-
#3560 by Rodal on 05 Jul, 2016 18:19
-
So locking the frustum in its own little Styrofoam box for a short time should eliminate all thermal convection thrust? Assuming you can safely keep it in there long enough to tell. Maybe some way for it to heat a small dish of water that is inside as a heat sink.
That or maybe holes where the air can exit and enter would give vertical thermal thrust, allowing air cooling, and eliminating horizontal air flow.
Certainly insulating the magnetron from the outside environment so as to at least further retard the heating of the air would be a good experiment to run (if one cannot run an experiment where the magnetron is kept from exciting the EM Drive by placing a plate inside the EM Drive).
Something should be done to quantify the effect of air heating.
The way the experiment is presently being run is just about the best way to maximize heating effects:
1) magnetron located at the big end of rfmwguy's EM Drive perpendicular to longitudinal axis of the torsional beam (instead of being on the side, along the longitudinal axis of the torsional beam)
2) magnetron fully exposed to the air -
#3561 by Rodal on 05 Jul, 2016 18:23
-
...
If you talk to Andrew tell him we miss him a lot and wish for him to come back too !
Just on a little vacation from work. Thought I'd drop in and see if anyone launched one off the ground yet.
Happy 4th!
-
#3562 by X_RaY on 05 Jul, 2016 19:37
-
Is there a raw data-list available for the public here for analyzing purposes?
The low pass seems a little overdosed, the decay seems to have a different sign in the blue chart after the magnetron is shutting down.
*The red line attached is drawn by hand, but it shows what I mean.
Compared to the orange it's a quite different pathway.
If data filters are used too extensive or in a inappropriate way one can generate things who aren't really there but it can look very impressive, such as an exponential decay from the top peak to zero.
It seems the pendulum deflects into the opposite direction after HF-power drops down.
Its a general human problem that people often tend to present their data in the way they like to see it.
-
#3563 by Tellmeagain on 05 Jul, 2016 20:37
-
Is there a raw data-list available for the public here for analyzing purposes?
The low pass seems a little overdosed, the decay seems to have a different sign in the blue chart after the magnetron is shutting down.
*The red line attached is drawn by hand, but it shows what I mean.
Compared to the orange it's a quite different pathway.
If data filters are used too extensive or in a inappropriate way one can generate things who aren't really there but it can look very impressive, such as an exponential decay from the top peak to zero.
It seems the pendulum deflects into the opposite direction after HF-power drops down.
Its a general human problem that people often tend
to present their data in the way they like to see it.
It seems the blue trace is the derivative of the read trace with respect to time, consistent with it being called "rate" by TheTraveller. However, the unit he presented, "mm/100", was obscure. I have learned to be accurate and precise at the first place in a forum post. Anything that might confuse people will definitely confuse some. -
#3564 by TheTraveller on 05 Jul, 2016 20:47
-
Is there a raw data-list available for the public here for analyzing purposes?
It includes my charting column additions but the original data columns are untouched.
You guys need to hang fire on the data analysis until Dave and a few other helpers sort out an anomaly.
-
#3565 by X_RaY on 05 Jul, 2016 20:58
-
Very enlightening, the blue one is effective an high pass filtered curve based on the upper one.Is there a raw data-list available for the public here for analyzing purposes?
It includes my charting column additions but the original data columns are untouched.
Thanks for this clarification
-
#3566 by TheTraveller on 05 Jul, 2016 21:02
-
Very enlightening, the blue one is effective an high pass filtered curve based on the upper one.Is there a raw data-list available for the public here for analyzing purposes?
Thanks for this clarification
The blue curve is the delta between the current tick's displacement minus the last tick's displacement, a measure of the amplitude and direction of change between ticks. Nothing fancy. -
#3567 by xyzzy on 05 Jul, 2016 21:34
-
Something should be done to quantify the effect of air heating.
From what I've been reading here for a while, the temperature rise of the magnetron itself could probably serve this purpose very well:
When the magnetron is cold, it can't lock (output is broadband).
When it's warm, it locks like a PLL (output becomes very narrowband and the cavity resonates).
When it heats up too much, it will unlock again (the cavity does not resonate anymore, output goes broadband again).
Based on this effect, Dave could probably do the following test:
1. Add a lock indication to the data. -> The indication does not have to be very precise in time. A second or two of uncertainty is no practical problem at all. A manual indication would be OK, like by pressing a button as soon as he sees the spectrum analyzer display "locking" and releasing the button again as soon as he sees it "unlocking", the signal form the button being recorded by his data acquisition system on an otherwise unused channel).
