doesn't really even talk to the hardware of your computer, but its cpu is simulated.
For simulation purposes I'm almost positive you want to install a native copy of Linux, not just boot up on a stick or any such thing. -
#1546 by RFPlumber on 10 Jan, 2016 07:40
-
While my frustum as-built ended up below cut-off on the small end, I figured it may still be useful to test it as-is before modifying And so I finally got to try out my test pendulum with a real frustum as opposite to a dummy load
Can you guess the result?
Well, I am happy to confirm the same findings EW has already reported (at least for a cavity below cut-off).
There is no thrust.
Protocol:
There have been total of 8 runs performed 4 runs with the big end facing East, and 4 runs with the big end facing West. For every 2 runs the frequency was first adjusted (within ~200 kHz) to obtain the minimum of reflected power (typically on the order of 0.3-0.7W, with forward power of 25-29W), then 2 runs of 80 seconds each have been initiated. Each run consisted of 20 seconds idle, followed by 20 seconds of RF power on, followed by 20 seconds idle, followed by 10 seconds high voltage pulse, followed by 10 seconds idle. After the second run the reflected power was measured again to make sure the cavity did not drift out of resonance.
Results for all 8 runs have been post-processed in Excel.
All data files are available at https://drive.google.com/open?id=0B3jbXEyEMvU8R3d3SEp3cmN4dnc. Pendulum platform weight was ~4 kg. Given the 3m suspension from the ceiling, the scale factor is ~13 uN / 1 um of mid-point displacement.
Issues:
The XBee wireless link to turn RF on was not stable. During each of the 20 seconds periods when RF has been commanded on there was typically 2..4 random disconnects, each was typically less than 1 second duration. During those disconnects the RF power amplifier has been commanded off, likely resulting in short drops in applied RF power.
Findings:
No directional force has been observed while applying RF power, even after post-processing the results with basic statistics.
Some uni-directional force (on the order of 60-130 uN) is present long after the RF power is turned OFF. It is possible this force starts at some point when RF power is still ON. The force is likely of thermal origin?
Regardless of whether the frustum is oriented East or West the mid-points of pendulum oscillation appear to follow the same pattern during the run (attached). There is some noise oscillations on the order of +- 5 um, both before and during the RF pulse, then there is a long unidirectional shift up to 50 um, which is interrupted by another unidirectional pulse (~25 um) from applied high voltage / electrostatic force.
Enjoy!
-
#1547 by TheTraveller on 10 Jan, 2016 08:05
-
2) reports of experiments showing what happens to the anomalous force when the EM Drive has a small diameter that is below the cut-off condition. On the contrary, all the NASA tests are for an EM Drive that has a small diameter that is below the cut-off condition (as pointed out by TT), and on top of that they use a dielectric insert which lowers the natural frequency even further. Yet, NASA reports an anomalous thrust. Now, somebody could answer "well that's why NASA reports thrust orders of magnitude lower than Shawyer and Yang", but there are problems with that explanation:
The use of a dielectric in the small end lowers the cutoff freq such that, assuming a dielectric constant of 2.2, the small end on the EW copper frustum is operating above cutoff and thus there is no apparent small end cutoff affecting the EW measured thrust. -
#1548 by oliverio on 10 Jan, 2016 08:12
-
While my frustum as-built ended up below cut-off on the small end, I figured it may still be useful to test it as-is before modifying And so I finally got to try out my test pendulum with a real frustum as opposite to a dummy load
Can you guess the result?
Well, I am happy to confirm the same findings EW has already reported (at least for a cavity below cut-off).
There is no thrust.
Protocol:
There have been total of 8 runs performed 4 runs with the big end facing East, and 4 runs with the big end facing West. For every 2 runs the frequency was first adjusted (within ~200 kHz) to obtain the minimum of reflected power (typically on the order of 0.3-0.7W, with forward power of 25-29W), then 2 runs of 80 seconds each have been initiated. Each run consisted of 20 seconds idle, followed by 20 seconds of RF power on, followed by 20 seconds idle, followed by 10 seconds high voltage pulse, followed by 10 seconds idle. After the second run the reflected power was measured again to make sure the cavity did not drift out of resonance.
