-
#1080 by SteveD on 01 Jan, 2016 07:23
-
Hi all. Earlier today I challenged somebody to prove their assertions about the South Korean video with readings and time stamps. I found that I couldn't sleep without doing this myself.
First, the second -- inverted -- run is useless, the tester flips the rig, changing the configuration of the balance beam, causing the weight reading to change by over 50 grams (he also does not allow enough time for the frustum to cool, meaning that if thermal expansion put it out of resonance, it was still out of resonance. At 9:18, during the second (inverted) run, the readings on the scale begin to oscillate wildly dropping from 315.79g to 167.89g and finally to 0.80g followed by a series of wide swings between around 300g and what looks like about 180g before stabilizing at about 308g at 10.05. By 10.45 there is a stable reading of 309.76 that begins to trend up. The test ends with the scale showing about 313g.
I wish there had been an analog backup (laser pointer) to confirm that the oscillations are RFI. The changes seem a bit much for the downdraft from a HVAC unit, but might be consistent with the tester accidently hitting the unit (or with a piece of fishing line being used to rig the results snapping).
Looking at the initial run I get:1:58 246.24 2:38 246.15 4:00 246.19 4:23 246.37 Loss or Resonance? 4:52 247.67 5:03 247.42 5:10 247.33 5:20 247.25 5:39 247.20 5:52 247.15
Without any of the wild swings seen in the second run.
Without a good base measurement to establish what the reading was before power on the data is basically junk. I can't tell if the base is 246.24 or 246.15. Without being able to translate, I cannot say for sure that the test begins at 1:58 and not 4:23. A, slight, actual downward force might be registering. I should point out that the downward force, if it exists, would be enough to register on a 0.1g resolution analog scale -- offering a possible alternative to the ad hoc balance beam setup.
The sudden movement upwards at 4:23 is interesting as it seems that some restraining force has been removed. This, and the subsequent upward movement, might be consistent with a loss of resonance causing the EMDrive effect to cease, allowing thermals to fully take over. The subsequent decrease in weight readings, could be consistent with air in the frustum cooling as the energy, without resonance, manifests as heat elsewhere on the experimental rig. Again without known the on time, this could be wrong.
I would very much like to know the dimensions of the frustum, frequency, rf source used and power output. Personally, I'm concerned that the best explanation of the wild oscillations is broken fishing line -- this video may very well be a hoax.
Could this be the rumored test of a cylindrical resonance cavity that was suppose to produce some strange results? -
#1081 by TheTraveller on 01 Jan, 2016 07:50
-
Hi all. Earlier today I challenged somebody to prove their assertions about the South Korean video with readings and time stamps. I found that I couldn't sleep without doing this myself.
Did the same thing with Iulian's test data. Here it is very clear to see the power on and off events as well as the declining Force as the maggie and the frustum warms up. Maggies decrease freq as they warm, so if the initial maggie freq was on the lower freq side of the frustum's bandwidth, as the maggie warms the output freq moves further and further away from frustum optimum freq and generated Force would decline. Should add that a warming frustum does decrease the resonance freq. How well the decreasing maggie freq will track the decreasing frustum freq as they both warm is yet to be determined.
Note the sharp increase / decrease in the measured Force as the maggie power is applied and removed. Hard to accept buoyancy would act that quick.
-
#1082 by qraal on 01 Jan, 2016 09:25
-
You're right, Traveller, buoyancy wouldn't happen that quick. More importantly, it wouldn't decline that quick.
-
#1083 by TheTraveller on 01 Jan, 2016 09:52
-
Thank you very much for your advice.I began to study EMdrive from 2013，with some help from Miss Yang. She have to develop the high temperature of YBCO superconducting cavity used in experiments, but unfortunately they can't get further recognition of the academic committee, lack of project funding. She has already retired, the project has ground to a halt in 2014. She was confirmed the results of the experiment, but cannot explain the conflict with existing physical theory.
Have emailed Prof Yang a link to the above post. Asked her to please comment on the claims. If she replies, will post it here.
All I know for sure is Oyzw's frustum design resonates, according to my spreadsheet, in TE013 mode at 2.528 GHz. Very close to the claimed 2.54 - 2.55 GHz.
That would suggest there was maybe knowledgeable frustum design involved. -
#1084 by frobnicat on 01 Jan, 2016 11:05
-
@Aero et al. Meepers analysts
I downloaded the long run (SeeShell's Yang 6 degree conic frustum model ...) provided here
https://drive.google.com/drive/folders/0B1XizxEfB23tRm41bVFtM1pVYlU
I believe this was done before the corrections to the copper model ?
