http://forum.nasaspaceflight.com/index.php?topic=38203.msg1416759#msg1416759
Got me thinking the quartz rod through the center could support some mode filters if needed.
Nice read Doc.
Shell
Surley you don't want a Q of 50 like Tajmar used? How do you intend to deal with using a waveguide feed, limited input bandwidth and getting all your magnetron energy inside your frustum?
I needed to get some sleep as I was getting tired but I'm awake (for a while) and wanted to answer this question on Tajmar's build vs Shell's "Crazy Eddie" Frustum dual symmetrically opposed waveguides.
Tajmar was trying to excite the TE01 mode
Wiki
"A mode with one half-wave of electric field across the height of the guide and uniform electric field (zero half-waves) across the width of the guide."
There is no place I can see where he had a uniform half-wave across the cavity (guide). I see heat and destructive/constructive bouncing waveforms out of phase and not a uniform electric field anywhere. There is no symmetry in the cavity to create a stable TE01 mode.
This is the issue with injecting microwaves into a frustum the only way I could see to create a stable mode generation was opposing symmetrical waveguides which one needs to be 180 out of phase for the generation of the TE12 patterns. This is something that hasn't been done before in a frustum and will require good tuning to be right but I think the rewards outweigh the negatives.
The biggest problem with a frustum is gettin frust that other people think is real.
I can get all the frust you want, it's thust that is costly.
http://forum.nasaspaceflight.com/index.php?topic=38203.msg1416759#msg1416759
Got me thinking the quartz rod through the center could support some mode filters if needed.
Nice read Doc.
Shell
...you are thinking about feeding the magnetron Rf into your frustum via a waveguide and not direct inject it. My concern with the waveguide is the frustum will then have an input bandwidth that may not wide enough to accept most of the magnetron output bandwidth. Surley you don't want a Q of 50 like Tajmar used? How do you intend to deal with using a waveguide feed, limited input bandwidth and getting all your magnetron energy inside your frustum?
True. But comparing different characteristics of transmission lines one has to trade off:
Characteristic Coaxial Waveguide Winner
Unloaded Q Medium High Waveguide
Power Capability Medium High Waveguide
Bandwidth Large Small Coaxial
Tajmar had measured loaded Q by using a Z=50 Ohm port of a network analyzer (S11). They simply use the 3dB bandwidth. They did not derive the unloaded Q!
Tajmar had measured loaded Q by using a Z=50 Ohm port of a network analyzer (S11). They simply use the 3dB bandwidth. They did not derive the unloaded Q!
What is the "load" on an 0 port EMDrive resonant cavity?
As far as I understand the topic, there is no load on a EMDrive cavity. It is a 0 port resonant cavity with a Rf feed point but no input nor output ports. Measuring the cavity Q via the Rf feed 3dB down points from the max return loss dB is the cavities unloaded Q.
This method to measure unloaded 0 port EMDrive resonant cavity Q is that used by Eagleworks, Shawyer, Prof Yang and Prof Tajmar.
Tajmar had measured loaded Q by using a Z=50 Ohm port of a network analyzer (S11). They simply use the 3dB bandwidth. They did not derive the unloaded Q!
What is the "load" on an 0 port EMDrive resonant cavity?
As far as I understand the topic, there is no load on a EMDrive cavity. It is a 0 port resonant cavity with a Rf feed point but no input nor output ports. Measuring the cavity Q via the Rf feed 3dB down points from the max return loss dB is the cavities unloaded Q.
This method to measure unloaded 0 port EMDrive resonant cavity Q is that used by Eagleworks, Shawyer, Prof Yang and Prof Tajmar.
The plot in the paper shows S11 Measurement. (headline of the plot)
I think they had measured the S11 and after that, the magnetron was connected.
Tajmar was trying to excite the TE01 mode
Where in his paper does it say Tajmar tried to excite TE01 mode or any mode? Have just rechecked the latest paper and there is no mention of any excitation mode.