EDIT: Make sure that the connection is safe! An acquisition box is no safety isolation against high voltages that may come from other channels! Put a wireless remote control (as an isolated signal path) between the button and the acquisition box, so that no wire or conductive part whatsoever, that is connected to the acquisition box, ever needs to be handled and none can ever be touched!
2. Let the magnetron run long enough until it "unlocks" again from heat plus maybe a minute.
This way, after the magnetron heats up out of lock, any effective EM-related force should disappear instantly while the thermal effects should continue climbing (after all, the mag is still on and the whole thing is still heating up, far from thermal equilibrium).
By continuing to run the experiment after the magnetron unlocks (and recording that moment to a second or two precision) one should get a clear separation between the EM effects (suddenly disappearing) and the thermal effects (still continuing to rise unabated).
Of course, from an EM drive builder's point of view, this sounds a lot like "It's not a bug, it's a feature!", but from a scientific point of view, this unintended side effect (unlock from overheating) could likely give the missing key data point to differentiate EM from thermal effects.
-
#3568 by TheTraveller on 05 Jul, 2016 21:38
-
Something should be done to quantify the effect of air heating.
From what I've been reading here for a while, the temperature rise of the magnetron itself could probably serve this purpose very well:
When the magnetron is cold, it can't lock (output is broadband).
When it's warm, it locks like a PLL (output becomes very narrowband and the cavity resonates).
When it heats up too much, it will unlock again (the cavity does not resonate anymore, output goes broadband again).
Based on this effect, Dave could probably do the following test:
1. Add a lock indication to the data. -> The indication does not have to be very precise in time. A second or two of uncertainty is no practical problem at all. A manual indication would be OK, like by pressing a button as soon as he sees the spectrum analyzer display "locking" and releasing the button again as soon as he sees it "unlocking", the signal form the button being recorded by his data acquisition system on an otherwise unused channel).
2. Let the magnetron run long enough until it "unlocks" again from heat plus maybe a minute.
This way, after the magnetron heats up out of lock, any effective EM-related force should disappear instantly while the thermal effects should continue climbing (after all, the mag is still on and the whole thing is still heating up, far from thermal equilibrium).
By continuing to run the experiment after the magnetron unlocks (and recording that moment to a second or two precision) one should get a clear separation between the EM effects (suddenly disappearing) and the thermal effects (still continuing to rise unabated).
Of course, from an EM drive builder's point of view, this sounds a lot like "It's not a bug, it's a feature!", but from a scientific point of view, this unintended side effect (unlock from overheating) could likely give the missing key data point to differentiate EM from thermal effects.
That is a good idea as the locked maggie spectrum clearly paints the frustum bandwidth curve as attached.
Dave does have 2 spare input channels on his data logger, so should be doable.
-
#3569 by RotoSequence on 05 Jul, 2016 21:42
-
That is a good idea as the locked maggie spectrum clearly paints the frustum bandwidth curve as attached.
This would still leave all parties in the position of trying to extract what is assumed to be signal from an uncharacterized noise floor with uncharacterized thermal dynamics. I believe we need a true null test of the magnetron and test setup to know what the pure noise part of this experiment looks like. -
#3570 by Rodal on 05 Jul, 2016 21:44
-
That is a good idea as the locked maggie spectrum clearly paints the frustum bandwidth curve as attached.
This would still leave all parties in the position of trying to extract what is assumed to be signal from an uncharacterized noise floor with uncharacterized thermal dynamics. I believe we need a true null test of the magnetron and test setup to know what the pure noise part of this experiment looks like.
I completely agree, no doubt about it. -
#3571 by xyzzy on 05 Jul, 2016 21:58
-
Dave does have 2 spare input channels on his data logger, so should be doable.
But he would definitely have to check and double-check (!!! and triple-check !!!) that everything is safe before touching (or pressing) anything. IIRC, one of the channels, he has used to connect a high voltage probe. The data logger offers NO safety isolation at the kV levels involved whatsoever. Before connecting anything touchable (like a button), make really sure that no other part of the data acquisition system can become hazardous. Again, the acquisition box itself offers no protection!
And then, in addition, still keep the button behind such isolation (preferably a large air gap via a wireless remote control), as if assuming that the 4 kV were applied directly to the acquisition system.
Safety First!!! -
#3572 by xyzzy on 05 Jul, 2016 22:08
-
Make the lock status switchable via a wireless remote control. That way nothing connected to the acquisition box ever needs to be touched at all. Given all the risks (grounds can fail, high voltage probes can slip, ground leads can fall off - or even worse - fall off and accidentally hit a high voltage wire, etc.), a NO-TOUCH option would really be the ONLY SAFE option. Don't connect anything touchable anywhere near high voltage wiring! Safety First!