Anyhow, concerning the energy content and its temporal evolution, for each of the time grabs @ 32 64 ... 2048, I summed on the first 10 time steps (approx. a whole period) the squares of all values for the E field and the H field. No fancy sinus fitting, just raw summing of values squared, that should give something proportional to energy content. Obviously it is quite not rigorous as I'm not integrating on regularly spaced nodes of the volumes, the absolute values are meaningless, but I find the shapes interesting as of latest discussions.
Raw values :
Big End injection
SUM (over ~ a period) of all E² values in function of time since power on (32 ... 2048)
32 3.383636e-002
64 2.128400e+001
128 2.269453e+005
256 1.488172e+008
512 1.791671e+008
1024 1.732074e+008
2048 1.766717e+008
SUM (over ~ a period) of all H² values in function of time since power on (32 ... 2048)
32 4.157975e-001
64 2.002878e+002
128 1.144931e+006
256 2.856278e+008
512 3.293681e+008
1024 3.404555e+008
2048 3.336691e+008
Small End injection
SUM (over ~ a period) of all E² values in function of time since power on (32 ... 2048)
32 3.125358e-001
64 2.052368e+002
128 2.313173e+006
256 1.525427e+009
512 1.837192e+009
1024 1.776081e+009
2048 1.811604e+009
SUM (over ~ a period) of all H² values in function of time since power on (32 ... 2048)
32 9.832593e-001
64 5.706325e+002
128 5.257881e+006
256 2.901258e+009
512 3.377364e+009
1024 3.491055e+009
2048 3.421467e+009
I don't know if it is a mere coincidence or if there is a reason relating to meep output units but sums of H² values are about twice sums of E² values. Just for convenience, in the attached plots I use 2 times E² (Blue) compared to H² (Red). Then an average between 2E² and H² (single Blue plots).
I'll stop here as I have no time to go further and this starts to look like sloppy numerology : I just wanted to point that from those very coarse analysis it looks like the energy content has already reached a plateau (and is stable) after step 512, which would come as a surprise as we are expecting much longer time constant, no ?
It would certainly be useful if (a much more rigorous) integration of total energy content could be done at each step of the time marching simulation, so as to have a smooth E(t) plot to summarize how fast the electromagnetic energy builds up, and then compare to what would be expected from Q value, assuming a constant rate of effective power P delivery to the cavity :
E(t)=τ2×P×(1-e-t/τ2) with τ2=Q/ω
http://forum.nasaspaceflight.com/index.php?topic=39004.msg1466547#msg1466547
-
#1085 by TheTraveller on 01 Jan, 2016 11:29
-
It would certainly be useful if (a much more rigorous) integration of total energy content could be done at each step of the time marching simulation, so as to have a smooth E(t) plot to summarize how fast the electromagnetic energy builds up, and then compare to what would be expected from Q value, assuming a constant rate of effective power P delivery to the cavity :
E(t)=τ2×P×(1-e-t/τ2) with τ2=Q/ω
http://forum.nasaspaceflight.com/index.php?topic=39004.msg1466547#msg1466547
The rate of power delivery to the frustum is not constant. Far from it.
Initially a resonant cavity being filled looks like a short and almost all the power is rejected with Forward power being almost zero and Reflected power being almost all that was generated.
Ok some of the discussion is accelerator cavity related but it still works.
I feel if MEEP is to reflect cavity reality, this 5 x TC fill time, due to Reflected power starting out at almost 100% and dropping to almost 0%, needs to be shown to occur.
-
#1086 by TheTraveller on 01 Jan, 2016 12:03
-
Very latest version of the EmDrive Design spreadsheet is here:
https://drive.google.com/file/d/0B7kgKijo-p0iY1FqemkzWXo1ZzQ/view?usp=sharing
There are additional goodies in this version such as Q unloaded via 2 industry methods, cavity TC and EM wave transmission time per resonant mode and number of end to end transits per 1 TC and per 5 TCs. Lots of end to end transits to make up 1 TC, yet to do 5 x TC.
Q model allows you to alter the resistivity and then calcs skin depth at the operational resonant freq and unloaded Q. Don't criticise the 2 Q models too much. I believe they are close approximations that allow one to vary frustum dimensions and see how that alters Q unloaded. Q for a spherical end cap model still needs work as it is only a very rough approximation.
Next major version will allow you to model a 3 section frustum which has 7 internal dimensions being:
1) small outer dia,
2) small inner dia,
3) small outer to inner length,
4) inner small to inner big length,
5) big inner dia,
6) big outer dia,
7) big inner to big outer length.