Results for all 8 runs have been post-processed in Excel.
All data files are available at https://drive.google.com/open?id=0B3jbXEyEMvU8R3d3SEp3cmN4dnc. Pendulum platform weight was ~4 kg. Given the 3m suspension from the ceiling, the scale factor is ~13 uN / 1 um of mid-point displacement.
Issues:
The XBee wireless link to turn RF on was not stable. During each of the 20 seconds periods when RF has been commanded on there was typically 2..4 random disconnects, each was typically less than 1 second duration. During those disconnects the RF power amplifier has been commanded off, likely resulting in short drops in applied RF power.
Findings:
No directional force has been observed while applying RF power, even after post-processing the results with basic statistics.
Some uni-directional force (on the order of 60-130 uN) is present long after the RF power is turned OFF. It is possible this force starts at some point when RF power is still ON. The force is likely of thermal origin?
Regardless of whether the frustum is oriented East or West the mid-points of pendulum oscillation appear to follow the same pattern during the run (attached). There is some noise oscillations on the order of +- 5 um, both before and during the RF pulse, then there is a long unidirectional shift up to 50 um, which is interrupted by another unidirectional pulse (~25 um) from applied high voltage / electrostatic force.
Enjoy!
Too much speculation required for much discussion before more testing, but let me suggest two things:
1) It is possible that (given the EMDrive effect is real, which we do not know is the case) that the decay of the evanescent waves inside the cavity rises through an ephemeral frequency/mode range in which your cavity is nominally thrust-producing. This could be the blip of "uni-directional force" or that could be a thermal effect.
That said, given the above, I find a predictable unidirectional nature a thermally cooling system to be unlikely. Certainly possible (especially with design), but I actually think of that blip on your force charts as pretty odd.
2) Try adding a dielectric to your current cavity. If, as some suspect to be the case (and have advocated recently) the cut-off frequency of the small end (w.r.t. the operating frequency) is very important to thrust generation, then placing a dielectric in the small end should change the generation of force (or not) in a noticeable way because it should raise the maximum dimensions required to reach the purported "cut off" frequency of the small end.
EDIT: I can also see that if there is some steady, net-directional thermal force applied (which appears consistent with the graphs), powering off might cause this "blip" as a pendulum system returns towards its midpoint. This would be consistent with the idea of blowing on a pendulum until your breath is exhaled entirely, over the course of which you would see a force that extends slowly in one direction until the force of gravity brings it towards the midpoint with a higher velocity than at any point of its impetus force. -
#1549 by TheTraveller on 10 Jan, 2016 08:37
-
2) Try adding a dielectric to your current cavity. If, as some suspect to be the case (and have advocated recently) the cut-off frequency of the small end (w.r.t. the operating frequency) is very important to thrust generation, then placing a dielectric in the small end should change the generation of force (or not) in a noticeable way because it should raise the maximum dimensions required to reach the purported "cut off" frequency of the small end.
Adding a dielectric disc at the small end will lower the cutoff freq, BUT it will also alter the resonance freq and reduce Q. -
#1550 by RFPlumber on 10 Jan, 2016 09:04
-
While my frustum as-built ended up below cut-off on the small end, I figured it may still be useful to test it as-is before modifying And so I finally got to try out my test pendulum with a real frustum as opposite to a dummy load
Can you guess the result?
Well, I am happy to confirm the same findings EW has already reported (at least for a cavity below cut-off).
There is no thrust.
...
Too much speculation required for much discussion before more testing, but let me suggest two things:
1) It is possible that (given the EMDrive effect is real, which we do not know is the case) that the decay of the evanescent waves inside the cavity rises through an ephemeral frequency/mode range in which your cavity is nominally thrust-producing. This could be the blip of "uni-directional force" or that could be a thermal effect.
...
