Author Topic: EM Drive Developments - related to space flight applications - Thread 2  (Read 3321534 times)

Offline frobnicat

  • Full Member
  • ****
  • Posts: 518
  • Liked: 500
  • Likes Given: 151
...

I don't think the tilt of the balance beam about the X axis would have to be very much for the resulting change in received light amplitude to register as a shift in position.   The LDS has fractional micron resolution.    Assuming the light pattern from the fiber optic cable has a circular gaussian distribution, for any given distance the maximum light level hitting the detector is when the plane of the mirror is perpendicular to the central axis of the light beam.   Any small deviation ( arc-second) will reduce the light amplitude; which registers as a displacement.     It is impossible to align the LDS perfectly so the angle between the mirror and the light beam is never exactly 90 degrees in X and Y.    The expansion of the cavity due to heating has a very long time constant.   The slow drift in the position between RF pulses looks like a thermal response.

Interesting hypothesis. You say that in this case the LDS readings would be more sensitive to a change in angle of the reflective mirror than a change in distance. The LDS is a small distance below the X axis, so a small angular deviation around X would amount to tiny displacement (d' - d) but the small angular deviation would itself tilt the mirror and that could change the LDS readings that is roughly proportional to the reflected light.

But then this coupling between angular effect and displacement would also be present for the main movements around Z. LDS is at 35cm from Z axis, I would say it is one order of magnitude more than the distance of mirror below X axis, the lever effect of displacement/angle ratio would be lower (less angle for same displacement) but still if angle deviation is to have a significant effect in case around X it should also have a significant effect around Z.

Unfortunately the documentation for the Philtec D63 displacement sensor seems to give no information on reflective surface angular deviation impact on measurements. It indicates Ø 1.6 mm Target Spot Size. This is 3 times more than nominal operating distance of 500µm (0.5mm) so your drawing is a bit misleading with proportions. In reality the reflected dot is big relative to distance from reflector, would an angular deviation still impact the amount of light swallowed back ?

Is it you drawing or you found it elsewhere ? I ask the question because you mention the reflecting surface as being mirror. It does appear as a mirror in the pictures of pendulum at Eagleworks and this is stated explicitly in Brady's report page 3 (together with that LDS is photon time of flight, which is not, but let's proceed). But I don't find in the D63 documentation that a mirror should or could be used. It does say that These specifications represent best case performance where: the target is flat, smooth and highly reflective
This implies that the target may not be highly reflective (at the cost of decreased performance) but does highly reflective imply that a mirror is the best case ? My reading (but I'm not native English reader) is that reflective can mean a matte white material, that is a surface that does bounce near 100% of incoming photons, but in a more or less scattering manner, not necessarily as per perfect reflection. See attached picture : top the situation at hand at Eagleworks, bottom the default situation for which the LDS might be initially calibrated.

This could explain a lot about the disparity between vertical readings in the charts and stiffness parameters : when charts indicate ~1µm LDS deviation for 29.1µnN calibration pulses for instance, all could be much better explained and consistent if it meant 10µm instead.

To me there is no doubt that the LDS is operating in the far range : the cal. pulses are attractive by nature, and correspond to a displacement to the right (Y+) and that lowers the measured distance, as in the charts. Also operating in the near range would not only reverse this consistent orientation but would also increase the sensitivity : that would go against what is to be explained (apparent higher µm/µN stiffness than compatible with other known parameters, for 2 separate predictions from the dynamic periods of oscillations and the known stiffness of flexure bearings)

So, what anyone wanting to understand clearly the charts need is a way to explain a reduced sensitivity of the LDS, that would show 1µm displacement when in fact it is 10µm. Following on the D63 doc, there is a calibration procedure :
The effect of changing target reflectance is to shift the voltage output higher or lower. Factory calibrations have
the Peak Voltage set to 5.000 volts. A gain control is provided for calibration of the sensor output to various target surfaces. In-situ calibration is performed simply, by adjusting the sensor’s tip-to-target gap until the peak output voltage is attained, and then by using the gain control to set the peak voltage to full scale (5.000 volts). After setting the peak to 5 volts, the factory gap calibration chart applies for the target being measured. This procedure allows the sensor to be used to perform precision linear motion measurements on most materials.


I read "most materials : including smooth but scattering surfaces", not only mirrors. Actually, mirrors may be a very special case. My question is now if the gain control of the system is consistent with the use of a mirror reflecting target. As seen qualitatively from attached figure : a perfect mirror would bounce back in the signal detector a lot more photons than a matte white. If the gain is calibrated against matte material (say, as hypothesis, default factory settings) the baseline distance from the mirror to have a tension that shows as 500µm would be much greater than 500µm.

The protocol of tuning for nominal 500µm distance between optical fibre head and mirror is on page 3 of Brady et al report (anomalous...) :
Quote from: Brady et al. Anomalous thrust...
Prior to a test run data take, the LDS is positioned to a known displacement datum (usually 500 micrometers) via mechanical adjustments to its mounting platform. Gross adjustments are performed via set screws. Fine adjustments are performed using manually - operated calibrated screw mechanisms and a remotely controlled motorized mechanism that can be operated with the chamber door closed and the chamber at vacuum. The remote adjustment capability is necessary since the LDS datum will change whenever a change to the test facility environment affects the roll - out table or the chamber – e.g., whenever the chamber door is closed or latched and whenever the chamber is evacuated. Once the LDS displacement is adjusted in the final test environment, further adjustment between test run data takes is usually not required.

