Paul March (then head EW engineer) did a scan of the cavity minus dielectrics. Resonant mode, with best Q was TE012, 2.167GHz, 3.85mN/kWrf
This is public info so why did Tajmar use a freq that was not cavity mode resonant & was not the highest Q?
Lorentz was measured & applied.
Those photos seems are about EW's 2014 paper. Their way of measuring Lorentz force was by replacing the cavity with a dummy load. However, the grounding schemes and lead configurations between the cavity exp and the dummy load exp were different, thus ground loop DC current were different, thus their way of measuring Lorentz force did not work. This was the main point of our 2015 arxiv paper, which was cited by Tajmar in his most recent paper.
What Paul did was to remove the cavity, replacing it with a dummy load at the same location, moving nothing else.
True but the grounding scheme and lead configuration were also changed.
This discussion is about Tajmar exciting his cavity at a freq it was not resonant at.
If you have cavity resonant data that supports a resonant mode at 1.865GHz for the cavity without a dielectric please post it.
I knew. I posted the unrelated reply because you added an unrelated statement to the discussion, "Lorentz was measured & applied." I just do not want people to be misled into believing that "3.85mN/kWrf" was real.
White et al. (2017) in "Measurement of Impulsive Thrust from a Closed Radio-Frequency Cavity in Vacuum", Journal of Propulsion and Power, 33(4), part C "Vacuum campaign", section 8 "Error Sources", point 3 page 838:
"The third error is magnetic interaction, which has the potential for a false positive resulting from dc currents in power cables interacting during test article operation with ambient magnetic fields (e.g., local Earth field, magnetic damper) to generate a torque displacement on the pendulum. All dc power cables are a twisted pair or twisted shielded pair to minimize magnetic interaction. The test article is tested in forward, reverse, and null thrust orientations, but dc power cable routing and orientation is the same for all three configurations (power cables come in from the top of the test article), meaning any false positives will be the same magnitude and polarity for all three tests. This is not observed during the test campaign."
Eagleworks did not address the potential issue of DC ground loops, but their wires and electronics being fixed, such currents would have been the same whatever the cavity direction, so they would have been detected and could have been quantified if they occurred and affected the cavity behavior during forward and reverse tests. In such two tests, the grounding scheme was the same.
This is their 2017 paper. We were talking about their 2014 paper. Seems they learned the lesson and in their 2017 paper there was probably no Lorentz problem but heat problem.
1,880.6 GHz at TE012 with 2 dielectric discs is the data Paul shared.
I can't find any non dielectric resonant mode at 1.865GHz. Can you?
Plus the way too low VNA rtn loss of -55dB strongly suggests it is a self resonant coupler freq.
Please check what I wrote again: " TE212 found at 1880 Mhz (1.88 Ghz)" I did not write TE012 at 1.88 Ghz.
However, I think you could be correct about them exciting a resonant coupler and not the cavity as if you look closely, there is a RL dip at 1.88 Ghz, where we would expect TE212. There shouldn't be anything at 1.865 Ghz.
Just got on this morning reading these posts and I must say you both seemed to have shown why the Tajmar's team only showed Lorentz forces. Very nice detective work.
My Best,
Shell
I am disappointed.
Monomorphic and SeeShells, you have demonstrated with your experiments a commitment to the scientific method. But now on the prompting of one poster here who has repeatedly failed at entry-level physics, you jump on an explanation without any empirical evidence.
Never mind that no one has ever measured a large peak in any experiment reported here due to a "resonant coupler."
Never mind that it makes no sense for such a coupler to have such a high Q despite all the difficulty that everyone here knows goes into making high Q systems.
Never mind that they have a 3 stub tuner keeping more energy in the RF cavity, leading to lower reflections, so they should have a deep trough at the tuned frequency.
Never mind that TT claimed "not confirmed by COMSOL" in a post where he had attached a picture of the COMSOL simulation they did.
You apparently are giving up on science, and taking the path of jumping on the first explanation you are given that contradicts an experiment with a result you don't like.
McCulloch's opinion of the media reaction to the Tajmar EmDrive test:
Is everyone who has looked at the emDrive losing their minds? McCulloch has seemed reasonable before, but unless this was accompanied by an actual detailed explanation of a problem with the experiment, this comic is the equivalent of him plugging his ears and loudly saying "LALALA I CAN'T HEAR YOU."