Somehow I am thinking of a much simpler explanation... That steady force is a force of air being either expelled out of the cavity (when first heating) or sucked back in (when cooling off) via all the gaps, and then reflecting off the platform surface and off the various components on it and hence creating "force". As long as air is reflecting from the same set of components, yet gaps are slightly different at each of the 2 sides of the frustum, one will likely observe what is being observed here - the force is uni-directional, yet slightly different between the 2 orientations. And it stops after a while. Btw, heating with a magnetron will certainly produce a bigger "force"... just not for very long.
The important point is that there is definitely no force (yet) on the order of ~1+ mN. I wouldn't bother speculating about all those < 100 uN forces, as there is really no limit on how far one can go wondering if any of those are truly anomalous. -
#1551 by frobnicat on 10 Jan, 2016 09:07
-
Quote
As I understand it, the objection here is that constant thrust at the levels claimed for the EM Drive lead directly to a violation of Conservation of Energy. However, suppose the thrust is NOT constant, but instead 'crashes' before reaching that point?
I am thinking here of a 'longer cycle.' Something on the order of 30-40 seconds of high 'thrust,' followed by an unavoidable 'crash' of at least several minutes with no thrust, yet still drawing power, then another 30-40 seconds of 'thrust,' followed by another 'crash.' A 'charge / discharge' cycle. To maintain 'constant acceleration,' you'd need several (10? 20?) EM Drives working in a timed sequence, and a corresponding increase in required power.
Does this represent a valid workaround for Conservation of Energy?QuoteSo, say you start at .1N/kW=100000µN/kW (where EM drive would start to be competitive with ion thruster, given contemporary flight ready sources of power -i.e. not advanced fission or fusion...-) and this is only valid for 1/10th of the time and during 9/10th each thruster still has to consume same nominal power (as per your hypothesis "yet still drawing power", which is hardly motivated : why not just switch off when 'crash' occurs and let things cool down to same state as that allowing a first pulse ?) then that gets us .01N/kW=10000µN/kW overall for the whole system (and added mass to the system, but total mass of EMdriven spacecraft doesn't play any role in the CoE issue). 10000µN/kW is not really interesting compared with ion thruster (given contemporary flight ready sources of power...), and would make power generation from apparent CoE breaking (breakeven) impractical (critical velocity>100km/s) , but still possible in principle, and that's obviously a problem with known physics.
I should have been clearer. The 'crash' (recharge with no apparent thrust) is mandatory. This is where the power for the active 'thrust generating' part of the cycle comes from. A sort of charge/discharge or capacitor effect. Without that long recharge/crash at one point, you don't get the 'thrust,' for a much shorter interval later on, even though you are still putting power into the frustum.
There is no indication that I'm aware of that, in the reported/claimed experimental positives, there is a "warm up" of more than a few seconds before an effect is registered for a first pulse. What we see, for instance at EagleWorks, is a relatively fast response (well, quite not fast enough to be unambiguously attributed to an electromagnetic effect, but this is another matter) at the first power-on of a sequence. When two consecutive pulses are recorded, the power is off between them, and while the second one shows slight variation in shape and drift relative to the first, it is still of same magnitude.
If a "warm up" or preliminary "charge up" is mandatory before the effect fires, this would show on first power on as a very delayed effect, which is apparently not the case. If a dead cold device does show some thrust at first power on, which is apparently the case, then what would prevent an unpowered device to get back to this initial "cold" state after some amount of time (at no energetic cost) ? You seem to require a phenomenon that kind of starts "charged" out of the factory (brand new device, never powered on) to explain an immediate first pulse, but that would need an active recharge before getting a second pulse : your hypothesis seems unnatural, what would make the difference between brand new copper cavity just assembled and a cavity that saw power for 30s and then was made sit on a shelf, unpowered, for 2 weeks ? What could possibly be responsible for such "memory effect" that would need active power to actually be "reset" ?QuoteWhat brings this on is I note that while there are multiple experiments showing 'thrust,' none have lasted for more than a minute or so. In several cases - most recently with Seashell's, the thrust 'crashes' after a short while. Currently, when thought of at all, it is in the context of a 'bug.' I suspect it is a built in, unavoidable feature.
