Chris has removed the general moderator - by mutual agreement - for this section and assigned me, an EM Drive regular, as the new moderator.- Dave (rfmwguy)
I did a little Googling and found that many meep users (not just here - everywhere) ran into the negative Q problem. It is an artifact of the way meep works, and is an indication that the simulation did not run long enough. Decreasing the bandwidth helps too.After some other attempts, hunting around, I finally set BW to 0.015, set the center frequency to 2.4959 GHz where it wanted to resonate anyway, and increased the runtime by a factor of 1.25. (aero: I did this by increasing 'gc' from 8 to 10) Now I get a positive Q, that is closer to Rodal's earlier result. I got Q=99,938. Also absolute amplitude went up from 4.9 to 23.5. Error is 4.12E-7. Runtime was about 2 hours. Resonant frequncy came out 2.4959-1.248E-5i.The equation I used for CU-D-conduct (which is the imaginary part of the relative complex permittivity (I should change the variable name) was posted earlier, -2*pi*2.4GHz*3.252698E+8i. That evaluates to -4.9083E+9. The "2.4GHz" term is actually in "meep units" so it is 2.4E+9 times 'a' divided by 'c'.
Quote from: rfmwguy on 12/11/2015 02:51 pmChris has removed the general moderator - by mutual agreement - for this section and assigned me, an EM Drive regular, as the new moderator.- Dave (rfmwguy)I, for one, welcome our new RF & microwave overlord!
The MEEP input properties for the conductivity of copper ( MEEP:-2*pi*2.4GHz*3.252698E+8i) are very high, corresponding to a copper of practically 100% purity.
Quote from: Rodal on 12/11/2015 03:08 pmThe MEEP input properties for the conductivity of copper ( MEEP:-2*pi*2.4GHz*3.252698E+8i) are very high, corresponding to a copper of practically 100% purity.I think I got that number 3.252698E+8i from one of your own posts. I have looked around the web for reference materials on more realistic real-world values but have not found a good one yet. What units is that in? It does not look like Siemens/meter, which is 58.6E6 for Cu.
only for f=2.4 E+09 Hertz one hasε“/εo = 3.252698 E+08while for f = 2.45 E+09 Hertz (for example) one hasε“/εo = 3.186316 E+08
The conductivity in SI Units that corresponds to epsilon"=0.00288 is:conductivity = omega * epsilon" = 2 Pi frequency 0.00288 = 2 Pi 2.4E+9 * 0.00288 = 4.342937 E+7which is almost 10 times smaller than 3.25E+8Please also recall that DeltaMass was conscious that it is incorrect to take 0.00288 as a constant !That value is a function of frequency.What is approximately constant in this regime is the conductivity itself.DeltaMass gave you explicit instructions to keep the conductivity constant, at other frequencies: for example, the conductivity at 1 GHz is also 4.342937 E+7, so at 1 GHz you should input into Meep (3.25...E+8 ) *2.4 = 7.8 E+8, for exampleSo, for pure Silver, for example (from the table in the link in your post), Conductivity = 6.090E+07So, instead of 3.25...E+8 for copper, you have to use ( 6.090E+07/4.342937 E+7 ) *3.25...E+8 at 2.4GHzin other words, at 2.4 GHz, your input to Meep for pure Silver should be 1.402276 times higher than for the copper value given by DeltaMass. About 40% higher, whether in SI units or in Meep units.In other words, everything else being the same, the quality of resonance (Q) should be about 40% higher with pure silver than with copper.