His comparison of this experiment to a dead chicken rather than a live one is significantly worse than any media treatment of this I have seen.
Forget your issues with TheTraveler, that is not the hot topic here.
Don't forget "There is no bad data", although it seems good data was interpreted badly here with Tajmar's team. And it seems
everyone has jumped on his null tests without seriously evaluating his work.
In response to your post. I was questioning
everything in Martin Tajmar's presentation and what was a resonate frequency 55db dip.
Including calculating out what mode they were running for the cavity dimensions. Cavity dimensions they closely and methodically built to match EagleWorks. Kept asking myself why they were
15MHz off in the frequency from EagleWorks resonate scans and operating mode. This was even before the issue with the coupler came up. As pointed out the numbers didn't match up to a resonate mode.
Whether I like or dislike test results (don't care either way if I'm objective) does not color the questions in lab test stand build or frustum build that need to be brought up. The builders here have been pushed and prodded to eliminate and categorize errors all of our home built test stands and we have taken years to do just that on a itsy bittsy budget. We should have the right to question other builds and data where we see glaring errors.
Shell
.......
A -55dB rtn loss from the S11 VNA scan represents a VSWR of 1.004:1 which is an impossible real world result.
Yet it is there.
One explication is it was caused by coupler self resonance, which ignores cavity eddy current losses.
BTW a -25dB S11 rtn loss is an excellent real world result, VSWR 1.2:1, when dealing with a high Q EmDrive.
Maybe ask Jamie what S11 rtn loss his cavity generates?
I emailed Martin Tajmar and he was kind enough to answer. I've put the input he was responding to in quotes. I've also stripped extraneous identifying email and other addresses.
You can pass on my response to the forum.
In general, I’m very surprised by this huge response. We posted preliminary results/work in progress, which was clearly stated in the paper. Work is ongoing, we will measure all the different frequencies/geometries/dielectric discs/higher power levels, etc. and hopefully publish the full analysis next year. So please be patient, more to come.
Regading your questions:
..... it would be that in none of the tests did RF energy reach the frustrum, thereby explaining why the application of the choke had no effect....
We did measure the RF-power going through our bi-directional coupler from the amplifier to the 3-stub-tuner and the cavity. We also measured the reflected power coming from the cavity. The input power was clearly bigger than the reflected power. So power can only be "lost" in the cavity. By tracking the minimum reflected power we always remained at resonance even during warming up of the cavity. Before test trials we did measure the reflection behaviour of the antenna inside the cavity (with NWA). Which means, that the antenna would radiate most of the inserted power (app. 80%) at certain frequencies which are the resonance frequencies of the cavity (standing wave, low power reflection back to antenna from cavity). Using the attenuator, we still measured the same input power, but the reflected power was so low, that our powermeters were not able to detect power anymore.
I was intruiged to note the suggestion that the Qfactor at that frequency was way to high, and note that Rodal and others had queried the Qfactor you noted in your earlier presentation.
In the design phase, we reached a Q-factor (with NWA, unstable) of about 300.000 and more. For test runs, a stabile Q-factor of app. 60,000 was measured with the NWA. This is the UNLOADED Q. Because of reflection at every RF-connector, lower radiation of the antenne due to warming up etc. the loaded Q is much smaller (around 700-1,000). This applies to the Eagleworks setup as well.
This is the VNA sweep that looks for resonant rtn loss dips. Note how far down is the 1.865GHz dip when compared to the others. I have identified the modes of the other dips but not the 1.865GHz dip. A 55dB dip is not real especially when not confirmed by COMSOL or other such resonance analysis.
This is a graph of the exported data from the NWA (unloaded case). The 1865MHz are matched with the 3-stub-tuner as good as possible. That means, the system antenna/cavity has a input resistance of nearly 50ohm. So there should be a minimum of power reflection at exact this frequency to the 50ohm-power-supply-system. So, most of provided power goes into cavity.
This suggests Tajmar was exciting a resonant coupler and not a resonant cavity mode.
Resonant coupling to what?
Thus it is very likely little Rf energy entered the cavity and any that did could not excite a resonant mode.
RF energy which enters the cavity and does not excite a resonant mode would be reflected to the antenna and increase the input resistance, the reflected power would be increased. That’s what we observe outside resonance. When we are at resonance, the RF power goes into the cavity as described above.
I'm sure others here can take the EW cavity dimensions that Tajmar replicated, minus the EW dielectric and search for the resonant mode he excited at 1.865GHz?