Well, thing is, even if its a built in feature, this doesn't save the EMdrive effect to need an invisible auxiliary power source from the background, at least if it is "frame agnostic" to the point of still thrusting with efficiency φ (in Newtons per electrical Watt) in a velocity window (relative to some background frame) larger than 1/φ (in m/s). Moreover, as I think I've shown in previous post linked to (constant thrust @ constant velocity) this problem appears for whatever short delta t. Intermittent functioning features won't save the device from that.
In short your idea, taken at face value, would save a frame invariant EMdrive effect from CoE (conservation of Energy) apparent problems if to get 30s of thrust at 100µN/kW you'd have to power it (with no thrust) for 900s (15 minutes) before thrusting again for 30s (etc...). Or if to get 30s of thrust at 100mN/kW you'd have to power it (with no thrust) for 900000s (~250 hours) before thrusting again for 30s (etc...).QuoteLikewise, with the MEEP simulations, Rodal and you both noted the effect goes exponential, but that we are seeing only the first 1/1000th (at the very most) of a much longer sequence. My suspicion is this exponential increase would eventually reverse itself.
Meep simulates known physics, known physics tells there is no thrust at all (short of minuscule "one shot" recoil from transfers of mass_energy from antennae to cavity, and radiated waste heat at 3.33µN/kW max. but that's not modeled). So if Meep tells otherwise, until someone includes new physics in the package (mutable quantum vacuum...) then there is a problem with Meep (or rather how it is being used).
The exponential fitted by Rodal (net Poynting vector, not integrating the side walls) and that I somehow saw also on a longer run (by Aero) concerns a very short transient at initial power-on (that is not even realistic as no microwave source will switch on full power on a nanosec.). Edit : my plots were about energy content, not Poynting vector. On the longer run what I see is that this initial transient exponential increase is followed by an inflection point (which is not the same as decreasing values, only decreasing rates) and then we have a well behaved asymptotic convergence to a plateau, with no indication of drop later on (given sustained power input).
If something special happens to a real device after ~30s, then it is most likely thermal in origin, as time constants representative of electromagnetic content is 6 orders of magnitude below, around a few µs at most. Meep doesn't model thermal effects.
-
#1552 by oliverio on 10 Jan, 2016 10:45
-
@frobnicat:
Two questions, forgive me if I am reiterating previous content..
In your analysis of the "over unity" conditions of the theoretically operant emdrive, do you consider the power of the device as its loaded energy content or the input power?
I ask this because for any given optical cavity, if there were a constant em-kinetic conversion taking place, there is not a linearly varying power requirement. I touched on this in a post about the notion of elasticity, and how the power input will always inelastically (at an increasing rate of inelasticity) vary for any rising photon-in-resonance schema.
More simply, if the amount of energy to double photons in resonance is "A", to double the amount of photons again is not "2A". So my question is whether you have been interpreting the analysis in this way also, and if not, how does that change the over-unity power requirement (if thrust is considered variant to cavity resonating power rather than power input to cavity, with the given that one photon must be absorbed by the cavity [or quantum plasma, etc] in order to generate 1 unit of thrust)?
--
Question two: if spacetime itself in the form of some Dirac sea formulation is given momentum by the drive, we cannot consider the drive to be frame-invariant any longer because the existence of a mutable quantum vacuum would essentially require a reformulation of special relativity as the Dirac sea, given that momentum can be imparted as per the theory of dr. White et. al is essentially proposing a shared privileged reference frame. Do you agree? -
#1553 by Rodal on 10 Jan, 2016 12:18
-
2) reports of experiments showing what happens to the anomalous force when the EM Drive has a small diameter that is below the cut-off condition. On the contrary, all the NASA tests are for an EM Drive that has a small diameter that is below the cut-off condition (as pointed out by TT), and on top of that they use a dielectric insert which lowers the natural frequency even further. Yet, NASA reports an anomalous thrust. Now, somebody could answer "well that's why NASA reports thrust orders of magnitude lower than Shawyer and Yang", but there are problems with that explanation:
The use of a dielectric in the small end lowers the cutoff freq such that, assuming a dielectric constant of 2.2, the small end on the EW copper frustum is operating above cutoff and thus there is no apparent small end cutoff affecting the EW measured thrust.