They we can see what adding sections to each end of the frustum does to resonance.
Frustum is modelled as approx 65,488 very small constant diameter cylinder sections stacked end to end. Guide wavelength is then calculated for each section, then numerically integrated to find the effective guide wavelength.
For a TE013 resonant calc that breaks each of the 3 x 1/2 waves down into 21,829.333 slices for each 1/2 wave. If the TE013 frustum is 240mm long, each end stacked constant diameter cylinder section is 0.003667 mm long.
Have fun.
BTW the furniture in storage in my workshop is now confirmed to be going late next week. Hooray, I will soon have my workshop back.
Then time to get to serious work, generating kick denier butt experimental data. 2016 will be such a good year.
Phil
-
#1087 by frobnicat on 01 Jan, 2016 12:54
-
It would certainly be useful if (a much more rigorous) integration of total energy content could be done at each step of the time marching simulation, so as to have a smooth E(t) plot to summarize how fast the electromagnetic energy builds up, and then compare to what would be expected from Q value, assuming a constant rate of effective power P delivery to the cavity :
E(t)=τ2×P×(1-e-t/τ2) with τ2=Q/ω
http://forum.nasaspaceflight.com/index.php?topic=39004.msg1466547#msg1466547
The rate of power delivery to the frustum is not constant. Far from it.
Initially a resonant cavity being filled looks like a short and almost all the power is rejected with Forward power being almost zero and Reflected power being almost all that was generated.
Ok, get that, so instant constant power delivery from t=0 would need (almost) infinite amplitude initially...
But I can't wrap my head around the distinction between "external Q" and "ohmic Q" and the corresponding β ratio, I'll leave that to the experts...QuoteOk some of the discussion is accelerator cavity related but it still works.
I feel if MEEP is to reflect cavity reality, this 5 x TC fill time, due to Reflected power starting out at almost 100% and dropping to almost 0%, needs to be shown to occur.
Well I don't know what TC we are talking about if there is 2 different Qs involved. Seems like the flattish energy buildup in the initial ~100 time slices (in the long run by Aero) could relate to this initial "refractory" period with almost 100% power reflected, then when we'd have an effective constant power delivery against losses (ohmic + reflected) proportional to energy, that'd give the 1st order (1-e-t/τ2) ...
I somehow get that the cavity is driven by a given amplitude, not by a given forward power, but at some point (as shown on your linked plots) P_forward reaches a plateau. Isn't the corresponding τ1 possibly much below the τ2 of the ensuing charge at (almost) constant P against losses proportional to E ? Sorry for the speculations, not my domain really.
-
#1088 by oyzw on 01 Jan, 2016 13:37
-
Very latest version of the EmDrive Design spreadsheet is here:
Hi I made a few aluminum cavity with spinning processing method, used for resonance frequency measurement. Copper cavity has been sent to do the test, the test data as follows
https://drive.google.com/file/d/0B7kgKijo-p0iY1FqemkzWXo1ZzQ/view?usp=sharing
There are additional goodies in this version such as Q unloaded via 2 industry methods, cavity TC and EM wave transmission time per resonant mode and number of end to end transits per 1 TC and per 5 TCs. Lots of end to end transits to make up 1 TC, yet to do 5 x TC.
Q model allows you to alter the resistivity and then calcs skin depth at the operational resonant freq and unloaded Q. Don't criticise the 2 Q models too much. I believe they are close approximations that allow one to vary frustum dimensions and see how that alters Q unloaded. Q for a spherical end cap model still needs work as it is only a very rough approximation.
Next major version will allow you to model a 3 section frustum which has 7 internal dimensions being:
1) small outer dia,
2) small inner dia,
3) small outer to inner length,
4) inner small to inner big length,
5) big inner dia,
6) big outer dia,
7) big inner to big outer length.
They we can see what adding sections to each end of the frustum does to resonance.
Frustum is modelled as approx 65,488 very small constant diameter cylinder sections stacked end to end. Guide wavelength is then calculated for each section, then numerically integrated to find the effective guide wavelength.
For a TE013 resonant calc that breaks each of the 3 x 1/2 waves down into 21,829.333 slices for each 1/2 wave. If the TE013 frustum is 240mm long, each end stacked constant diameter cylinder section is 0.003667 mm long.
Have fun.
BTW the furniture in storage in my workshop is now confirmed to be going late next week. Hooray, I will soon have my workshop back.
Then time to get to serious work, generating kick denier butt experimental data. 2016 will be such a good year.