Tajmar had measured loaded Q by using a Z=50 Ohm port of a network analyzer (S11). They simply use the 3dB bandwidth. They did not derive the unloaded Q!
What is the "load" on an 0 port EMDrive resonant cavity?
As far as I understand the topic, there is no load on a EMDrive cavity. It is a 0 port resonant cavity with a Rf feed point but no input nor output ports. Measuring the cavity Q via the Rf feed 3dB down points from the max return loss dB is the cavities unloaded Q.
This method to measure unloaded 0 port EMDrive resonant cavity Q is that used by Eagleworks, Shawyer, Prof Yang and Prof Tajmar.
The plot in the paper shows S11 Measurement. (headline of the plot)
I think they had measured the S11 and after that, the magnetron was connected.
That Tajmar cavity has no holes in it, other than the waveguide feed port in the frustum side wall and at the other end, the hole to allow the magnetron antenna to get inside the connecting waveguide.
Which would suggest they removed the magnetron and built a sample probe to fit in the magnetron hole and measured the S11 return loss dB curve, which Tajmar shows in the paper and used the 3bd down points to calc the bandwidth and unloaded cavity Q.
If you look closely at the vac photographs, you can see the magnetron is sometimes on one side or the other of the frustum ends. Would guess they rotated the magnetron and it's coupling waveguide assembly to make placing it on the vac table more stable.
Tajmar had measured loaded Q by using a Z=50 Ohm port of a network analyzer (S11). They simply use the 3dB bandwidth. They did not derive the unloaded Q!
What is the "load" on an 0 port EMDrive resonant cavity?
As far as I understand the topic, there is no load on a EMDrive cavity. It is a 0 port resonant cavity with a Rf feed point but no input nor output ports. Measuring the cavity Q via the Rf feed 3dB down points from the max return loss dB is the cavities unloaded Q.
This method to measure unloaded 0 port EMDrive resonant cavity Q is that used by Eagleworks, Shawyer, Prof Yang and Prof Tajmar.
The plot in the paper shows S11 Measurement. (headline of the plot)
I think they had measured the S11 and after that, the magnetron was connected.
That Tajmar cavity has no holes in it, other than the waveguide feed port in the frustum side wall and at the other end, the hole to allow the magnetron antenna to get inside the connecting waveguide.
Which would suggest they removed the magnetron and built a sample probe to fit in the magnetron hole and measured the S11 return loss dB curve, which Tajmar shows in the paper and used the 3bd down points to calc the bandwidth and unloaded cavity Q.
If you look closely at the vac photographs, you can see the magnetron is sometimes on one side or the other of the frustum ends. Would guess they rotated the magnetron and it's coupling waveguide assembly to make placing it on the vac table more stable.
IMHO the simplest way is to use a coaxial to waveguide connector instead the magnetron for the measurements.
And no, they measured Q not Q_0, its not possible to measure Q_0 directly, one have to derive/calculate that from the complex measurement data.
And no, they measured Q not Q_0, its not possible to measure Q_0 directly, one have to derive/calculate that from the complex measurement data.
Thanks for the paper.
After going through it side by side to a google translate version it seems you are saying the unloaded cavity Q is:
Q
0 = (2 * Pi * Stored Energy) / Energy loss per cycle
and that the Q
0 value can't be accurately measured via S11 max return loss dB at the 3dB down points?
Correct?
And no, they measured Q not Q_0, its not possible to measure Q_0 directly, one have to derive/calculate that from the complex measurement data.
Thanks for the paper.
After going through it side by side to a google translate version it seems you are saying the unloaded cavity Q is:
Q0 = (2 * Pi * Stored Energy) / Energy loss per cycle
and that the Q0 value can't be accurately measured via S11 max return loss dB at the 3dB down points?
Correct?
Only for the coupling factor equals to 1 the full 1/sqrt(2) BW is direct usable. The coupling factor can be derived from the measurements. Sorry the paper is german but i never find a better instructions manual.