How this displacement of 500µm is known ? Has the procedure followed the step (as per D63 documentation) of homing to the distance corresponding to peak signal (whatever it's absolute level), calibrating the gain so that at this peak the signal is now 5V, and then stepping back until meeting ~4.3 V corresponding to 500µm nominal distance ? If this was not followed, or if the D63 is not mean to be operated with perfect mirror target (doubtful it wouldn't be mentioned in documentation, but who knows ?), it is possible the LDS is operating in a far range that has a much lower sensitivity (and linearity) than the nominal -2.7 mv/µm.

Edit : BTW, how comes that at some very near distance there start to be less photons bouncing back in the signal detector ? Thank zen-in for bringing up interesting questions about LDS...
« Last Edit: 03/06/2015 02:17 pm by frobnicat »

Offline DIYFAN

  • Member
  • Posts: 48
  • Liked: 28
  • Likes Given: 149
Same here.  The EBay units I've got are #201065780928 and #131442703325 so far in case anyone want to try the same system.

I've decided to try and live up to my screen name and throw my lot in with the replicators.  I have a few questions before I kick off my effort:

1) Why did you decide to go with #201065780928 rather than a frustrum of a cone?  The #201065780928 part looks more like a rectangular slit shape rather than a cone shape.
2) How do you plan on hooking the MA86751B X band oscillator up to the waveguide assembly?
3) It looks like the MA86751B X band oscillator is tunable from 9.9 GHz to 10.6 GHz with power output levels from 10 mW to 100 mW powered by 9-10 Volts DC.  Did you pick this particular oscillator for a reason?
4) Do you expect there to be a resonant frequency within the 9.9 GHz to 10.6 GHz frequency band?
5) Do you plan on putting a dielectric toward one end of the waveguide assembly?
6) With what material do you plan on capping each end of the waveguide assembly?

Although I direct this friendly set of questions to Notsosureofit, Mulletron, and others who are attempting replications, I welcome any forum members to chime in with recommendations or insights.  Can you imagine what would happen if the effect can be shown on such a small scale?  It could cause some ripples and raise a quite a few eyebrows around the world.

1.  it matches the osc. waveguide and 1st cheap one on EBay.
2.  just bolt them together
3.  I had this one already
4.  probably,  have to calculate when I get the waveguide
5. that's an option
6. TBD  flat pate, detector fitting etc etc

I'll just try a long ( 20 ft ?) pendulum First and see what happens.   Vacuum later.

Okay, thanks for those answers/thoughts.  Following up on 2 and 6, given that they just bolt together, doesn't that mean that the oscillator occupies one of the ends of the cavity?  It may sound like an obvious question, but I wonder how this would impact the potential effect when other models and equations seem to rely on there being two flat plates on each end of the cavity to reflect the waves.  I'm writing this without the luxury of having the MA86751B X band oscillator in hand, so I don't know the shape or material of the underside (which is not shown in the picture posted on ebay and I can't seem to find other pictures of it online).  Maybe it is just time to start ordering parts and see what it all looks like.   :D

Edit: what do you think of something like this, which would allow a flat plate to be placed on either end?

http://www.ebay.com/itm/WAVEGUIDE-WR-90-T-/131444026986?pt=LH_DefaultDomain_0&hash=item1e9aacae6a
« Last Edit: 03/06/2015 03:25 pm by DIYFAN »

Online Notsosureofit

  • Full Member
  • ****
  • Posts: 691
  • Liked: 747
  • Likes Given: 1729
Hopefully the whole thing acts as a cavity oscillator.  The intrusion acts as the "antennas". (my old cavity oscillator had a "lighthouse" tube stuck into it.  Wish I could find that thing)

Don't know what to do w/ a "Tee" ?

While the 1/f^3 factor is in there, the geometry of the smaller cavity at higher frequency should help.

The power is only ~50mW though.


Offline frobnicat

  • Full Member
  • ****
  • Posts: 518
  • Liked: 500
  • Likes Given: 151
http://www.philtec.com/downloadssupport/documentlibrary/documents/philtecmanuals/newanalogD.pdf
This answers some of the questions I just asked about LDS, reflectivity, calibration...  :D

Yes the D63 can measure displacements from matte surfaces (even black rubber !), it can also work with metal mirror polished surfaces. The later (mirror vs matte) having the higher reflectance, that is the higher signals, to be compensated by proper gain adjustment. The interpretation charts µm in function of analog volts output depends on the gain adjustment 5V at peak reflexion (between "near side" and "far side").

If expecting to see ~4.3V corresponding to 500µV with a correct gain (black curve and magenta tangent) then a bad gain much too high would get us more to the right on the red curve (believing we are measuring 500µm when actually it is more than 1mm) but this wouldn't give a lower sensitivity (slope actually higher). To get a slope lower, a gain smaller than the ideal one (result in green) would bring the functioning point (believed to be 500µm) closer to the peak. That would lower sensitivity and could explain a huge discrepancy between vertical readings and known actual stiffness bounds.

This could be easily checked by looking for the actual raw peak analog tension output, when slowly closing the gap the max should be 5V. And when inputing ~4.3V to the display it should show ~500.000 on vertical scale. If both hold, then I'm in mental confusion again.

Offline aero

  • Senior Member
  • *****
  • Posts: 3629
  • 92129
  • Liked: 1145
  • Likes Given: 360
I just finished running a meep case this morning which gave results that I don't understand at all. The detected forces were so large as to indicate an error in the model set-up. Probably a resolution issue again with the detectors within the same cells as the structure. But I can handle that.

What was weird were the field patterns. The cylindrical cavity appeared to be in perfect resonance with the energy range of the complex fields being about 3 times higher than the energy of the real fields. But the complex fields lagged the real fields by at least a sixth of a cycle. That is about 60 degrees.

Any guesses as to what that might mean and could it relate to the larger forces detected?
Retired, working interesting problems

Offline frobnicat

  • Full Member
  • ****
  • Posts: 518
  • Liked: 500
  • Likes Given: 151
...