Re: EM Drive Developments - related to space flight applications - Thread 5« Reply #2538 on: Today at 02:54 PM »Quote from: ThereIWas3I did a little Googling and found that many meep users (not just here - everywhere) ran into the negative Q problem. It is an artifact of the way meep works, and is an indication that the simulation did not run long enough. Decreasing the bandwidth helps too.After some other attempts, hunting around, I finally set BW to 0.015, set the center frequency to 2.4959 GHz where it wanted to resonate anyway, and increased the runtime by a factor of 1.25. (aero: I did this by increasing 'gc' from 8 to 10) Now I get a positive Q, that is closer to Rodal's earlier result. I got Q=99,938. Also absolute amplitude went up from 4.9 to 23.5. Error is 4.12E-7. Runtime was about 2 hours. Resonant frequncy came out 2.4959-1.248E-5i.The equation I used for CU-D-conduct (which is the imaginary part of the relative complex permittivity (I should change the variable name) was posted earlier, -2*pi*2.4GHz*3.252698E+8i. That evaluates to -4.9083E+9. The "2.4GHz" term is actually in "meep units" so it is 2.4E+9 times 'a' divided by 'c'. That's EXCELLENT work, @ThereIWas3, I also got Q's of ~90,000 using the exact solution for a (previous ?) geometry of the Shell Frustum (she has posted more than one geometrical shape for possible testing in the last year), depending on the input parameters (conductivity of copper, etc.). Please let us know what mode shape is present in the EM Drive when you have a chance, or post some images of the fields when you have a chance.It is great to have more people running Meep. Excellent work !The MEEP input properties for the conductivity of copper ( MEEP:-2*pi*2.4GHz*3.252698E+8i) are very high, corresponding to a copper of practically 100% purity. Using impure copper, together with an imperfect geometry and surface reflectivity, will bring down the quality of resonance Q to a lower value in actual testing.______PS: Again: no negative mass/negative energy, no dark mass, no dark energy, no leaky fields, no strange quantum effects, no microwave black magic, were responsible for a negative Q. It was just a numerical artifact of the finite difference solution.
Quote from: ThereIWas3 on 12/11/2015 03:29 pmQuote from: Rodal on 12/11/2015 03:08 pmThe MEEP input properties for the conductivity of copper ( MEEP:-2*pi*2.4GHz*3.252698E+8i) are very high, corresponding to a copper of practically 100% purity.I think I got that number 3.252698E+8i from one of your own posts. I have looked around the web for reference materials on more realistic real-world values but have not found a good one yet. What units is that in? It does not look like Siemens/meter, which is 58.6E6 for Cu.The background of that number is in my previous posts, with luxury of details, including a discussion of units and a discussion of a conversion to usual units.My posts (as any user posts) can be searched by clicking on my NSF membership, and clicking "Show Posts"For example (one of many posts on this subject):http://forum.nasaspaceflight.com/index.php?topic=38577.msg1453316#msg1453316Basically it is based on a model proposed, for pure copper, by @DeltaMass.It corresponds to a conductivity in SI units of 4.342937 E+7As I have explained in previous messages:1) For other material conductivities you simply have to ratio this input by the material conductivities2) The input to MEEP should be changed for different frequencies. The pure copper input is ONLY valid for 2.4 GHz. At other frequencies the input should be linearly ratioed by the frequency ratio, so that the conductivity stays constantAt resonating frequencies higher than 2.4 GHz, you should input a correspondingly LOWER number to keep the conductivity constant.At lower frequencies than 2.4 GHz you should input a correspondingly HIGHER number, so that the conductivity stays at the correct constant value.As I wrote in http://forum.nasaspaceflight.com/index.php?topic=38577.msg1453316#msg1453316:Quoteonly for f=2.4 E+09 Hertz one hasε“/εo = 3.252698 E+08while for f = 2.45 E+09 Hertz (for example) one hasε“/εo = 3.186316 E+08Quote from: RodalThe conductivity in SI Units that corresponds to epsilon"=0.00288 is:conductivity = omega * epsilon" = 2 Pi frequency 0.00288 = 2 Pi 2.4E+9 * 0.00288 = 4.342937 E+7which is almost 10 times smaller than 3.25E+8Please also recall that DeltaMass was conscious that it is incorrect to take 0.00288 as a constant !That value is a function of frequency.What is approximately constant in this regime is the conductivity itself.DeltaMass gave you explicit instructions to keep the conductivity constant, at other frequencies: for example, the conductivity at 1 GHz is also 4.342937 E+7, so at 1 GHz you should input into Meep (3.25...E+8 ) *2.4 = 7.8 E+8, for exampleSo, for pure Silver, for example (from the table in the link in your post), Conductivity = 6.090E+07So, instead of 3.25...E+8 for copper, you have to use ( 6.090E+07/4.342937 E+7 ) *3.25...E+8 at 2.4GHzin other words, at 2.4 GHz, your input to Meep for pure Silver should be 1.402276 times higher than for the copper value given by DeltaMass. About 40% higher, whether in SI units or in Meep units.In other words, everything else being the same, the quality of resonance (Q) should be about 40% higher with pure silver than with copper.
