All this talk of Q is interesting, and it does concern me in the sense that we need to see long runs from meep but the question to me becomes, "How long should the run be?" And there, the definition of Q becomes a problem. I have attached an image of what Wikipedia says about it. Which one are we using?The problem, as it looks to me is that it will take several thousand cycles for the energy stored to reach steady state, everything before that is transient. So again, "How long should the run be?"
This is what I'm using...0.0625 holes with 3/32 stagger spacing copper perforated sheet ~.020 thick. Have thicker on order but needs to be made..0625 inch hole works out to 188.9 GHz wavelength.
Excuse my ignorance of the basic algorithm used in this Finite Difference technique, but is it the case that, as the name implies, total simulation error accumulates with the number of cycles simulated? That being so, is there a way of knowing how many cycles it takes before the error causes significant non-physical divergence of the simulation results? I ask because a few hundred thousand cycles might be out of reach for that reason.
Quote from: aero on 07/11/2015 05:36 pmAll this talk of Q is interesting, and it does concern me in the sense that we need to see long runs from meep but the question to me becomes, "How long should the run be?" And there, the definition of Q becomes a problem. I have attached an image of what Wikipedia says about it. Which one are we using?The problem, as it looks to me is that it will take several thousand cycles for the energy stored to reach steady state, everything before that is transient. So again, "How long should the run be?"The generic calculation formulas you posted from Wikipedia, are not an issue, as they give similar results. The issue, as excellently discussed by zen-in, rfcavity, rfmwguy and others has to do with how to experimentally measure power and calculate loaded Q based on S11 (and/or S21 -there are papers on figuring out loaded Q using both S11 and S21). Dr. Notsosureofit who has very long-term experience in this field, has posted that he has always used phase rather than return loss for calculating Q.Yang instead of using the S11 zero dB reference plane to measure their -3dB down bandwidths from, as is done elsewhere, uses the most negative dB S11 value located at the resonance frequency and measure up 3dB toward the S11 zero dB plane. Therefore, of course, the bandwidth figures used by the Chinese in this unorthodox calculation are going to be ridiculously small which yields correspondingly artificially large values of the calculated Q-factor.The issue of how long you would have to march the Finite Difference solution to reach steady-state (if steady-state is achievable with the RF feed on ) can only be addressed by solving the transient solution for a truncated cone cavity, and that cannot be done exactly because there is no such exact solution. Others have posted that you would have to get close to a microsecond. This means that you would have to have from 1 to 2 orders of magnitude greater number of time steps, as your present run represents only 0.013063 microseconds of transient response.That means that instead of the 6,527 time steps now, you would need from 65,000 to 650,000 FD time steps.By the way, it is known in the numerical solution literature than experimental Q's are usually much lower than calculated Q's, as losses are usually underestimated.
In physics and engineering the quality factor or Q factor is a dimensionless parameter that describes how under-damped an oscillator or resonator is, as well as characterizes a resonator's bandwidth relative to its center frequency.
This was discussed back in May:http://forum.nasaspaceflight.com/index.php?topic=36313.msg1369553#msg1369553Quote from: Rodal Paul March has addressed and explained this as follows: Chinese (Prof. Yang) calculated loaded Q factors are much higher than the Q's reported by Shawyer and by NASA' Eagleworks because of the unorthodox way that the Chinese calculate their loaded Q factors. Instead of using the S11 zero dB reference plane to measure their -3dB down bandwidths from, as is done elsewhere, the Chinese use the most negative dB S11 value located at the resonance frequency and measure up 3dB toward the S11 zero dB plane. Therefore, of course, the bandwidth figures used by the Chinese in this unorthodox calculation are going to be ridiculously small which yields correspondingly artificially large values of the calculated Q-factor. .
Paul March has addressed and explained this as follows: Chinese (Prof. Yang) calculated loaded Q factors are much higher than the Q's reported by Shawyer and by NASA' Eagleworks because of the unorthodox way that the Chinese calculate their loaded Q factors. Instead of using the S11 zero dB reference plane to measure their -3dB down bandwidths from, as is done elsewhere, the Chinese use the most negative dB S11 value located at the resonance frequency and measure up 3dB toward the S11 zero dB plane. Therefore, of course, the bandwidth figures used by the Chinese in this unorthodox calculation are going to be ridiculously small which yields correspondingly artificially large values of the calculated Q-factor. .