Every COMSOL simulation uses ideal geometry and material properties. COMSOL Eigenfrequency simulations also do not respect the antenne position and orientation which also supports certain modes.
I hope that helps. Again, more to come – please wait for the full analysis.
Best regards,
Martin Tajmar.
Gesendet: Donnerstag, 24. Mai 2018 12:44
An: Tajmar, Martin
Betreff: Comment in Nasa Spaceflight
Dr. Tajmar -
I was interested to see a comment today in the NASA Spaceflight/EM Drive blog, which I've pasted in below.
I'm sure everyone there would be very grateful if you were able to respond. If I were to try and paraphrase the argument, it would be that in none of the tests did RF energy reach the frustrum, thereby explaining why the application of the choke had no effect. I was intruiged to note the suggestion that the Qfactor at that frequency was way to high, and note that Rodal and others had queried the Qfactor you noted in your earlier presentation.
I could pass on any response you have, if you don't want to comment directly on the blog.
Regards,
Forget your issues with TheTraveler, that is not the hot topic here.
My post wasn't about TT, it was about your bandwagon jumping reaction to his post.
Don't forget "There is no bad data", although it seems good data was interpreted badly here with Tajmar's team. And it seems everyone has jumped on his null tests without seriously evaluating his work.
Your claim "it seems good data was interpreted badly" doesn't have any
facts behind it, so you are not being objective.
We should have the right to question other builds and data where we see glaring errors.
You do have that right, but you haven't pointed out any glaring errors, and seem to have come tho the conclusion that the experiment must have been done wrong. Some difference in the frequency from a different build is not unexpected, and there isn't really a plausible alternative to the cavity being excited in a resonant mode for the VNA plot.
.......
A -55dB rtn loss from the S11 VNA scan represents a VSWR of 1.004:1 which is an impossible real world result.
You don't get to talk about "impossible real world result" when you are claiming conservation of momentum does not work.
You didn't actually respond to any of my points, anything other than cavity resonance producing those features on the plot is what is implausible.
You do have that right, but you haven't pointed out any glaring errors, and seem to have come tho the conclusion that the experiment must have been done wrong. Some difference in the frequency from a different build is not unexpected, and there isn't really a plausible alternative to the cavity being excited in a resonant mode for the VNA plot.
Claim made in Tajmar's paper:
"Considering the magnetic field strength of the Earth’s magnetic field of 48 μT with an inclination of 70° in middle Europe, a few centimeters of cables and a current of 2 A (similar to what is needed to power the amplifier), we obtain Lorentz forces of a few μN, which is similar to our observed “thrust” values."The image I posted clearly shows that there is probably over a meter of main power cables to the amplifier and then there is all the DC power that goes to the VNA, variable attenuator, stepper motor, etc. There is no way there is only a "few centimeters of cables."
That seems to be a pretty large glaring error. As pointed out earlier, the entire experiment box rotates, including the long wires. This was not made clear in the paper and doesn't make sense if we want to isolate the contribution of the frustum alone.
Then there is the Return Loss trace that shows a smaller dip very close to the main dip (which we have always been told to avoid). Without an infrared camera image of the frustum walls to show a specific mode being excited, the RF could have gone any number of places besides the cavity. This would be obvious by a hot spot in the IR camera where it shouldn't be. That is the huge draw-back to my 3D printed frustum, it is impossible to see mode with the IR camera since PLA is insulative. Luckily Ozyw sent me his solid copper frustum and I should be able to confirm mode there.
We should probably mention the makeshift aluminum foil gasket held in place with two clamps. Tajmar admitted that the frustum leaked RF as he stated that in his presentation (he was worried the European version of the FCC would coming knocking). That -55dB return loss could be most the RF leaking out for that given frequency. This should be quantified. I use an external antenna to monitor ambient RF from very near the frustum. I even completely rebuilt the small end twice because of unacceptable levels of RF leaking.
X_Ray has said that right angle connectors are known to have issues with reflections. It should be noted that Tajmar uses a right angle connector. I do not think anyone has used a right angle connector for the main RF input.
You do have that right, but you haven't pointed out any glaring errors, and seem to have come tho the conclusion that the experiment must have been done wrong. Some difference in the frequency from a different build is not unexpected, and there isn't really a plausible alternative to the cavity being excited in a resonant mode for the VNA plot.