Please look at page 21 in the green box..
And Q_0=Q*(1+coupling factor)
Back from the Dresden front! Martin Tajmar sent me an email today where he says he measured the internal height of his frustum. It seems he went to the lab to measure it himself before his student came back from holidays 
Quote from: Martin Tajmar
I measured it: the internal height is 72.8 mm (after adjustment for better resonance). Between the Cavity and the waveguide we used an adapter. The measures are all correct. We simulated it in COMSOL and also Shawyer with his calculation program assured us that the dimensions we used were correct.
So the official internal dimensions from Tajmar are now:
Db = 108.2 mm
Ds = 77 mm
Height = 72.8 mm
Can you guys verify this cavity resonates in your frustumator software?
EDIT: A WR340 waveguide measures 86.36 x 43.18 mm so Tajmar used a coupling adapter reducing the waveguide.
http://forum.nasaspaceflight.com/index.php?topic=37642.msg1412868#msg1412868According to TheTraveller's spreadsheet, based on those dimensions
Db = 108.2 mm
Ds = 77 mm
Height = 72.8 mm
with spherical ends, what mode shape did Tajmar excite?According to my calculations it is (the equivalent of cylindrical mode)
TM010 at ~2.45 GHz (*)
the lowest transverse magnetic mode possible.
I have modified the EM Drive wiki (that had TE111 based on a longer length of 0.1008 m instead of 0.0728m, and Db=0.1062m instead of 0.1082m and Ds=0.075m instead of 0.077m) to show these dimensions and mode shape.
_____________
(*) I write "the equivalent" because this mode is not constant in the longitudinal direction. It is impossible for a mode to be constant in the longitudinal direction for a truncated cone. I write "the equivalent" because this is the lowest TM mode for the truncated cone. It is a degenerate mode that corresponds to TM010 for a cylinder.
(**) This is based on the stated assumption by FluxCapacitor that Tajmar used a coupling adapter reducing the waveguide WR340 that measures 86.36 x 43.18 mm.
with spherical ends, what mode shape did Tajmar excite?[/b]
Where did Prof Tajmar say the end plates had a spherical curve?
Did I miss some info here on the forum as I can't find any such mention in the original or updated paper?
with spherical ends, what mode shape did Tajmar excite?[/b]
Where did Prof Tajmar say the end plates had a spherical curve?
Did I miss some info here on the forum as I can't find any such mention in the original or updated paper?
My recollection is that the information (spherical ends and dimensions) is the product of personal e-mails exchanged between FluxCapacitor and Tajmar, if my memory is correct.
FluxCapacitor to confirm...
Back from the Dresden front! Martin Tajmar sent me an email today where he says he measured the internal height of his frustum. It seems he went to the lab to measure it himself before his student came back from holidays
Quote from: Martin Tajmar
I measured it: the internal height is 72.8 mm (after adjustment for better resonance). Between the Cavity and the waveguide we used an adapter. The measures are all correct. We simulated it in COMSOL and also Shawyer with his calculation program assured us that the dimensions we used were correct.
So the official internal dimensions from Tajmar are now:
Db = 108.2 mm
Ds = 77 mm
Height = 72.8 mm
Can you guys verify this cavity resonates in your frustumator software?
EDIT: A WR340 waveguide measures 86.36 x 43.18 mm so Tajmar used a coupling adapter reducing the waveguide.
http://forum.nasaspaceflight.com/index.php?topic=37642.msg1412868#msg1412868
According to TheTraveller's spreadsheet, based on those dimensions
Db = 108.2 mm
Ds = 77 mm
Height = 72.8 mm
with spherical ends, what mode shape did Tajmar excite?
According to my calculations it is (the equivalent of cylindrical mode) TM010 at ~2.45 GHz (*)
the lowest transverse magnetic mode possible.