Please show us the mathematical equation and the source of the equation you are using to support your conclusion regarding "cut-off" in a resonant tapered cavity with a dielectric insert at the small end !
____________
(*) We have shown that there is no doubt that NASA's test without a dielectric was in resonance: see https://forum.nasaspaceflight.com/index.php?topic=39214.msg1469685#msg1469685 and subsequent posts.
The equations (and references) you have posted previously regarding cut-off are:
1) applicable only to open waveguides, while Shawyer considers the EM Drive a closed cavity. This is a most important conflict I have never seen addressed: why is a closed cavity being analyzed as is if it would be an open waveguide?
2) applicable only to open waveguides without dielectric inserts asymmetrically placed inside them
3) applicable only to open waveguides with constant cross-section
-
#1554 by VAXHeadroom on 10 Jan, 2016 12:23
-
This is incorrect. aero has shown that meep running on VirtualBox (added: in Ubuntu) only has about a 1% runtime degradation over a native operating system. And VirtualBox DOES NOT emulate a CPU, it merely provides a 'blank slate' for another operating system to run on the same hardware at the same time....
I have a machine at work dedicated to running big/long engineering simulations - a quad core XEON with 16GB RAM. I just installed VirtualBox on it and have the Ubuntu VM with meep on an SD card to take in and install on it. It can run as long as we like - I don't care if it runs a meep run for a few weeks, and nobody but me uses it
This is a case where more computer, or longer running times could be helpful. Perhaps Tajmar's cavity model will give an opportunity to test higher resolution meep/Harminv runs compared to published data. . Let me know what you'd like to run on it.
I am not sure if this is correct... but I don't think you're ever going to get great cpu performance out of virtualbox. As I understand virtualbox doesn't really even talk to the hardware of your computer, but its cpu is simulated.
For simulation purposes I'm almost positive you want to install a native copy of Linux, not just boot up on a stick or any such thing.
-
#1555 by Rodal on 10 Jan, 2016 12:33
-
While my frustum as-built ended up below cut-off on the small end, I figured it may still be useful to test it as-is before modifying And so I finally got to try out my test pendulum with a real frustum as opposite to a dummy load
Can you guess the result?
Well, I am happy to confirm the same findings EW has already reported (at least for a cavity below cut-off).
There is no thrust.
...
Thank you very much for carrying out your test and for reporting your Null results. As previously discussed there appear to be several Null, or "inconclusive", results by other people that have not been reported in the EM Drive wiki (*):
http://emdrive.wiki/Experimental_Results
It would be very much appreciated it if you could enter your experimental results in this EM Drive wiki.
_____________________________________________
(*) if Null and "inconclusive" reports are not reported in the same table where positive results are posted, then the table becomes biased towards claimed positive results. We should strive to have all results posted in the table, whether null, positive or "inconclusive" -
#1556 by glennfish on 10 Jan, 2016 12:40
-
For these runs, can you define what each channel is measuring in the run .csv file?
Results for all 8 runs have been post-processed in Excel.
All data files are available at https://drive.google.com/open?id=0B3jbXEyEMvU8R3d3SEp3cmN4dnc. Pendulum platform weight was ~4 kg. Given the 3m suspension from the ceiling, the scale factor is ~13 uN / 1 um of mid-point displacement.
And
Can you clarify your file naming convention so we have a better idea which file means what? -
#1557 by TheTraveller on 10 Jan, 2016 12:41
-
2) reports of experiments showing what happens to the anomalous force when the EM Drive has a small diameter that is below the cut-off condition. On the contrary, all the NASA tests are for an EM Drive that has a small diameter that is below the cut-off condition (as pointed out by TT), and on top of that they use a dielectric insert which lowers the natural frequency even further. Yet, NASA reports an anomalous thrust. Now, somebody could answer "well that's why NASA reports thrust orders of magnitude lower than Shawyer and Yang", but there are problems with that explanation:
The use of a dielectric in the small end lowers the cutoff freq such that, assuming a dielectric constant of 2.2, the small end on the EW copper frustum is operating above cutoff and thus there is no apparent small end cutoff affecting the EW measured thrust.