Quote from: Rodal on 12/11/2015 03:32 pmQuote from: ThereIWas3 on 12/11/2015 03:29 pmQuote from: Rodal on 12/11/2015 03:08 pmThe MEEP input properties for the conductivity of copper ( MEEP:-2*pi*2.4GHz*3.252698E+8i) are very high, corresponding to a copper of practically 100% purity.I think I got that number 3.252698E+8i from one of your own posts. I have looked around the web for reference materials on more realistic real-world values but have not found a good one yet. What units is that in? It does not look like Siemens/meter, which is 58.6E6 for Cu.The background of that number is in my previous posts, with luxury of details, including a discussion of units and a discussion of a conversion to usual units.My posts (as any user posts) can be searched by clicking on my NSF membership, and clicking "Show Posts"For example (one of many posts on this subject):http://forum.nasaspaceflight.com/index.php?topic=38577.msg1453316#msg1453316Basically it is based on a model proposed, for pure copper, by @DeltaMass.It corresponds to a conductivity in SI units of 4.342937 E+7As I have explained in previous messages:1) For other material conductivities you simply have to ratio this input by the material conductivities2) The input to MEEP should be changed for different frequencies. The pure copper input is ONLY valid for 2.4 GHz. At other frequencies the input should be linearly ratioed by the frequency ratio, so that the conductivity stays constantAt resonating frequencies higher than 2.4 GHz, you should input a correspondingly LOWER number to keep the conductivity constant.At lower frequencies than 2.4 GHz you should input a correspondingly HIGHER number, so that the conductivity stays at the correct constant value.As I wrote in http://forum.nasaspaceflight.com/index.php?topic=38577.msg1453316#msg1453316:Quoteonly for f=2.4 E+09 Hertz one hasε“/εo = 3.252698 E+08while for f = 2.45 E+09 Hertz (for example) one hasε“/εo = 3.186316 E+08Quote from: RodalThe conductivity in SI Units that corresponds to epsilon"=0.00288 is:conductivity = omega * epsilon" = 2 Pi frequency 0.00288 = 2 Pi 2.4E+9 * 0.00288 = 4.342937 E+7which is almost 10 times smaller than 3.25E+8Please also recall that DeltaMass was conscious that it is incorrect to take 0.00288 as a constant !That value is a function of frequency.What is approximately constant in this regime is the conductivity itself.DeltaMass gave you explicit instructions to keep the conductivity constant, at other frequencies: for example, the conductivity at 1 GHz is also 4.342937 E+7, so at 1 GHz you should input into Meep (3.25...E+8 ) *2.4 = 7.8 E+8, for exampleSo, for pure Silver, for example (from the table in the link in your post), Conductivity = 6.090E+07So, instead of 3.25...E+8 for copper, you have to use ( 6.090E+07/4.342937 E+7 ) *3.25...E+8 at 2.4GHzin other words, at 2.4 GHz, your input to Meep for pure Silver should be 1.402276 times higher than for the copper value given by DeltaMass. About 40% higher, whether in SI units or in Meep units.In other words, everything else being the same, the quality of resonance (Q) should be about 40% higher with pure silver than with copper.OK, as an example, let's calculate some numbers:1) Since the resonant frequency you calculated was 2.4959 GHz instead of 2.4 GHz, the input instead of 3.252698 E+08 should have been:(2.4/2.4959)*3.252698 E+08 = 3.127719 E+08Therefore, your MEEP run's output quality of resonance Q instead of Q=99,938, would have been, at 2.4959 GHz, for DeltaMass pure copper:Q = (2.4/2.4959)*99,938 = 96,098__________________________________________2) If instead of copper's conductivity, as proposed by DeltaMass, you would have used the value for conductivity for Bronze, Commercial (Annealed) in (http://eddy-current.com/conductivity-of-metals-sorted-by-resistivity/): 2.552E+07, the Q output would have