I have modified the EM Drive wiki (that had TE111 based on a longer length of 0.1008 m instead of 0.0728m, and Db=0.1062m instead of 0.1082m and Ds=0.075m instead of 0.077m) to show these dimensions and mode shape.
_____________
(*) I write "the equivalent" because this mode is not constant in the longitudinal direction. It is impossible for a mode to be constant in the longitudinal direction for a truncated cone. I write "the equivalent" because this is the lowest TM mode for the truncated cone. It is a degenerate mode that corresponds to TM010 for a cylinder.
If this is true (TM010) i don't understand the orientation of the coupling waveguide, its the complete wrong orientation to excite this mode! May be thats why they have so small Q value 
In a cylinder this mode do not depend on the length.
The calculated TE111 resonance with this short length is at a higher frequency: 2.815 GHz. The TM010 resonance is at 2.43GHz with the dimensions provided by FluxCapacitor.
Maybe they didn't excite TE111 or TM010 well? Maybe they had TE111 mode participation at a small amplitude?
Maybe Tajmar's measurement is incorrect, the real length is 0.1 m and therefore they excited TE111 ?
Concerning <<In a cylinder this mode do not depend on the length.>> that's why I wrote the note (this TM mode for a truncated cone DOES depend on length):
(*) I write "the equivalent" because this mode is not constant in the longitudinal direction. It is impossible for a mode to be constant in the longitudinal direction for a truncated cone. I write "the equivalent" because this is the lowest TM mode for the truncated cone. It is a degenerate mode that corresponds to TM010 for a cylinder.
And no, they measured Q not Q_0, its not possible to measure Q_0 directly, one have to derive/calculate that from the complex measurement data.
Thanks for the paper.
After going through it side by side to a google translate version it seems you are saying the unloaded cavity Q is:
Q0 = (2 * Pi * Stored Energy) / Energy loss per cycle
and that the Q0 value can't be accurately measured via S11 max return loss dB at the 3dB down points?
Correct?
Only for the coupling factor equals to 1 the full 1/sqrt(2) BW is direct usable. The coupling factor can be derived from the measurements. Sorry the paper is german but i never find a better instructions manual.
Please look at page 21 in the green box..
And Q_0=Q*(1+coupling factor)
That paper is GOLD, even if it is in technical German. Split screening with the Google translation it reads fine.
Nice how it ties together the coupling factor k, the reflection coefficient p and the various forms of Q.
I note the reflection coefficient p can be calculated from the S11 max return loss dB at resonance and from that p value, the coupling factor k can be calculated and from that k value, the unloaded cavity Q can be calculated from the Q measured via the 3dB off the max return loss dB.
Correct?
...
On the top of my rotary table but under all the equipment will be a thin copper layer that will serve as my reference ground plane. Everything will be grounded to this high frequency ground reference plane.
...
Anybody who has ever designed and debugged a high frequency pcb knows the value of a solid copper ground plane.
So yes ground loops can be a problem but not if using a large area solid copper ground plane with additional high frequency decoupling capacitors and ferrite donuts.
At 2.4 GHz ferrite donuts are not needed. Instead an air core inductor inline with each power lead that is not at ground potential is all that is needed. Filter caps on either side of the inductor are used to prevent voltage sag from current transients. They are a high impedance path to RF. But since the inductor only needs to be a few nH. caps are not needed.
The fustrum is your ground plane. No thin copper sheet is needed. Any conductors going to this ersatz ground plane will be many wavelengths long. So it is impossible to make this copper sheet a ground at RF.
The biggest problem with driving a fustrum from a coax feed is matching the loop inside the cavity. It is virtually impossible to make the loop exactly the right length so it will almost never be a 50 Ohms resistive load at the frequency used. It will always be capacitive because of the adjacent cavity walls. So a large part of the RF sent into the cavity will just get reflected back on the shield. Ferrites are only useful up to VHF frequencies so putting ferrite donuts on the coax will have no effect. Maybe coiling the coax will help but a lot of RF will still get radiated because of the mismatch. Matching networks can be used but in that case the return wave is just dissipated in the matching network.
