R Shawyer wrote:

(a)
F ig7 from the 2021 paper now gives the dimensions of the dielectricdiscs and the loop orientation, which enabled the true operating modeto be determined.


The SPR designspreadsheet, described in our IAC-20 paper, was used to calculate theresonant frequency at the interface axial length of 188.6mm usingTM012 mode. (Note the loop orientation is for TM modes)
The resultingresonant frequency is 1894 MHz, compared to the measured frequency of1914 MHz, which is a typical offset due to manufacturing tolerances.
This showsclearly that the interface between the dielectric and vacuum sectionis not impedance matched, as suggested to Dresden in 2017.


(b)
The actualresonant geometry is therefore a tapered cavity with two flat endplates. The problems with flat end plates were explained in the SPRseminar given at Dresden on 11 July 2018. The simple geometry shownin slide 15 of that seminar, is reproduced below.








For flat endplates
Path lengtherror = Ls-La
Total error=Q(Ls-La) leads to wavefront phase error and reduction of Q
End plates mustbe shaped and aligned to minimise mean path length error.




Application ofthis geometry to the Dresden cavity shows that the wave-front phaseerror reaches half a wavelength (87.9mm), at a Q of 14, so the cavitywill suffer a negligible unloaded Q. The hugely distorted wave-frontwill result in multiple reflections off the sidewalls rendering thecavity unable to support linear travelling waves. Any trivial forcesin the end plates will be balanced by sidewall forces, resulting inzero net thrust.
(c)
In section 5.2,the paper states:
For fast andprecise tuning, a single port Vector Network Analyzer (VNA) AnritsuMS46121A was used.
Therefore theapparent Qu of 23,000 and input return loss that have been measuredis simply the performance of the input loop because there was noattempt at impedance matching between the loop and the cavity. Thisrequires two port measurement using a detector probe. The importanceof Impedance matching between the loop and cavity was emphasised inslide 14 of the 2018 Dresden seminar.


(d)
From Fig 8 in thepaper, it appears that loop tuning merely comprised a 3 stub coaxialtuner between the circulator and input. There is no method ofactually tuning the loop itself. Thus virtually all the input poweris dissipated in the loop, and there is no coupling of anysignificant power into the cavity itself.
This is clearlyshown by the reported very high return loss of 45dB, which isindicative of an input loop matched only to the source impedance.Correct coupling into the cavity usually gives a much lower returnloss, typically around 20dB.
A copy of thedrawing of the SPR Flight Thruster loop assembly was sent to Dresdenon 10 October 2018 to illustrate the typical tuning mechanismrequired.


4.Conclusions
Whilst therational for manufacturing and testing a replica NASA cavity,following our correspondence with Dresden since 2017, is notunderstood, the result of almost 4 years of effort has merelyconfirmed our original advice. The cavity, so carefully tested,clearly cannot operate as an EmDrive thruster, and should not even bedescribed as such.
This result isalso confirmed by three other organisations, NRL, Toulouse Universityand a private US Research organisation.
All attempts to produce working EmDrive thrusters must be carried outwith a clear understanding of the theory of EmDrive, and withsufficient experience of microwave engineering to avoid the mistakesidentified in these notes.