Frobnicat:

Your analysis and comments made me go back and look at the current alignment of the Eagleworks torque pendulum and the attached picture indicates that my recollection of the leveling of the bottom beam of the TP being a quarter bubble high at the vacuum chamber door end was a bit exaggerated due to parallax error and a just plain bad memory.  It looks now like less than tenth of bubble low at the door end of the vacuum chamber, but part of that apparent tilt may in reality be due to actual bending of the 1.50 inch square aluminum beam.

Best, Paul M.

The X+ (test article) beam end appears lower at Height2= 3.97'' than toward X- (under electronic stack) at Height1= 4.00, as if the assembly is tilted downward toward test article. This appears in contradiction to the previous hypothesis that the electronic stack was lower, and the bubble up orientation to the right if the top image is oriented with the test article (X+) on the right, ie has same orientation as bottom pictures (hard to tell). Also a tilt with downward slope toward test article would lower the stability of the system as the global centre of mass of rotating assembly is toward the backward electronic stack relative to axis (~6cm behind using your latest data on positions and weights). I will try to incorporate those new data points in the model.

Your answers to questions about factual aspects of the ongoing experiments are greatly appreciated here on this thread. I understand you wouldn't want or have time to respond to all interrogations about possible models, but to proceed with my attempts at quantifying thermal expansions impact on sustained displacement signals, I really need your point of view on the immediately appearing contradictions of stiffness around Z :

A calibration pulse of 29.1µN / 4.45 = 6.54µLbf applied at 11.25'' gives a torque of 6.54*11.25 = 73.6 µLbf-In.
Two 0.007 Lb-In/° flexure bearing in tandem give 0.014 Lb-In/°. At rest equilibrium the angular displacement of the arm to equilibrate the cal. pulse torque should be  73.6 * 10^-6 / 0.014 = 5.26 * 10^-3 ° = 9.17 * 10^-5 rad. This would correspond to a displacement at the LDS of 9.17*10^-5*0.352 = 32.3 µm (LDS distance from Z is 13.88''=.352m)

In most charts, the vertical scale reads between 1µm to 2.5µm deviation in response to 29.1µN pulses. In one chart we even have 5µm. See attached pictures. Why is there an apparent disparity from chart to chart ? Why do we have an order of magnitude less actual stiffness (from the flexure bearings, but that would agree also with oscillation periods) than apparent stiffness (apparent LDS readings relative to µN cal. pulses). ? Could it be that the vertical scale numbers are not correct and should read ~10µm when saying 1µm ?

My previous post makes an attempt at a possible explanation (output gain of LDS...) for a 10 times factor on the vertical scale.

The disparity from chart to chart could then be explained by a varying Z tilt angle theta and/or CoM's of rotating assembly position D behind Z axis (by adding or removing small weights). Gravitational effects around a Z tilted pendulum would be equivalent to an horizontal pendulum with added stiffness k' = D M g sin(theta). This component could change enough (with differing test articles...) to explain the disparity between charts. But if the vertical scale remain as is, then it would imply variations of tilts of more than 10° !

Many thanks.
« Last Edit: 03/06/2015 06:10 pm by frobnicat »

Online Notsosureofit

  • Full Member
  • ****
  • Posts: 691
  • Liked: 747
  • Likes Given: 1729
@ RODAL

Still struggling as to how to get the exact equation into the form;

f^2 = k^2 + [asymmetric terms(x)] such that, (f(0)^2 - f(L)^2) = [asymmetric terms(0)] - [asymmetric terms(L)]

That is, how to evaluate at constant k.   Any ideas ?

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5911
  • USA
  • Liked: 6124
  • Likes Given: 5564
@ RODAL

Still struggling as to how to get the exact equation into the form;

f^2 = k^2 + [asymmetric terms(x)] such that, (f(0)^2 - f(L)^2) = [asymmetric terms(0)] - [asymmetric terms(L)]

That is, how to evaluate at constant k.   Any ideas ?
The force derivation for a truncated cone seems to have been of a physical, intuitive character (if not so, please correct me).  From a physical standpoint, one can observe the following:

1) My understanding of your prior derivation is that it modeled the truncated cone as being traveled by plane waves (*), just like a plane wave travels a cylindrical cavity from one end to the other.  The difference is that in the truncated cone, the plane wave continuously, gradually experiences a lower natural frequency as it travels from the small diameter (cross-section with the highest natural frequency) end to the big diameter end (cross-section with the lowest natural frequency).  It should also continuously, gradually experience attenuation (loss in intensity) as it travels from the small end to the big end.'

___________
(*) Note: truncated cones have spherical waves traveling inside them, rather than plane waves.  So the model assumes a very large radius of curvature, such that the spherical wave is approximately flat.

CONICAL PIPES (like the truncated cone: continuous gradual change in diameter)



2) The behavior in a cylindrical pipe containing two dielectric mediums is fundamentally different: instead of the continuous, gradual change in natural frequency of the truncated cone, the  cylindrical pipe containing two dielectric mediums experiences a discontinuous, step-change, a brutal change, going from medium 1 to dielectric medium 2.  This would be more like a pipe experiencing a discontinuous step-change in diameter all of a sudden, causing a sudden change in natural frequency and a sudden change in attenuation.


STEP-PIPE (the geometrical analog of the discontinuous step-change produced by a pipe having two dielectric mediums)


« Last Edit: 03/06/2015 10:16 pm by Rodal »

Offline Stormbringer

  • Full Member
  • ****
  • Posts: 1340
  • Liked: 239
  • Likes Given: 92
I know its a stretch thinking this fits in here but you guys have to see this:  http://www.sciencedaily.com/releases/2015/03/150306091617.htm

because QM and GRT reconciled? and gravitons. mustn't forget gravitons.
When antigravity is outlawed only outlaws will have antigravity.