Quite right - there is a very significant difference between power grounding and RF grounding. At the appoximately 12cm wavelengths most folks are working with (2.45 Ghz or thereabout) any ground lead longer than about 6cm is also a pretty decent radiator. Google RF grounding - there is wealth of info about it. Power and safety ground are also very important.
Just a comment from those using loop antennas. Loop antennas are magnetic antenna and work differently from antennas like dipoles and monopoles. The common circumference for loop antennas is on the order of 1 lambda not lambda/2. A half wavelength circmumference will have a VERY high input impedance and will be hard to couple. A full wavelength loop will be significantly closer to 50 ohms, although wavelength variations due to excited mode and geometry of the cavity will also have significant effects. I will try to model up some example loops and patterns later today/tonight when I have access to the software.
Herman
My recollection is that the information (spherical ends and dimensions) is the product of personal e-mails exchanged between FluxCapacitor and Tajmar, if my memory is correct.
FluxCapacitor to confirm...
Found it.
http://forum.nasaspaceflight.com/index.php?topic=37642.msg1411835#msg1411835Spherical end plates seems so wrong when Prof Tajmar's frustum need a low Q to be able to accept all the wide band Rf the magnetron can produce. Something is not right.
...
On the top of my rotary table but under all the equipment will be a thin copper layer that will serve as my reference ground plane. Everything will be grounded to this high frequency ground reference plane.
...
Anybody who has ever designed and debugged a high frequency pcb knows the value of a solid copper ground plane.
So yes ground loops can be a problem but not if using a large area solid copper ground plane with additional high frequency decoupling capacitors and ferrite donuts.
At 2.4 GHz ferrite donuts are not needed. Instead an air core inductor inline with each power lead that is not at ground potential is all that is needed. Filter caps on either side of the inductor are used to prevent voltage sag from current transients. They are a high impedance path to RF. But since the inductor only needs to be a few nH. caps are not needed.
The fustrum is your ground plane. No thin copper sheet is needed. Any conductors going to this ersatz ground plane will be many wavelengths long. So it is impossible to make this copper sheet a ground at RF.
The biggest problem with driving a fustrum from a coax feed is matching the loop inside the cavity. It is virtually impossible to make the loop exactly the right length so it will almost never be a 50 Ohms resistive load at the frequency used. It will always be capacitive because of the adjacent cavity walls. So a large part of the RF sent into the cavity will just get reflected back on the shield. Ferrites are only useful up to VHF frequencies so putting ferrite donuts on the coax will have no effect. Maybe coiling the coax will help but a lot of RF will still get radiated because of the mismatch. Matching networks can be used but in that case the return wave is just dissipated in the matching network.
Quite right - there is a very significant difference between power grounding and RF grounding. At the appoximately 12cm wavelengths most folks are working with (2.45 Ghz or thereabout) any ground lead longer than about 6cm is also a pretty decent radiator. Google RF grounding - there is wealth of info about it. Power and safety ground are also very important.
Just a comment from those using loop antennas. Loop antennas are magnetic antenna and work differently from antennas like dipoles and monopoles. The common circumference for loop antennas is on the order of 1 lambda not lambda/2. A half wavelength circmumference will have a VERY high input impedance and will be hard to couple. A full wavelength loop will be significantly closer to 50 ohms, although wavelength variations due to excited mode and geometry of the cavity will also have significant effects. I will try to model up some example loops and patterns later today/tonight when I have access to the software.
Herman
The ground plane is approx 0.6m in diameter, as a solid copper sheet. It forms the bottom of the fine copper mesh Faraday Cages around the frustum and the Rf amp. Should be fairly difficult for such a large ground plane to resonate and radiate or form significant Dc ground offset nodes at 2.45GHz. Seems I will find out.
Look forward to your loop examples.