Claim made in Tajmar's paper: "Considering the magnetic field strength of the Earth’s magnetic field of 48 μT with an inclination of 70° in middle Europe, a few centimeters of cables and a current of 2 A (similar to what is needed to power the amplifier), we obtain Lorentz forces of a few μN, which is similar to our observed “thrust” values."
The image I posted clearly shows that there is probably over a meter of main power cables to the amplifier and then there is all the DC power that goes to the VNA, variable attenuator, stepper motor, etc. There is no way there is only a "few centimeters of cables." That seems to be a pretty large glaring error.
Since increasing the length of cables would increase the expected magnetic force, and the point of what you are quoting is that magnetic field interactions are a plausible reason for the observed force, I fail to see how this is an error of any sort. Since they are using twisted leads it makes sense that the effective length producing magnetic force is less than the total cable length anyway. (And if that didn't makes sense the question would be why didn't they measure
more force.)
As pointed out earlier, the entire experiment box rotates, including the long wires. This was not made clear in the paper and doesn't make sense if we want to isolate the contribution of the frustum alone.
Maybe you need to read the paper again. I had no problem understanding that from the paper, and there is a reason they used the attenuator test.
Sure I am, when he stated that the test stand and frustum were a compass, implying that they had taken extraordinary efforts to eliminate error, there were still build errors as we pointed out. A few were... Running a magnetic stepper motor in close proximity to the frustum and wiring, improper wiring (twisted pairs), limited to 2 watts RF because of thermal mitigation issues. Questions in modes run or the strange high Q (ask Dr. Rodal about that).
At least on the NRL test stand they realized the issues with the thermal (and Lorentz) causing errors, I believe Tajmar could take a few ideas from their build to incorporate into the next try.
Did you even read the paper? Things listed in the paper as "we will run more tests after fixing these issues" are not things you can fairly criticize the paper for. You are making it sound like they aren't aware of these issues, and that they are not already planning to fix them.
Let me emphasize something from Tajmar's response above:
In general, I’m very surprised by this huge response. We posted preliminary results/work in progress, which was clearly stated in the paper. Work is ongoing, we will measure all the different frequencies/geometries/dielectric discs/higher power levels, etc. and hopefully publish the full analysis next year. So please be patient, more to come.
As he said, I found it clear in the paper that it was preliminary work with more to come, but it seems like some people (in particular emDrive supporters in this forum) keep missing that. He should get the same patience that the DIY builders in this forum are given.
X_Ray has said that right angle connectors are known to have issues with reflections. It should be noted that Tajmar uses a right angle connector. I do not think anyone has used a right angle connector for the main RF input.
They tracked the reflected power, so they can know just how much power is lost due to various sources of reflections.
You do have that right, but you haven't pointed out any glaring errors, and seem to have come tho the conclusion that the experiment must have been done wrong. Some difference in the frequency from a different build is not unexpected, and there isn't really a plausible alternative to the cavity being excited in a resonant mode for the VNA plot.
Claim made in Tajmar's paper: "Considering the magnetic field strength of the Earth’s magnetic field of 48 μT with an inclination of 70° in middle Europe, a few centimeters of cables and a current of 2 A (similar to what is needed to power the amplifier), we obtain Lorentz forces of a few μN, which is similar to our observed “thrust” values."
The image I posted clearly shows that there is probably over a meter of main power cables to the amplifier and then there is all the DC power that goes to the VNA, variable attenuator, stepper motor, etc. There is no way there is only a "few centimeters of cables." That seems to be a pretty large glaring error.
...
I think what he meant was (EVEN) a few centimeters of (unpaired) cable would cause problem large enough. At least this was my reaction the first time I read the statement a few days ago. It should not be regarded as a large glaring error, since most of his cables are paired/twisted; the net effect may be a just few centimeters of unpaired cable.
Plus where is the Smith Chart showing a nice round circle, crossing the centre line at 50 ohm impedance and showing equal amounts of L and C energy on both sides of resonance?
Either their VNA doesn't have that capability, they didn't have time, or they chose not to show it. With that second mode so close it will look something like this.
As he said, I found it clear in the paper that it was preliminary work with more to come, but it seems like some people (in particular emDrive supporters in this forum) keep missing that. He should get the same patience that the DIY builders in this forum are given.