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5911
  • USA
  • Liked: 6124
  • Likes Given: 5564
@ RODAL

Still struggling as to how to get the exact equation into the form;

f^2 = k^2 + [asymmetric terms(x)] such that, (f(0)^2 - f(L)^2) = [asymmetric terms(0)] - [asymmetric terms(L)]

That is, how to evaluate at constant k.   Any ideas ?
The force derivation for a truncated cone seems to have been of a physical, intuitive character (if not so, please correct me).  From a physical standpoint, one can observe the following:

1) My understanding of your prior derivation is that it modeled the truncated cone as being traveled by plane waves (*), just like a plane wave travels a cylindrical cavity from one end to the other.  The difference is that in the truncated cone, the plane wave continuously, gradually experiences a lower natural frequency as it travels from the small diameter (cross-section with the highest natural frequency) end to the big diameter end (cross-section with the lowest natural frequency).  It should also continuously, gradually experience attenuation (loss in intensity) as it travels from the small end to the big end.'

___________
(*) Note: truncated cones have spherical waves traveling inside them, rather than plane waves.  So the model assumes a very large radius of curvature, such that the spherical wave is approximately flat.

CONICAL PIPES (like the truncated cone: continuous gradual change in diameter)



2) The behavior in a cylindrical pipe containing two dielectric mediums is fundamentally different: instead of the continuous, gradual change in natural frequency of the truncated cone, the  cylindrical pipe containing two dielectric mediums experiences a discontinuous, step-change, a brutal change, going from medium 1 to dielectric medium 2.  This would be more like a pipe experiencing a discontinuous step-change in diameter all of a sudden, causing a sudden change in natural frequency and a sudden change in attenuation.


STEP-PIPE (the geometrical analog of the discontinuous step-change produced by a pipe having two dielectric mediums)


@Mulletron and @Star-Drive advocate dielectrics inside these microwave cavity resonators.

So far these dielectrics (I assume for expediency) have had uniform dielectric properties.

Thinking along the lines of the discussion above, one could think of whether a functionally-graded dielectric, with a continuous change in dielectric properties along the longitudinal axis, may provide superior functionality than a dielectric with uniform properties.

Although a functionally-graded dielectric will be more expensive to procure, this could be tested by approximation, by using several coupled dielectric material segments with increasing dielectric constants, their total (additive) length equaling the length of the single dielectric used so far with uniform properties.
« Last Edit: 03/06/2015 10:28 pm by Rodal »

Offline RotoSequence

  • Senior Member
  • *****
  • Posts: 2208
  • Liked: 2068
  • Likes Given: 1535
I know its a stretch thinking this fits in here but you guys have to see this:  http://www.sciencedaily.com/releases/2015/03/150306091617.htm

because QM and GRT reconciled? and gravitons. mustn't forget gravitons.

I found the full paper online, if anyone wants to take a look: https://www.academia.edu/8604226/The_Schwarzschild_Solution_to_the_Nexus_Graviton_Field

Online Notsosureofit

  • Full Member
  • ****
  • Posts: 691
  • Liked: 747
  • Likes Given: 1729

2) The behavior in a cylindrical pipe containing two dielectric mediums is fundamentally different: instead of the continuous, gradual change in natural frequency of the truncated cone, the  cylindrical pipe containing two dielectric mediums experiences a discontinuous, step-change, a brutal change, going from medium 1 to dielectric medium 2.  This would be more like a pipe experiencing a discontinuous step-change in diameter all of a sudden, causing a sudden change in natural frequency and a sudden change in attenuation.



Yes, the case of a gradually changing dielectric constant is very similar to that of the tapered cavity.

The step-change involves a much more complicated situation w/ the necessity of including the reflected waves from the discontinuity.  But, it may well offer enhanced performance. (?)  These types of calculations have been done w/ iterative matrix math.  I'm going to try some ideas w/ programs I have to see if I can get a better feel for this case.  (1D calculations at least)

Added: 1D modes suggest that we only have to consider terms which contain b  (ie. X or X')  The others are plain waves and cancel.

 
« Last Edit: 03/07/2015 02:25 am by Notsosureofit »

Offline zen-in

  • Full Member
  • ****
  • Posts: 541
  • California
  • Liked: 483
  • Likes Given: 371




Interesting hypothesis. You say that in this case the LDS readings would be more sensitive to a change in angle of the reflective mirror than a change in distance. The LDS is a small distance below the X axis, so a small angular deviation around X would amount to tiny displacement (d' - d) but the small angular deviation would itself tilt the mirror and that could change the LDS readings that is roughly proportional to the reflected light.

But then this coupling between angular effect and displacement would also be present for the main movements around Z. LDS is at 35cm from Z axis, I would say it is one order of magnitude more than the distance of mirror below X axis, the lever effect of displacement/angle ratio would be lower (less angle for same displacement) but still if angle deviation is to have a significant effect in case around X it should also have a significant effect around Z.

I read "most materials : including smooth but scattering surfaces", not only mirrors. Actually, mirrors may be a very special case. My question is now if the gain control of the system is consistent with the use of a mirror reflecting target. As seen qualitatively from attached figure : a perfect mirror would bounce back in the signal detector a lot more photons than a matte white. If the gain is calibrated against matte material (say, as hypothesis, default factory settings) the baseline distance from the mirror to have a tension that shows as 500µm would be much greater than 500µm.


Edit : BTW, how comes that at some very near distance there start to be less photons bouncing back in the signal detector ? Thank zen-in for bringing up interesting questions about LDS...