Quote from: Rodal on 07/11/2015 06:01 pmThis was discussed back in May:http://forum.nasaspaceflight.com/index.php?topic=36313.msg1369553#msg1369553Quote from: Rodal Paul March has addressed and explained this as follows: Chinese (Prof. Yang) calculated loaded Q factors are much higher than the Q's reported by Shawyer and by NASA' Eagleworks because of the unorthodox way that the Chinese calculate their loaded Q factors. Instead of using the S11 zero dB reference plane to measure their -3dB down bandwidths from, as is done elsewhere, the Chinese use the most negative dB S11 value located at the resonance frequency and measure up 3dB toward the S11 zero dB plane. Therefore, of course, the bandwidth figures used by the Chinese in this unorthodox calculation are going to be ridiculously small which yields correspondingly artificially large values of the calculated Q-factor. .Here is where they went wrong...under no industrial RF standard does anyone measure Q on return loss, S11. It is done on S21, forward power in the frequency domain for cavities. I stand by my claim that "Specsmanship" was used to create an unnaturally large Q, either by unfamiliarity or intent.Note that S21 requires a 2 port measurement, input and output (note the sampling port on the frustums will provide the output). I'd bet a six-pack of craft beer that realistic Qs are in the 4 digit range for both shawyer and yang. And yes Doc, Yang should have used the -3dB points below 0 insertion, not -3dB above best return loss...not RF types IMHO.
I will state my case again:Why does the waveguide industry NOT make cavities of copper mesh?The EMDrive is a tapered waveguide.
...I will state my case again:Why does the waveguide industry NOT make cavities of copper mesh?The EMDrive is a tapered waveguide.
@aero et al - regarding longer MEEP runsIf we can figure out how to offload the MEEP computation to the cloud, I would be willing to contribute server time (within reason, of course). I suspect others may be willing to donate server time, too.Have you heard of the FEFF project? Link here: http://www.feffproject.org/feffproject-scc-caseexamples-meep.html. Seems they've done some of the heavy lifting to link MEEP and Amazon AWS. Might be helpful..
Somebody posted this a couple of hours ago in another EM Drive forum:<<You need to ask yourself why copper mesh is not used in the microwave industry to build waveguides? Would be heaps lighter, lower weight and cost.Might be because mesh it is good at absorbing microwave energy that strikes it but bad at reflecting / propogating microwave energy that strikes it.The inside of your cavity needs to be very highly polished, ding & scratch free rigid copper that reflects and propogates microwave energy with VERY little energy loss, instead of absorbing the energy and turning it into heat.>>That's not correct, actually in the aerospace industry the use of a mesh is quite common:" Why is my Satellite Dish full of holes?" http://www.thenakedscientists.com/forum/index.php?topic=16208.0"A study of microwave transmission perforated flat plates" http://ipnpr.jpl.nasa.gov/progress_report2/II/IIO.PDFWikipedia <<With lower frequencies, C-band for example (IEEE C 4 – 8 GHz), dish designers have a wider choice of materials. The large size of dish required for lower frequencies led to the dishes being constructed from metal mesh on a metal framework. At higher frequencies, mesh type designs are rarer though some designs have used a solid dish with perforations>>http://www.yldperforatedmetal.com/Perforated-Metal-Screen.htm
Microwave magnetrons can have a +-30MHz bandwidth. If you designed a cavity bandwidth wide enough to suck all that up, the Q would be around 41.
I'm just looking for a possible hybrid mode on cavity. But it is not so easy.There are some possibles candidates on Nasa's paper.Some formulas of sensitivity can be used for adjust the dimensions of cavity and to control the frequencys.Very cool.By the way. In corrugated waveguides, hybrid modes have very low losses. In cavity, perhaps they produce more higher Q.