This whole discussion was started because someone said these specific results were "case closed." And it seems that a number of media outlets have also run with the same story. We are merely pointing out the reasons we think this is unlikely. You are welcome to participate in that discussion, but let's discuss the experiment, not your feelings.
I welcome more tests from Tajmar, but it seems he has most of his work ahead of him.
As he said, I found it clear in the paper that it was preliminary work with more to come, but it seems like some people (in particular emDrive supporters in this forum) keep missing that. He should get the same patience that the DIY builders in this forum are given.
This whole discussion was started because someone said these specific results were "case closed." And it seems that a number of media outlets have also run with the same story. We are merely pointing out the reasons we think this is unlikely. You are welcome to participate in that discussion, but let's discuss the experiment, not your feelings.
I welcome more tests from Tajmar, but it seems he has most of his work ahead of him.
Someone saying "case closed" has nothing to do with the conversation I was involved in today. I quite clearly never said that.* I got involved when you and SeeShells jumped on TT's nonsensical explanation that the drive wasn't resonating. The difference between resonating and not resonating is important because it is a difference between the experiment just identifying what error sources still need to be eliminated, and the experiment setting an upper limit on emDrive thrust. The second one is what is happening here based on the presented evidence.
I have been talking about the experiment not my feelings. I said one sentence and then moved on to directly discuss either the experiment, or statements and reactions of others to the experiment. If I am welcome to discuss the experiment, why aren't you addressing the primary content of any of my posts? (Hint: this question is best answered indirectly by responding to my statements other statements.)
*There are some things that are "case closed" for the emDrive, such as most things related to Shawyer. Lets not bother getting into those right now, they aren't particularly relevant.
OK, one more comment more related to feelings rather than the experiment.
On the contrary, I enjoyed McCulloch's cartoon. He has a good sense of humor.
In general, that is a funny comic. In this context, it seems to be a direct insult to Tajmar (effectively baselessly claiming that he didn't build the drive right), which is not funny at all.
.......
A -55dB rtn loss from the S11 VNA scan represents a VSWR of 1.004:1 which is an impossible real world result.
Yet it is there.
One explication is it was caused by coupler self resonance, which ignores cavity eddy current losses.
BTW a -25dB S11 rtn loss is an excellent real world result, VSWR 1.2:1, when dealing with a high Q EmDrive.
Maybe ask Jamie what S11 rtn loss his cavity generates?
To me the simple answer, in the case of an S11 measurement is that each line crossing the center of the smith chart (almost exactly) will result in a huge return loss at that point. In the LogMag plot you can read how excact the condition Z=50Ω+j0Ω is satisfied for each frequency point. In theory the possible return loss can be infinite. In real world experiments it is finite because of several reasons but none of them is the Q of the cayity itself which is represented by the bandwidth not if there is any point on the curve where the impedance matches Z. You can have a low Q mode but with a proper couppling coefficient the return loss will be big.
The missing plot in the complex plane is a major problem here.
As first i would ask how well the calibration of the VNA was done, if there was a misstake the dB values say almost nothing.
As for the resonance dip in the plot i am with Jamie and his first thought taken from the analyzes of Frank Davies.
.......
A -55dB rtn loss from the S11 VNA scan represents a VSWR of 1.004:1 which is an impossible real world result.
Yet it is there.
One explication is it was caused by coupler self resonance, which ignores cavity eddy current losses.
BTW a -25dB S11 rtn loss is an excellent real world result, VSWR 1.2:1, when dealing with a high Q EmDrive.
Maybe ask Jamie what S11 rtn loss his cavity generates?
To me the simple answer is, in the case of an for a S11 measurement that each line crossing the center of the smith chart (almost exactly) will result in a huge return loss at that point. In the LogMag plot you can read how excact the Z=Z0=50Ω+j0Ω condition is present for each frequency point. In theory the possible return loss can be infinite. In real world experiments it is finite because of several reasons but none of them is the Q of the cayity itself which is represented by the bandwidth not if there is any point on the curve where the impedance matches Z0. You can have a low Q mode but with a proper couppling coefficient the return loss will be big.
The missing plot in the complex plane is a major problem here.
As first i would ask how well the calibration of the VNA was done, if there was a misstake the dB values say almost nothing.
As for the resonance dip in the plot i am with Jamie and his first thought taken from the analyzes of Frank Davies.