Here is a revised drawing that better illustrates how I think the LDS works.   The distance from the end of the LDS to the mirror would actually be much less but the proportion of the beam width at the mirror to distance is similar.  The light from the optical fiber, (green) has a higher dispersion near the target than a laser beam would.   The reflected pattern is equally dispersive, which has the effect of magnifying the reflected dot.  So in effect the receiver is sampling just a small area of the dot.   Small angular changes will produce a variation in amplitude of the received light if the dot pattern is gaussian because of this magnifying effect.   So just a few arc-seconds of rotation will register as a displacement even though the actual distance has not changed.

I think the reason why the LDS output has an opposite slope at close distances (likely < 10 µm) is because the two surfaces start acting like a waveguide when they are in close proximity.   Beamsplitters are made using this principle.
« Last Edit: 03/07/2015 04:35 am by zen-in »

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5911
  • USA
  • Liked: 6124
  • Likes Given: 5564

2) The behavior in a cylindrical pipe containing two dielectric mediums is fundamentally different: instead of the continuous, gradual change in natural frequency of the truncated cone, the  cylindrical pipe containing two dielectric mediums experiences a discontinuous, step-change, a brutal change, going from medium 1 to dielectric medium 2.  This would be more like a pipe experiencing a discontinuous step-change in diameter all of a sudden, causing a sudden change in natural frequency and a sudden change in attenuation.



Yes, the case of a gradually changing dielectric constant is very similar to that of the tapered cavity.

The step-change involves a much more complicated situation w/ the necessity of including the reflected waves from the discontinuity.  But, it may well offer enhanced performance. (?) ....

If one argues that the truncated cone has enhanced performance over a cylindrical cavity containing two adjoining coupled dielectrics, then it can be argued that the dielectrics should be functionally graded such that the dielectric constant is a continuous function of the axial length, and the dielectric constant changes gradually in that direction. Or, as an approximation, to have short segments of dielectric materials with gradually increasing dielectric constant in the axial direction.





Conversely, if the discontinuous jump in dielectric constant encountered in a cavity  containing two adjoining coupled dielectrics offers enhanced performance over the gradual geometrical change of a truncated cone, then it can be argued that the optimum geometrical shape would be a step-change in geometry, going from a small diameter to suddenly, discontinuously, to a larger diameter, for the same reason that it is argued that the discontinuous change in dielectric constant offers enhanced performance.






Something like this is accomplished in the Cannae cavity design (practically a sudden change in geometry, going from a small diameter to a large diameter).  However, the dielectric in the Cannae cavity was located at the wrong location: with the dielectric, for example located at the right end of the cylindrical section:

CANNAE:

Tested design: natural frequency reaches a maximum between the ends:

Large Diameter (lower natural frequency) --> Small Diameter (higher natural frequency) --> Small Diameter & Dielectric (lower natural frequency)



It would have made more sense to have the dielectric at the large diameter end, in order to have this:

Natural frequency increases from left end to right end:

Large Diameter & Dielectric (lowest natural frequency)-->Large Diameter (intermediate natural frequency) --> Small Diameter (highest natural frequency) 


« Last Edit: 03/07/2015 01:20 pm by Rodal »

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5911
  • USA
  • Liked: 6124
  • Likes Given: 5564
...maybe he envisions this device as some sort of 'turbo-charger?'   That is something that adds to an already existing velocity, but won't function well, if at all when 'at rest.'  ...
A turbocharger, is a device that increases an engine's efficiency and power by forcing extra air into the combustion chamber. 

What is the "engine" that the EM Drive is turbocharging?

The bizarre nature of something needing to be free to accelerate for it to produce a force doesn't apply to the turbocharger or to the engine:  the engine that is being turbocharged does not need to be accelerating or even be in rigid body motion. Its center of mass can be completely stationary, and the turbocharged engine can then be used for electric power generation, for example, instead of for transporting people as in an automobile.

Of course, there are many kinds of EM-Drives:

here is an older type of EM-Drive    :)     :

Notice that "The strings of the guitar need to be moving for the EM Drive to produce a sound"   :)



« Last Edit: 03/07/2015 01:27 pm by Rodal »

Offline Star-Drive

  • Member
  • Full Member
  • ****
  • Posts: 925
  • TX/USA
  • Liked: 1031
  • Likes Given: 31
...

I don't think the tilt of the balance beam about the X axis would have to be very much for the resulting change in received light amplitude to register as a shift in position.   The LDS has fractional micron resolution.    Assuming the light pattern from the fiber optic cable has a circular gaussian distribution, for any given distance the maximum light level hitting the detector is when the plane of the mirror is perpendicular to the central axis of the light beam.   Any small deviation ( arc-second) will reduce the light amplitude; which registers as a displacement.     It is impossible to align the LDS perfectly so the angle between the mirror and the light beam is never exactly 90 degrees in X and Y.    The expansion of the cavity due to heating has a very long time constant.   The slow drift in the position between RF pulses looks like a thermal response.

Interesting hypothesis. You say that in this case the LDS readings would be more sensitive to a change in angle of the reflective mirror than a change in distance. The LDS is a small distance below the X axis, so a small angular deviation around X would amount to tiny displacement (d' - d) but the small angular deviation would itself tilt the mirror and that could change the LDS readings that is roughly proportional to the reflected light.

But then this coupling between angular effect and displacement would also be present for the main movements around Z. LDS is at 35cm from Z axis, I would say it is one order of magnitude more than the distance of mirror below X axis, the lever effect of displacement/angle ratio would be lower (less angle for same displacement) but still if angle deviation is to have a significant effect in case around X it should also have a significant effect around Z.

Unfortunately the documentation for the Philtec D63 displacement sensor seems to give no information on reflective surface angular deviation impact on measurements. It indicates Ø 1.6 mm Target Spot Size. This is 3 times more than nominal operating distance of 500µm (0.5mm) so your drawing is a bit misleading with proportions. In reality the reflected dot is big relative to distance from reflector, would an angular deviation still impact the amount of light swallowed back ?