Phil
-
#1089 by TheTraveller on 01 Jan, 2016 14:08
-
Hi I made a few aluminum cavity with spinning processing method, used for resonance frequency measurement. Copper cavity has been sent to do the test, the test data as follows
Hi Oyzw,
Thanks for the images and test data. Most appreciated.
I suspect your full loop antenna design and placement position in the side wall is not ideal. Most cavity excitation methods using a loop, use a 1/2 loop that has the return loop current flowing through the cavity wall. So the loop looks like a "U" and not a "O".
The S11 rtn loss peak of 8.4466 is a VSWR of 2.216 which means with 400W output from your Rf amp, only 350W will make in inside the frustum, with 50W being rejected as reflected wasted power. If your Rf amp can't handle that much reflected Rf, you will need to use a circulator and Rf dummy load to protect your Rf amp from overheating or damage.
Section 2.2.3 of the attached document may be of interest in designing a 1/2 round coupling loop.
-
#1090 by oyzw on 01 Jan, 2016 14:13
-
Your point of view is very correct, I need to continue to improve the antenna. I also need to purchase lock resonance frequency signal source, do you have any good Suggestions?Hi I made a few aluminum cavity with spinning processing method, used for resonance frequency measurement. Copper cavity has been sent to do the test, the test data as follows
Hi Oyzw,
Thanks for the images and test data. Most appreciated.
I suspect your full loop antenna design and placement position in the side wall is not ideal. Most cavity excitation methods using a loop, use a 1/2 loop that has the return loop current flowing through the cavity wall. So the loop looks like a "U" and not a "O".
The S11 rtn loss peak of 8.4466 is a VSWR of 2.216 which means with 400W output from your Rf amp, only 350W will make in inside the frustum, with 50W being rejected as reflected wasted power. If your Rf amp can't handle that much reflected Rf, you will need to use a circulator and Rf dummy load to protect your Rf amp from overheating or damage.
-
#1091 by TheTraveller on 01 Jan, 2016 14:30
-
Your point of view is very correct, I need to continue to improve the antenna. I also need to purchase lock resonance frequency signal source, do you have any good Suggestions?Hi I made a few aluminum cavity with spinning processing method, used for resonance frequency measurement. Copper cavity has been sent to do the test, the test data as follows
Hi Oyzw,
Thanks for the images and test data. Most appreciated.
I suspect your full loop antenna design and placement position in the side wall is not ideal. Most cavity excitation methods using a loop, use a 1/2 loop that has the return loop current flowing through the cavity wall. So the loop looks like a "U" and not a "O".
The S11 rtn loss peak of 8.4466 is a VSWR of 2.216 which means with 400W output from your Rf amp, only 350W will make in inside the frustum, with 50W being rejected as reflected wasted power. If your Rf amp can't handle that much reflected Rf, you will need to use a circulator and Rf dummy load to protect your Rf amp from overheating or damage.
This is a very custom design job. A very rough sketch of my design is attached.
This design needs no circulator and Rf dummy load as the real time Rf environment is monitored at least 100 times a second and will never allow the Rf amp to be over stressed. Rf power is programmed from approx 200mW to 100W with real time Forward and Reflected power feedback. If reflected power exceeds programmed limits, power output levels are dropped back in real time.
-
#1092 by oyzw on 01 Jan, 2016 14:37
-
I find this
Your point of view is very correct, I need to continue to improve the antenna. I also need to purchase lock resonance frequency signal source, do you have any good Suggestions?Hi I made a few aluminum cavity with spinning processing method, used for resonance frequency measurement. Copper cavity has been sent to do the test, the test data as follows
Hi Oyzw,
Thanks for the images and test data. Most appreciated.
I suspect your full loop antenna design and placement position in the side wall is not ideal. Most cavity excitation methods using a loop, use a 1/2 loop that has the return loop current flowing through the cavity wall. So the loop looks like a "U" and not a "O".
The S11 rtn loss peak of 8.4466 is a VSWR of 2.216 which means with 400W output from your Rf amp, only 350W will make in inside the frustum, with 50W being rejected as reflected wasted power. If your Rf amp can't handle that much reflected Rf, you will need to use a circulator and Rf dummy load to protect your Rf amp from overheating or damage.
This is a very custom design job. A very rough sketch of my design is attached.
This design needs no circulator and Rf dummy load as the real time Rf environment is monitored at least 100 times a second and will never allow the Rf amp to be over stressed. Rf power is programmed from approx 200mW to 100W with real time Forward and Reflected power feedback. If reflected power exceeds programmed limits, power output levels are dropped back in real time.