Xray,
I'm sure you are aware that couplers can have a self resonant freq. When excited at that freq the coupler will get very hot as its resistance is expending all the input energy as heat. The rtn loss value is so low because there is no energy coupled into the cavity and no eddy currents. So the VNA sees the resistance of the loop antenna, which at self resonance reflects very little Rf energy, instead getting very hot, well depending on the input power. A mini thermal load.
Tajmar needs a Rf engineer that will explain to him why a -55dB S11 rtn loss is saying "Danger Martin Tajmar, Danger".
.......
A -55dB rtn loss from the S11 VNA scan represents a VSWR of 1.004:1 which is an impossible real world result.
Yet it is there.
One explication is it was caused by coupler self resonance, which ignores cavity eddy current losses.
BTW a -25dB S11 rtn loss is an excellent real world result, VSWR 1.2:1, when dealing with a high Q EmDrive.
Maybe ask Jamie what S11 rtn loss his cavity generates?
To me the simple answer is, in the case of an for a S11 measurement that each line crossing the center of the smith chart (almost exactly) will result in a huge return loss at that point. In the LogMag plot you can read how excact the Z=Z0=50Ω+j0Ω condition is present for each frequency point. In theory the possible return loss can be infinite. In real world experiments it is finite because of several reasons but none of them is the Q of the cayity itself which is represented by the bandwidth not if there is any point on the curve where the impedance matches Z0. You can have a low Q mode but with a proper couppling coefficient the return loss will be big.
The missing plot in the complex plane is a major problem here.
As first i would ask how well the calibration of the VNA was done, if there was a misstake the dB values say almost nothing.
As for the resonance dip in the plot i am with Jamie and his first thought taken from the analyzes of Frank Davies.
Xray,
I'm sure you are aware that couplers can have a self resonant freq. When excited at that freq the coupler will get very hot as its resistance is expending all the input energy as heat. The rtn loss value is so low because there is no energy coupled into the cavity and no eddy currents. So the VNA sees the resistance of the loop antenna, which at self resonance reflects very little Rf energy, instead getting very hot, well depending on the input power. A mini thermal load.
Tajmar needs a Rf engineer that will explain to him why a -55dB S11 rtn loss is saying "Danger Martin Tajmar, Danger".
TT,
high return loss (large negative dB value) means the detector in your measurement device detects almost no reflected signal. The hf-power somewhere disapeers, this can resulting from destructive interference, radiation loss or of course resonances in the system with associated thermal losses, either of single parts but more likely because RLC values of each network component act together with the others nearby. So the resulting RLC values (of cable, feed antenna, connectors, couppler, tuner, cavity, ...) as seen by the detector then may equivalent to the system impedance of the measuring device at defined frequency points in the spectrum.
What you describe is at least possible if a additional high resonant structure is part of the network. A way to test your theory could be to open up the cavity resonator and measure again. If the dip is still present it may be what you say, if it is gone it is most likely a cavity resonance.
What you describe is at least possible if a additional high resonant structure is part of the network. A way to test your theory could be to open up the cavity resonator and measure again. If the dip is still present it may be what you say, if it is gone it is most likely a cavity resonance.
Opening up the cavity by say removing one end cap would make it basically a feedhorn antenna, which likely would effectively radiate near drive resonances. Better to replace the cavity with an open circuit, or otherwise a load known to be very mismatched. Might be what you meant, but I wanted to clarify.
What you describe is at least possible if a additional high resonant structure is part of the network. A way to test your theory could be to open up the cavity resonator and measure again. If the dip is still present it may be what you say, if it is gone it is most likely a cavity resonance.
Opening up the cavity by say removing one end cap would make it basically a feedhorn antenna, which likely would effectively radiate near drive resonances. Better to replace the cavity with an open circuit, or otherwise a load known to be very mismatched. Might be what you meant, but I wanted to clarify.
To find "in system" resonances all the way along the cable path i would agree with you using a high reflective part like short or open (load not really because it will decrese the Q of possible in system resonances too..)
TT assumed a antenna resonance, to find out if this really happens it may be an easy to do thing what i said. Most of the possible standing wave pattern** in the frustum itself will be not present then. Moveing a plate of the truncated frustum could be enough to explore if the resonant frequency of the marked dip at 1.88GHz shifts in frequency. If so it is most likely no antenna resonance but a resonant mode in the cavity.
**Modes with index TXmn0 will still present.
A poor man's way of achieving the same result without a thermal camera. It glows in response to heat. Should mirror the mode being excited. Paint it on the outside of the cavity.