Is it you drawing or you found it elsewhere ? I ask the question because you mention the reflecting surface as being mirror. It does appear as a mirror in the pictures of pendulum at Eagleworks and this is stated explicitly in Brady's report page 3 (together with that LDS is photon time of flight, which is not, but let's proceed). But I don't find in the D63 documentation that a mirror should or could be used. It does say that These specifications represent best case performance where: the target is flat, smooth and highly reflective
This implies that the target may not be highly reflective (at the cost of decreased performance) but does highly reflective imply that a mirror is the best case ? My reading (but I'm not native English reader) is that reflective can mean a matte white material, that is a surface that does bounce near 100% of incoming photons, but in a more or less scattering manner, not necessarily as per perfect reflection. See attached picture : top the situation at hand at Eagleworks, bottom the default situation for which the LDS might be initially calibrated.

This could explain a lot about the disparity between vertical readings in the charts and stiffness parameters : when charts indicate ~1µm LDS deviation for 29.1µnN calibration pulses for instance, all could be much better explained and consistent if it meant 10µm instead.

To me there is no doubt that the LDS is operating in the far range : the cal. pulses are attractive by nature, and correspond to a displacement to the right (Y+) and that lowers the measured distance, as in the charts. Also operating in the near range would not only reverse this consistent orientation but would also increase the sensitivity : that would go against what is to be explained (apparent higher µm/µN stiffness than compatible with other known parameters, for 2 separate predictions from the dynamic periods of oscillations and the known stiffness of flexure bearings)

So, what anyone wanting to understand clearly the charts need is a way to explain a reduced sensitivity of the LDS, that would show 1µm displacement when in fact it is 10µm. Following on the D63 doc, there is a calibration procedure :
The effect of changing target reflectance is to shift the voltage output higher or lower. Factory calibrations have
the Peak Voltage set to 5.000 volts. A gain control is provided for calibration of the sensor output to various target surfaces. In-situ calibration is performed simply, by adjusting the sensor’s tip-to-target gap until the peak output voltage is attained, and then by using the gain control to set the peak voltage to full scale (5.000 volts). After setting the peak to 5 volts, the factory gap calibration chart applies for the target being measured. This procedure allows the sensor to be used to perform precision linear motion measurements on most materials.


I read "most materials : including smooth but scattering surfaces", not only mirrors. Actually, mirrors may be a very special case. My question is now if the gain control of the system is consistent with the use of a mirror reflecting target. As seen qualitatively from attached figure : a perfect mirror would bounce back in the signal detector a lot more photons than a matte white. If the gain is calibrated against matte material (say, as hypothesis, default factory settings) the baseline distance from the mirror to have a tension that shows as 500µm would be much greater than 500µm.

The protocol of tuning for nominal 500µm distance between optical fibre head and mirror is on page 3 of Brady et al report (anomalous...) :
Quote from: Brady et al. Anomalous thrust...
Prior to a test run data take, the LDS is positioned to a known displacement datum (usually 500 micrometers) via mechanical adjustments to its mounting platform. Gross adjustments are performed via set screws. Fine adjustments are performed using manually - operated calibrated screw mechanisms and a remotely controlled motorized mechanism that can be operated with the chamber door closed and the chamber at vacuum. The remote adjustment capability is necessary since the LDS datum will change whenever a change to the test facility environment affects the roll - out table or the chamber – e.g., whenever the chamber door is closed or latched and whenever the chamber is evacuated. Once the LDS displacement is adjusted in the final test environment, further adjustment between test run data takes is usually not required.

How this displacement of 500µm is known ? Has the procedure followed the step (as per D63 documentation) of homing to the distance corresponding to peak signal (whatever it's absolute level), calibrating the gain so that at this peak the signal is now 5V, and then stepping back until meeting ~4.3 V corresponding to 500µm nominal distance ? If this was not followed, or if the D63 is not mean to be operated with perfect mirror target (doubtful it wouldn't be mentioned in documentation, but who knows ?), it is possible the LDS is operating in a far range that has a much lower sensitivity (and linearity) than the nominal -2.7 mv/µm.

Edit : BTW, how comes that at some very near distance there start to be less photons bouncing back in the signal detector ? Thank zen-in for bringing up interesting questions about LDS...

Folks:

The Philtec fiber-optic cable used in their D63 reflective displacement sensor uses TWO (2) fiber optic bundles in the armored cable with one bundle being used as the transmit and the other bundle the receiver.  A description of how Philtec uses these two fiber-optic bundles to measure a distance with sub-micron resolutions is provided at their website: 

http://www.philtec.com/downloadssupport/documentlibrary/documents/applicationnotes/AboutTheSensors.pdf

Hope that helps

Best, Paul M.
Star-Drive

Offline zen-in

  • Full Member
  • ****
  • Posts: 541
  • California
  • Liked: 483
  • Likes Given: 371


Folks:

The Philtec fiber-optic cable used in their D63 reflective displacement sensor uses TWO (2) fiber optic bundles in the armored cable with one bundle being used as the transmit and the other bundle the receiver.  A description of how Philtec uses these two fiber-optic bundles to measure a distance with sub-micron resolutions is provided at their website: 

http://www.philtec.com/downloadssupport/documentlibrary/documents/applicationnotes/AboutTheSensors.pdf

Hope that helps

Best, Paul M.

That would explain the near side of the response curve better than my theory.   The response to small angular shifts of the mirror would be similar.   If the change in CM from heating caused the angle of the mirror to shift closer to a perpendicular wrt to the LDS, it would register as a decreasing displacement.   This is an attempt to explain the small negative deviations in position as shown in the curve below.