EDIT: changed the original post to mention safety. With the risks involved, isolation must really be the top priority. -
#3573 by TheTraveller on 05 Jul, 2016 23:35
-
Dave does have 2 spare input channels on his data logger, so should be doable.
But he would definitely have to check and double-check (!!! and triple-check !!!) that everything is safe before touching (or pressing) anything. IIRC, one of the channels, he has used to connect a high voltage probe. The data logger offers NO safety isolation at the kV levels involved whatsoever. Before connecting anything touchable (like a button), make really sure that no other part of the data acquisition system can become hazardous. Again, the acquisition box itself offers no protection!
And then, in addition, still keep the button behind such isolation (preferably a large air gap via a wireless remote control), as if assuming that the 4 kV were applied directly to the acquisition system.
Safety First!!!
Dave uses an industry standard 5kV 100:1 high voltage step down probe to feed the data logger & monitor the -4kV on channel 2 of his data logger.
Channel 1 is the Laser Displacement input
-
#3574 by otlski on 06 Jul, 2016 01:27
-
I think part of the characterization problem we are having is that many of us are expecting this to act like a pure torsion pendulum. For many reasons, I believe it is not. As pointed out by Rodal previously, we need to consider the torque required to twist the torsion rod (this is the pure model parameter).
We also have to know the mass moment of inertia and determine angular acceleration to calculate the torque that gets “used” to achieve angular acceleration and ultimately angular displacement over time. This T=Ia relationship provides information regarding the expected bandwidth response; perhaps more so than the torsion rod constant does. Frankly, I am not sure it’s significance, that’s why information is needed – just to get a sense of what is going on, and be backed notionally by maths.
The effect of the oil dampener will cause a modified fundamental frequency, unintentionally lengthening the oscillation period as well as providing the intended dampening. We should attempt to isolate this lengthening to properly model the system.
Based upon prior experiences, I can guess that air damping has a significant affect both in damping but especially in period lengthening. Unfortunately, given the DIY nature, this is harder to deal with (but ultimately not impossible).
And finally, the umbilical. This thing is a basket case of trouble. It impacts the system response in so many undesirable ways. It is a spring element that adds to the torsion wire’s spring rate and quickens the oscillation period. It has a non-linear spring rate whereas a well-executed torsion wire is around 0.03% linearity. It is a mechanical dampener whose characteristics are unreliable. It is a source of mechanical hysteresis (angular stiction). It is quasi point mass that affects the MOI with an mR^2 relationship, but does so in an unclear way because half the umbilical mass is suspended by the moving beam, and half by the rod holding the other end of the cable. It is a thermal heat accumulator (copper) surrounded by a heat insulator (cable insulation). It is potentially a thermal actuator. Notice how the spring element aspect shortens the period, many of the others likely lengthen the period.
Understanding what the period should be, and what the additive elements provide will help us determine what any magnitude of force is, and over what time duration should cause what displacement response.
Some basic info to complete the picture.
What is the diameter of the torsion wire? Need to know to 0.001” because the torsion rod constant is a fourth power relationship.
What is the length of the torsion wire? With the diameter and length, we can calculate the torsion constant in torque units per radian.
What is the natural oscillation period without the umbilical and oil damper? This will tell us the MOI of the beam assembly. This in turn will tell us the linear portion of the expected system response. Show plot of known input verses displacement response over time.
What is the beam assembly’s all up mass and length. Indirectly we can use these to estimate a radius of gyration to see if it sensibly matches the calculated MOI.
What is the oscillation period with the oil dampener in place? Show plot of known input verses displacement response over time.
What is the oscillation period of the system with the umbilical in place but unpowered? ? Show plot of known input verses displacement response over time. Keeping in the DIY’ers spirit, I get that doing away with the umbilical is one of the next steps; after proof of concept has been suitably demonstrated.
Once we quantitatively understand each of our DIY’ers systems, we can better analyze thermal and other anomalies in order to accept or dismiss their relevance.
-
#3575 by masterharper1082 on 06 Jul, 2016 01:59
-
Sort of... a high-pass filter should have unity gain for high frequencies, and attenuate at lower frequencies. A derivative has constantly increasing gain as frequency goes up. That's why there is so much noise in the signal. Later on, it might be worth trying more of a Kalman filter or observer structure to get a cleaner velocity estimate, and perhaps even an acceleration estimate.
Very enlightening, the blue one is effective an high pass filtered curve based on the upper one.Is there a raw data-list available for the public here for analyzing purposes?
It includes my charting column additions but the original data columns are untouched.
Thanks for this clarification -
#3576 by frobnicat on 06 Jul, 2016 10:58
-
Follow-up on experimental characterization of horizontal pendulums : it's not trivial to make a calibration system able to deliver a clean square off/on/off forcing function (like the electrostatic actuator used at EagleWorks) to compare the allure of result from such forcing function with what the powered frustum generates. I suggested a contactless actuator based on electromagnet (fixed relative to lab) pulling on small mass of iron (fixed on pendulum).