I can write the opposite. Rather than argue with words, let's be constructive. Please show us the mathematical equation and the source of the equation you are using to support your conclusion!
!
The equations are standard microwave engineering equations to calculate the cutoff and guide wavelength in a circular waveguide. But I think you already do know that.
Next you will say those equations don't apply as this is not a constant diameter waveguide.
I will reply saying that is your unproven theory and we will go around and round.
Bottom line is the DIY EmDrive builders are here because of Roger Shawyer and his invention. His advice is that if you calc the small end cutoff, as you would a constant diameter waveguide and it is at or below cutoff, the small end needs to be enlarged so it is operating above cutoff or you will not generate any thrust. That advise is backed up if you use Roger's Df equation and calc the Df of the frustum as attached. Maybe you should try it and be sure to use a cutoff equation that includes the Dielectric Constant when calculating a small end cutoff with a dielectric. Hint: the cutoff freq drops when you add in the dielectric.
My spreadsheet has been updated to support dielectrics inside the frustum. Attached see runs with no dielectric (small end in cutoff) and a dielectric (small end not in cutoff).
-
#1558 by Tellmeagain on 10 Jan, 2016 12:44
-
Somehow I am thinking of a much simpler explanation... That steady force is a force of air being either expelled out of the cavity (when first heating) or sucked back in (when cooling off) via all the gaps, and then reflecting off the platform surface and off the various components on it and hence creating "force". As long as air is reflecting from the same set of components, yet gaps are slightly different at each of the 2 sides of the frustum, one will likely observe what is being observed here - the force is uni-directional, yet slightly different between the 2 orientations. And it stops after a while. Btw, heating with a magnetron will certainly produce a bigger "force"... just not for very long.
The important point is that there is definitely no force (yet) on the order of ~1+ mN. I wouldn't bother speculating about all those < 100 uN forces, as there is really no limit on how far one can go wondering if any of those are truly anomalous.
Thank you for carrying out the experiment. I think if you have time you may test a symmetric cylinder as a control. -
#1559 by Rodal on 10 Jan, 2016 13:11
-
...The exponential fitted by Rodal (net Poynting vector, not integrating the side walls) and that I somehow saw also on a longer run (by Aero) concerns a very short transient at initial power-on (that is not even realistic as no microwave source will switch on full power on a nanosec.). Edit : my plots were about energy content, not Poynting vector. ...
Correction: I think that you are referring to a nonlinear fit (with an R^2 exceeding 0.99) (with exponential,linear, and harmonic components) of the net force vs. time on both flat ends, based on the stress tensor components normal to the flat ends, instead of an exponential fit to a Poynting vector.
(Going from memory here
)
The net force on the flat ends was variable with time, over the last two cycles of the Meep run (or a little longer), such that its time variation could be fit to that expression. This behavior was found, to a high degree of fit (R^2>0.99) over several runs. As I recall all those runs were at a very early transient stage (0.01 microseconds or so), nowhere near the point where one would expect steady state conditions (requiring tens of microseconds).
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:
what I fitted was the components of this stress tensor normal to the flat ends, integrated over the surface. I think you are referring to this fit, and not to a fit of the Poynting vector time evolution.
Notice that in the balance equation, the gradient of the stress is balanced by the time rate of the Poynting vector. (Assuming that the charge terms and the current terms in the balance equation are negligible).
Or, using the divergence theorem of calculus, the surface integral of the stress (and hence the net force on the surface enclosing the cavity) is balanced by the volume integral of the time rate of the Poynting vector.
One would have to include the surface integration of the stress on the lateral conical walls, calculate its vector resultant and add it to the net force resultant on the flat walls, and this overall net force should be balanced by the volume integral of the time rate of the Poynting vector.
For example, if the Poynting vector were to be exponentially increasing during the early transient regime, its time rate would also be exponentially increasing (the derivative of an exponential is an exponential), and the volume integral of its time rate would be balanced by an exponentially increasing stress, integrated over the surface.
. Let me know what you'd like to run on it.
)