Offline Rodal

  • Senior Member
  • *****
  • Posts: 5911
  • USA
  • Liked: 6124
  • Likes Given: 5564

2) The behavior in a cylindrical pipe containing two dielectric mediums is fundamentally different: instead of the continuous, gradual change in natural frequency of the truncated cone, the  cylindrical pipe containing two dielectric mediums experiences a discontinuous, step-change, a brutal change, going from medium 1 to dielectric medium 2.  This would be more like a pipe experiencing a discontinuous step-change in diameter all of a sudden, causing a sudden change in natural frequency and a sudden change in attenuation.



Yes, the case of a gradually changing dielectric constant is very similar to that of the tapered cavity.

The step-change involves a much more complicated situation w/ the necessity of including the reflected waves from the discontinuity.  But, it may well offer enhanced performance. (?)  These types of calculations have been done w/ iterative matrix math.  I'm going to try some ideas w/ programs I have to see if I can get a better feel for this case.  (1D calculations at least)

Added: 1D modes suggest that we only have to consider terms which contain b  (ie. X or X')  The others are plain waves and cancel.

For the geometry and dielectric considered by @aero:

diameter = 8.278945 centimeter;
length = 12.24489 centimeter;
dielectricLength2 = 2.7282494103102 centimeter;
cVacuum = 299792458 meter/second;
cMedium1 = cVacuum*100 centimeter/meter;
cMedium2 = cVacuum*100/Sqrt[2.3] centimeter/meter;

The first two terms inside the first square root are:

(+(3.96427*10^(-19))* b^2)/(6.67825*10^(-39))

and

(+(1.19817*10^(-19)* p^2)/(6.67825*10^(-39))

The two terms inside the second square root (which is the third term inside the first square root) are

+4.65027*10^(-39)* (b^2)* (p^2) /((6.67825*10^(-39))^2)

and

+1.91947*10^(-39)*(p^4) /((6.67825*10^(-39))^2)

So, for this geometry and dielectric, all these terms are positive.



Where

b := If[modetype == "TM", xbesselzeros[[m + 1, n]]/Pi,  If[modetype == "TE", xprimebesselzeros[[m + 1, n]]/Pi]]

Where  Xmn=xbesselzeros[[m+1,n]] (the zeros of the Bessel function)

xbesselzeros = {{2.40483, 5.52008, 8.65373, 11.7915, 14.9309}, {3.83171, 7.01559,
  10.1735, 13.3237, 16.4706}, {5.13562, 8.41724, 11.6198, 14.796,
  17.9598}, {6.38016, 9.76102, 13.0152, 16.2235, 19.4094}, {7.58834,
  11.0647, 14.3725, 17.616, 20.8269}, {8.77148, 12.3386, 15.7002,
  18.9801, 22.2178}, {9.93611, 13.5893, 17.0038, 20.3208,
  23.5861}, {11.0864, 14.8213, 18.2876, 21.6415, 24.9349}, {12.2251,
  16.0378, 19.5545, 22.9452, 26.2668}, {13.3543, 17.2412, 20.807,
  24.2339, 27.5837}, {14.4755, 18.4335, 22.047, 25.5095, 28.8874}}

and where  X'mn=xprimebesselzeros [[m+1,n]] (the zeros of the derivative of the Bessel function)

xprimebesselzeros = {{3.83171, 7.01559, 10.1735, 13.3237, 16.4706}, {1.84118, 5.33144,
  8.53632, 11.706, 14.8636}, {3.05424, 6.70613, 9.96947, 13.1704,
  16.3475}, {4.20119, 8.01524, 11.3459, 14.5858, 17.7887}, {5.31755,
  9.2824, 12.6819, 15.9641, 19.196}, {6.41562, 10.5199, 13.9872,
  17.3128, 20.5755}, {7.50127, 11.7349, 15.2682, 18.6374,
  21.9317}, {8.57784, 12.9324, 16.5294, 19.9419, 23.2681}, {9.64742,
  14.1155, 17.774, 21.2291, 24.5872}, {10.7114, 15.2867, 19.0046,
  22.5014, 25.8913}, {11.7709, 16.4479, 20.223, 23.7607, 27.182}}

For example, for mode TE01, b= xprimebesselzeros[[1,1]]/Pi= 3.83171 / Pi =  3.83171 / 3.14159
« Last Edit: 03/07/2015 08:26 pm by Rodal »

Online Notsosureofit

  • Full Member
  • ****
  • Posts: 691
  • Liked: 747
  • Likes Given: 1729
@ RODAL

Miss. rambling looking for terms that might drop out:

f^2 = (1/4)*((-4*b^2*(dD1^2-dD2^2)*((L1/c1)^2-(L2/c2)^2))/(dD1^2*dD2^2)

      +(L1^2/c1^2 + L2^2/c2^2)*p^2

      (-/+)2*Sqrt[(-4*b^2*((L1/c1)^2-(L2/c2)^2)*((dD1*L1/c1)^2-(dD2*L2/c2)^2)*p^2)

      /(dD1^2*dD2^2)

      +(L1^2*L2^2*p^4)/(c1^2*c2^2))]

      /((L1/c1)^2-(L2/c2)^2)^2

If b = 0 then;

f^2 = [((L1/c1)^2+(L2/c2)^2)*p^2 (-/+) 2*(L1/c1)*(L2/c2)*p^2)] /(4*(L1^2/c1^2-L2^2/c2^2)^2)

If c1 = c2 = 1 then;

f^2 = [(L1^2+L2^2)*p^2 (-/+) 2*L1*L2*p^2] /(4*(L1^2-L2^2)^2)

    = [p^2/4] * [L1^2 (-/+) 2*L1*L2 + L2^2] / [(L1^2-L2^2)*(L1^2-L2^2)]

    = [p^2/4] * [(L1 (-/+) L2)^2] / [(L1-L2)^2*(L1+L2)^2]

    = [p^2/4] / [(L1 (-/+) L2)^2]

implies this p ==> p*2*pi,  that is to say my "f's" are angular frequencies and probably should be "w's".