On the other hand it seems relatively straightforward to see what the decay (pendulum restoring to rest position) from a constant forcing function dropping suddenly to 0 (on/off) should look like : have a beam of wood or metal sit on its end on ground and pushing on the pendulum (or on the frustum) to a given shift typical of displacement with em-drive runs. Let the system settle, and then pull a thin light thread from a few feet away to make the beam topple, hence releasing the pendulum from a shifted position to record its trajectory back to equilibrium. See attached illustrations. I think that would allow for a clear assessment of the time constant and shape expected from a down to 0 excitation force, and whether or not (and how much) it looks like what is observed at emdrive power off.
Intuitively from looking at the plots from rfmwguy I would say (as other commentators here) that at least 80% of the forces that drove the pendulum arm away from its rest position are thermal in nature because of the shape and length of the decay back to rest, i.e. after power off the restoring torque of the suspension wire is not the only force involved, the force(s) that drove away the pendulum in the first place are still active and slowly decaying, and are thermal in nature. My suggested low cost experiment could confirm or infirm this impression, not perfectly but more precisely.
Bottom drawing : top figure as seen from the left
-
#3577 by Gilbertdrive on 06 Jul, 2016 13:03
-
Do you know if Dave already tried with the Polymer insert ?
As it was told earlier, if he gets different results with the insert, since the thermal and Lorentz should remain the same, it sould be a strong argument for a real anomalous force, and would help to give a lower bound to the anomalous force.
for example, if the mesured force becomes 5 mN higher with the polymer insert, we can reasonably think that the anomalous part of the force is, at least, 5 mN. Of course it does not replace vacuum and battery powered test, it should not be a definitive proof, but at least it should be a strong clue.
At the opposite, if there is no significant difference with the insert installed, we can not conclude, and, in that case, Dave should probably think to a null device. -
#3578 by SteveD on 06 Jul, 2016 16:09
-
Here are the Force calculations from each of the 3 test runs in Dave's latest data.
Very consistent 15mN, which suggest Dave could leave the maggie power on for a longer time and achieve a higher force result?
I'm afraid, since run 3 was a null, this chart shows that the vast majority of any reported effect must be thermal in nature. Since the magnetron did not fire during test 3, that means we are dealing with a cause other than heat from the magnetron. Given all the data to date, thermal expansion of the wires leading to the test apparatus would be the most likely candidate. Before writing the entire thing off (1.5 grams of force seems a bit much for heated wires) I would suggest testing to see if the effect isn't simply caused by heating a magnet feeding into the can with the various rf issues being largely irrelevant. -
#3579 by Rodal on 06 Jul, 2016 16:24
-
Here are the Force calculations from each of the 3 test runs in Dave's latest data.
Very consistent 15mN, which suggest Dave could leave the maggie power on for a longer time and achieve a higher force result?
I'm afraid, since run 3 was a null, this chart shows that the vast majority of any reported effect must be thermal in nature. Since the magnetron did not fire during test 3, that means we are dealing with a cause other than heat from the magnetron. Given all the data to date, thermal expansion of the wires leading to the test apparatus would be the most likely candidate. Before writing the entire thing off (1.5 grams of force seems a bit much for heated wires) I would suggest testing to see if the effect isn't simply caused by heating a magnet feeding into the can with the various rf issues being largely irrelevant.
<< Since the magnetron did not fire during test 3>>
This is all news to me !
I didn't see that information in TheTraveller's post as he posted the data. On the contrary, the communication was that this was a demonstration of Shaywer's invention: the EM Drive, working as it should.
On the contrary, TheTraveller posted:Here are the Force calculations from each of the 3 test runs in Dave's latest data.
Very consistent 15mN, which suggest Dave could leave the maggie power on for a longer time and achieve a higher force result?
Which did not make any sense to me:Here are the Force calculations from each of the 3 test runs in Dave's latest data.
Very consistent 15mN, which suggest Dave could leave the maggie power on for a longer time and achieve a higher force result?
Isn't that another argument (besides the fact that there is an exponential fast rise and a slow exponential decay) that this force is thermal in nature?
That the longer he heats the more displacement he gets?
Wasn't the EM Drive supposed to give a constant force for a constant power input?
Isn't his power input a constant during the test ? 900 watts?
Can you please explain on what basis are you stating that the magnetron did not fire on test 3
I have not seen any such information from the people that have been posting rfmwguy's results here (TheTraveller and SeeShells, to my recollection)
An explanation of what is going on with the data presented by rfmwguy would be appreciated ...