been:Q = (2.4/2.4959)*99,938*(2.552E+07)/(4.342937 E+7) =56,469and the input should have been:(2.4/2.4959)*((2.552E+07)/(4.342937 E+7))*3.252698 E+08 = 1.837912 E+08__________________________________________3) If instead of copper's conductivity, as proposed by DeltaMass, you would have used the value for conductivity for Pure Silver in (http://eddy-current.com/conductivity-of-metals-sorted-by-resistivity/): 6.090E+07, the Q output would have been:Q = (2.4/2.4959)*99,938*(6.090E+07)/(4.342937 E+7) =134,756and the input should have been:(2.4/2.4959)*((6.090E+07)/(4.342937 E+7))*3.252698 E+08 =
Quote from: Rodal on 12/11/2015 05:12 pmQuote from: Rodal on 12/11/2015 03:32 pmQuote from: ThereIWas3 on 12/11/2015 03:29 pmQuote from: Rodal on 12/11/2015 03:08 pmThe MEEP input properties for the conductivity of copper ( MEEP:-2*pi*2.4GHz*3.252698E+8i) are very high, corresponding to a copper of practically 100% purity.I think I got that number 3.252698E+8i from one of your own posts. I have looked around the web for reference materials on more realistic real-world values but have not found a good one yet. What units is that in? It does not look like Siemens/meter, which is 58.6E6 for Cu.The background of that number is in my previous posts, with luxury of details, including a discussion of units and a discussion of a conversion to usual units.My posts (as any user posts) can be searched by clicking on my NSF membership, and clicking "Show Posts"For example (one of many posts on this subject):http://forum.nasaspaceflight.com/index.php?topic=38577.msg1453316#msg1453316Basically it is based on a model proposed, for pure copper, by @DeltaMass.It corresponds to a conductivity in SI units of 4.342937 E+7As I have explained in previous messages:1) For other material conductivities you simply have to ratio this input by the material conductivities2) The input to MEEP should be changed for different frequencies. The pure copper input is ONLY valid for 2.4 GHz. At other frequencies the input should be linearly ratioed by the frequency ratio, so that the conductivity stays constantAt resonating frequencies higher than 2.4 GHz, you should input a correspondingly LOWER number to keep the conductivity constant.At lower frequencies than 2.4 GHz you should input a correspondingly HIGHER number, so that the conductivity stays at the correct constant value.As I wrote in http://forum.nasaspaceflight.com/index.php?topic=38577.msg1453316#msg1453316:Quoteonly for f=2.4 E+09 Hertz one hasε“/εo = 3.252698 E+08while for f = 2.45 E+09 Hertz (for example) one hasε“/εo = 3.186316 E+08Quote from: RodalThe conductivity in SI Units that corresponds to epsilon"=0.00288 is:conductivity = omega * epsilon" = 2 Pi frequency 0.00288 = 2 Pi 2.4E+9 * 0.00288 = 4.342937 E+7which is almost 10 times smaller than 3.25E+8Please also recall that DeltaMass was conscious that it is incorrect to take 0.00288 as a constant !That value is a function of frequency.What is approximately constant in this regime is the conductivity itself.DeltaMass gave you explicit instructions to keep the conductivity constant, at other frequencies: for example, the conductivity at 1 GHz is also 4.342937 E+7, so at 1 GHz you should input into Meep (3.25...E+8 ) *2.4 = 7.8 E+8, for exampleSo, for pure Silver, for example (from the table in the link in your post), Conductivity = 6.090E+07So, instead of 3.25...E+8 for copper, you have to use ( 6.090E+07/4.342937 E+7 ) *3.25...E+8 at 2.4GHzin other words, at 2.4 GHz, your input to Meep for pure Silver should be 1.402276 times higher than for the copper value given by DeltaMass. About 40% higher, whether in SI units or in Meep units.In other words, everything else being the same, the quality of resonance (Q) should be about 40% higher with pure silver than with copper.OK, as an example, let's calculate some numbers:1) Since