EDIT:  Nope, just me forgetting I made c = 1.   Must be gettin' (really) old !
« Last Edit: 03/08/2015 01:24 am by Notsosureofit »

Offline frobnicat

  • Full Member
  • ****
  • Posts: 518
  • Liked: 500
  • Likes Given: 151


Folks:

The Philtec fiber-optic cable used in their D63 reflective displacement sensor uses TWO (2) fiber optic bundles in the armored cable with one bundle being used as the transmit and the other bundle the receiver.  A description of how Philtec uses these two fiber-optic bundles to measure a distance with sub-micron resolutions is provided at their website: 

http://www.philtec.com/downloadssupport/documentlibrary/documents/applicationnotes/AboutTheSensors.pdf

Hope that helps

Best, Paul M.

That would explain the near side of the response curve better than my theory.   The response to small angular shifts of the mirror would be similar.   If the change in CM from heating caused the angle of the mirror to shift closer to a perpendicular wrt to the LDS, it would register as a decreasing displacement.   This is an attempt to explain the small negative deviations in position as shown in the curve below.

Yes that makes perfect sense as for the change of slope when approaching nearer than 175µm absolute distance from target (not close enough for wavelength effects of such magnitude). But if the hypothesis is that a change of CoM can make a tilt around an axis (my bet would be around Z much greater than X) then the question to answer quantitatively is the ratio of measurement change due to angle and change due to linear displacement. Since we are talking about tilts around an axis, the two amounts of variations are mechanically linked, we can't have an angle variation of delta_alpha radians without having a linear displacement of delta_d = D delta_alpha, where D is the distance from the axis. In other words there can't be a significant delta_alpha without a significant delta_d for either rotation around X (D=a bit above 1'' ?) or Z (D=13.88''). With such constraint, what would impact more the measurement, the delta_d or the delta_alpha, or both about same order of magnitude ? I would tend to say that the delta_d influence is linear delta_measure = cst1 * delta_d, while the delta_alpha would be quadratic (cosine like) delta_measure = - cst2 * delta_alpha² since it is symmetric around a maximum. For small deltas that would strongly favor delta_d influence, but that don't preclude -cst2 * delta_alpha² being the leader depending on constant factors cst1 and cst2, and this is only valid for a delta_alpha starting from a perfect perpendicularity...

Quote from: zen-in
If the change in CM from heating caused the angle of the mirror to shift closer to a perpendicular wrt to the LDS, it would register as a decreasing displacement.

Interestingly, depending on the starting angle Alpha (0 for perpendicular), a strong delta_alpha dependent variation could be either in a direction or in the opposite direction : for cosine like effect measure1=cst3 -cst2*Alpha², measure2=cst3-cst2*(Alpha+delta_alpha)²  =>  delta_measure=-cst2*(delta_alpha+2*Alpha)*delta_alpha. If starting angle Alpha is not strictly 0 and delta_alpha is kept small relative to this rest default bias then delta_measure=-2*cst2*Alpha*delta_alpha. The sense of variation of delta_measure relative to delta_alpha would depend on the direction in which the initial non perpendicularity is biased. So no argumentation about orientations could be settled without knowing the direction of the angle bias relative to perfect perpendicularity.

But it is still unsure if the magnitude of the delta_angle effect is enough to be significant at all relative to the delta_d effect, need to be quantified, from the documentation. Unfortunately the doc kindly pointed to by Paul March give quantitative perpendicularity dependence information only for models D20 and D170, we are interested in an intermediate D63. Bottom of post I attached screen grab of the relevant charts. Is 63 to be interpolated between 20 and 170 ? From clearer D170 charts, a quadratic regression from the values of the peaks depending on angle  (0°->5, 1°->4.82, 2°->4.57, 3°->3.88, 4°->3.15, 5°->2.17) gives peak_tension_in_volts = 4.9559 - 0.1128*alpha² with alpha in degrees. I overlay the outputs of this quadratic model as red horizontal lines above the chart, the fit is consistent with a cosine effect, seems a good model as a first approximation. D20 is murkier... and how would we be supposed to interpolate the quadratic parameters of D63 between the quadratic parameters of D20 and D170 ? Or anyone with happy google karma to find those curves for D63 ?

Anyhow, if the LDS was calibrated perpendicularly and later used at another angle that could explain the apparent lack of punch of the vertical scale : indicating delta 1µm when really it should rather indicate more like delta 10µm. I will insist on that aspect until it is clarified, as it is of paramount importance to validate any sound mechanical model to rule out or not thermal displacements as sources of sustained displacements interpreted as sustained thrusts. The yellow curve at 2° on the perpendicularity dependence chart (for D170, bottom) looks a lot like the situation I drawn in green for a weak gain :


A gain smaller than the ideal one (result in green) would bring the functioning point (believed to be 500µm) closer to the peak. That would lower sensitivity and could explain a huge discrepancy between vertical readings and known actual stiffness bounds.

This could be easily checked by looking for the actual raw peak analog tension output, when slowly closing the gap the max should be 5V. And when inputing ~4.3V to the display it should show ~500.000 on vertical scale.





« Last Edit: 03/07/2015 10:13 pm by frobnicat »

Tags:
 

Advertisement NovaTech
Advertisement Northrop Grumman
Advertisement
Advertisement Margaritaville Beach Resort South Padre Island
Advertisement Brady Kenniston
Advertisement NextSpaceflight
Advertisement Nathan Barker Photography
0