-
#1093 by OnlyMe on 01 Jan, 2016 14:58
-
When trying to compare meep results, to what are believed to be (at this point) real conditions, I keep running into a disconnect... Not being an expert...
It seems to me that meep models an ideal microwave source and a frustum that does not deform as it warms up. A magnetron is not, an ideal source and it appears that a frustum's resonance is not fixed. It drifts as it warms up and perhaps with, as yet unidentified changing conditions....
It takes time for a magnetron to warm up and hit a frequency lock (is that right)? And meep is locked on frequency..? Meep models an ideal and yet static relationship between the frustum and source, an instant? While in practice the relationship between a frustum and magnetron is dynamic? They both change...
This suggests to an inexpert observer that the best that can be expected from meep or any other current modeling system, is a snap shot in time and ideal conditions.
You might be able to determine what resonance looks like for that instant in time, but not what is happening over time, in a dynamic system that is changing over time.
Even given a tunable microwave source, it must be dynamically tuned as the frustum resonance changes.
How can any modeling software that does not model changes in the frustum, we cannot yet quantify, be expected to represent anything real? Since meep begins with an ideal source and frustum, you should not have to run the model for any longer than it takes to establish a stable resonant pattern, to get all you can get from meep.
I suggest setting up the simplest run possible, maybe even just a cylinder and running that for a longer time frame to see just how long it takes meep to reach a stable resonance. That should tell you how long you need to run the simulation for a frustum, before the output becomes redundant.
I also suspect, that unless I have missed some dynamics within meep, that critical time will be close to the current time frames modeled by meep.., all things going in being idealistic and fixed. -
#1094 by oyzw on 01 Jan, 2016 15:03
-
Who can calculate internal surface lorentz force of the high Q cavity using Superfish？
-
#1095 by TheTraveller on 01 Jan, 2016 15:04
-
I find this
Your paper in English as pdf is attached.
I do note there is no protection against an experimental event that could spike reflected power and release the "Magic Smoke" from your Rf amp.
This system need a 2nd hole in the frustum for the sense antenna, which means increased loss and lower Q. Eagleworks use a PLL system similar to this.
-
#1096 by glennfish on 01 Jan, 2016 15:06
-
You're right, Traveller, buoyancy wouldn't happen that quick. More importantly, it wouldn't decline that quick.
I would expect buoyancy to propagate at just under 340 m/s. What propagation delay would you expect? -
#1097 by SeeShells on 01 Jan, 2016 15:19
-
http://www.popularmechanics.com/science/a18801/3d-printed-wonder-ceramics-wont-shatter/
I want a FRUSTUM from this!
Shell -
#1098 by OnlyMe on 01 Jan, 2016 15:25
-
You're right, Traveller, buoyancy wouldn't happen that quick. More importantly, it wouldn't decline that quick.
I would expect buoyancy to propagate at just under 340 m/s. What propagation delay would you expect?
If you are talking balloon like buoyancy, I have no idea how fast any air inside the frustum will heat.., other than by contact with the thermal heating of the frustum itself.
As far as the frustum heating and any internal ballooning or external convection, don't you have to account for how long it take the frustum's heat capacity to be filled before it begins to transfer heat?
Unless you expect that microwave heating of the frustum is uniform, not consistent with Shell's unrecorded observations, any heating of the interior surface of the metal frustum will not be transferred out with 100% efficiency until the frustum is uniformly heated to its full heat capacity.
Shell observed that her side walls felt cool while the endplates felt warm.
Until the whole frustum reaches a stable and uniform temperature, you cannot assume heat will be conducted or dissipated, inside or outside the frustum, in an ideal manner. Some of the initial dissipation will be into the metal frustum itself.
-
#1099 by TheTraveller on 01 Jan, 2016 15:27
-
You're right, Traveller, buoyancy wouldn't happen that quick. More importantly, it wouldn't decline that quick.
I would expect buoyancy to propagate at just under 340 m/s. What propagation delay would you expect?
Glenn,
You need to understand that when Iulian pointed the small end down, he measured weight increase on his scale, so the downward force that was generated was greater than the upward buoyancy force.
When you say propagate at 340 m/sec, hey man that is a fast energy wave propagation in air from a internally heated frustum from maggie generated Rf. Bit hard to understand the logic behind that statement. Also you need to quality your statement with the amount of the upward Force that would be generated and in what time frame and response time.
Maybe generate a Glenn upward force overlay as against the measured data that can also explain the very rapid drop in upward force being generated when the maggie was switched off?
Maybe I'm missing something here but how can the upward Force that was being thermally generated (as you suggest) drop so rapidly when the maggie was switched off? This is not a small variation in Force generation.