the resonant frequency you calculated was 2.4959 GHz instead of 2.4 GHz, the input instead of 3.252698 E+08 should have been:(2.4/2.4959)*3.252698 E+08 = 3.127719 E+08Therefore, your MEEP run's output quality of resonance Q instead of Q=99,938, would have been, at 2.4959 GHz, for DeltaMass pure copper:Q = (2.4/2.4959)*99,938 = 96,098__________________________________________2) If instead of copper's conductivity, as proposed by DeltaMass, you would have used the value for conductivity for Bronze, Commercial (Annealed) in (http://eddy-current.com/conductivity-of-metals-sorted-by-resistivity/): 2.552E+07, the Q output would have been:Q = (2.4/2.4959)*99,938*(2.552E+07)/(4.342937 E+7) =56,469and the input should have been:(2.4/2.4959)*((2.552E+07)/(4.342937 E+7))*3.252698 E+08 = 1.837912 E+08__________________________________________3) If instead of copper's conductivity, as proposed by DeltaMass, you would have used the value for conductivity for Pure Silver in (http://eddy-current.com/conductivity-of-metals-sorted-by-resistivity/): 6.090E+07, the Q output would have been:Q = (2.4/2.4959)*99,938*(6.090E+07)/(4.342937 E+7) =134,756and the input should have been:(2.4/2.4959)*((6.090E+07)/(4.342937 E+7))*3.252698 E+08 =How thick would the silver plating (over) need to be, to model the frustum as pure silver? And how might a thin playing of gold, to protect the silver plating affect things?
Ref long thread... http://forum.nasaspaceflight.com/index.php?topic=39004.msg1455988#msg1455988From what I've read I'm around 40-50 um and may put another coat but I don't think it's needed as RF penetration is under 5um.I think I'll keep silver for now. 1) it develops oxides quite slowly (takes sulfur around it) and it also cleans very easy. 2) To lay down a electroplated uniform layer of gold that remains under 1um is very hard to do with a hand method and it will need to wait for a professional plating shop specializing in waveguides.Shellmodded, Stupiddd speeeling korector
Stupid question: how does a spacecraft propelled by an EMDrive (if it works) slow down? The basic answer is by creating a force in the opposite direction. What happens to the energy that the drive has been loading into the spaceship? It can't simply be destroyed, it has to go somewhere. Where?
Quote from: SeeShells on 12/11/2015 05:51 pmRef long thread... http://forum.nasaspaceflight.com/index.php?topic=39004.msg1455988#msg1455988From what I've read I'm around 40-50 um and may put another coat but I don't think it's needed as RF penetration is under 5um.I think I'll keep silver for now. 1) it develops oxides quite slowly (takes sulfur around it) and it also cleans very easy. 2) To lay down a electroplated uniform layer of gold that remains under 1um is very hard to do with a hand method and it will need to wait for a professional plating shop specializing in waveguides.Shellmodded, Stupiddd speeeling korectorI don't think from the information above, that gold plating would be worthwhile, funtionally or cost wise, as far as a test design is concerned. It would be good to know if silver electroplating can be modeled as pure silver or if it needs to be adjusted in some way.
I will correct the computation of permittivity to use the actual signal frequency rather than the fixed 2.4 GHz it is now. I am trying to eliminate "magic constants" in the code as much as possible, so everything automatically tracks the input model data.Edit: meep does not do well at simulating very thin layers, unless you set the lattice size really small, which increases computation time enormously. So I would keep the current thickish material specification and set the permitivity somewhere between Cu and Ag. The current thickness is greater than real life for the same reason. We do not need to simulate the escape of fields outside for the current purposes.