Author Topic: EM Drive Developments - related to space flight applications - Thread 5  (Read 965610 times)

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2367
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 3101
  • Likes Given: 2668
" http://forum.nasaspaceflight.com/index.php?topic=38577.msg1439633#msg1439633
This whole approach requires some thought but one way or another I really think areo has a breakthrough idea here!  :D "

Yep he does, aero is one sharp man. It's called a controlled micro environment. Add a a flex cooling line and a fan... bingo!

The cooling hose only connects to the fulcrum beam and the fan isn't attached to the beam but sits on the base of the testing stand. Done correctly this would have little or no impact on the movement of the beam and remove most of the thermal balloon effects.
http://masterduct.com.tempdomain.com/SearchProducts/SearchResults/tabid/116/CategoryID/22/List/1/Level/a/ProductID/73/language/en-US/Default.aspx
I think it is something that just might work... what do you think?

Shell

Added: Sorry this is such a crude drawing and I'm sure there are mods that need to be considered, air insertion points and attachments to the beam and materials. I was simply excited.

One more thing... http://www.homedepot.com/p/Tripp-Lite-Portable-Cooling-Unit-or-Air-Conditioner-3-4-kW-120-Volt-60-Hz-12K-BTU-SRCOOL12K/203796126

Well, there is that old HVAC law that relates input temperature to output temperature and calculates airflow required to achieve X BTU of cooling or heating.

How many BTU of heat are you producing?  What degree of temperature reduction is required?  Is the airflow directed at convection, lift, outgassing, all?   

Methinks active cooling adds complexity and additional error sources.  You will STILL  have to characterize the thermal effects because active cooling can't get all of them.

If you still have to characterize thermal, your life is simpler if you don't bother with active cooling.

I like simple.  Fewer parts to break or calibrate.

In my minds eye, you have two data sets derived from say, 100 - 10second runs per data set.

Data set one is with the frustum in bad resonance mode.   You get a set of lift curves and come up with a range of "normal" lift curves with error bars around sample points.

Data set two is with the frustum in good resonance mode, and again, you get a set of lift curves and come up with a range of "normal" lift curves with error bars around sample points.

Now you have a comparison that can be made.  If the two curves differ outside the range of their respective error bars, then something needs to be explained.  If the only change is the position of a dialectric and its impact on resonance, then either its position and the corresponding change on the center of mass has to be reviewed, or, resonance creates more heat than no resonance, or it's vacuum chamber time.   That would definitely call for an independent replication in any case.

If you add a fan and vent tubes and intentional air flow, all critiques will focus on the magnetic character of the fan, the stability of airflow (you need a second set of data to prove the airflow is constant at all times), you need a third set of data to monitor the temperature at multiple locations in the system, you need to verify that there were no vertical or horizontal breaches in your duct-work before and after each run, etc. etc. etc.  You could easily most of your of time defending your air conditioner whether you got positive or negative results.

Simplicity, young weedhopper, simplicity.

Still thinking about horizontal vs. vertical ... if you have a vertical design, you're close to first light.  If you have to go horizontal, all kinds of new issues enter the mix.

You're right Glen. The more complex the easier it is to break or introduce error. Looking to dissipate around 100-200 watts this time, which is about .1 BTU/Sec or 341 BTU/hr. Not a great deal. And in this first light test I'll not be doing it.

I have designs that ramp input power into the drive into the KW ranges and we should consider something like this.


I'm planning a extended 100% high power duty cycle run in the future and (added cat on keyboard, she thought I was finished typing and should pay attention to her) need to consider some form of cooling.
« Last Edit: 10/27/2015 12:56 PM by SeeShells »

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177
" http://forum.nasaspaceflight.com/index.php?topic=38577.msg1439633#msg1439633
This whole approach requires some thought but one way or another I really think areo has a breakthrough idea here!  :D "

Yep he does, aero is one sharp man. It's called a controlled micro environment. Add a a flex cooling line and a fan... bingo!

The cooling hose only connects to the fulcrum beam and the fan isn't attached to the beam but sits on the base of the testing stand. Done correctly this would have little or no impact on the movement of the beam and remove most of the thermal balloon effects.
http://masterduct.com.tempdomain.com/SearchProducts/SearchResults/tabid/116/CategoryID/22/List/1/Level/a/ProductID/73/language/en-US/Default.aspx
I think it is something that just might work... what do you think?

Shell

Added: Sorry this is such a crude drawing and I'm sure there are mods that need to be considered, air insertion points and attachments to the beam and materials. I was simply excited.

One more thing... http://www.homedepot.com/p/Tripp-Lite-Portable-Cooling-Unit-or-Air-Conditioner-3-4-kW-120-Volt-60-Hz-12K-BTU-SRCOOL12K/203796126

Well, there is that old HVAC law that relates input temperature to output temperature and calculates airflow required to achieve X BTU of cooling or heating.

How many BTU of heat are you producing?  What degree of temperature reduction is required?  Is the airflow directed at convection, lift, outgassing, all?   

Methinks active cooling adds complexity and additional error sources.  You will STILL  have to characterize the thermal effects because active cooling can't get all of them.

If you still have to characterize thermal, your life is simpler if you don't bother with active cooling.

I like simple.  Fewer parts to break or calibrate.

In my minds eye, you have two data sets derived from say, 100 - 10second runs per data set.

Data set one is with the frustum in bad resonance mode.   You get a set of lift curves and come up with a range of "normal" lift curves with error bars around sample points.

Data set two is with the frustum in good resonance mode, and again, you get a set of lift curves and come up with a range of "normal" lift curves with error bars around sample points.

Now you have a comparison that can be made.  If the two curves differ outside the range of their respective error bars, then something needs to be explained.  If the only change is the position of a dialectric and its impact on resonance, then either its position and the corresponding change on the center of mass has to be reviewed, or, resonance creates more heat than no resonance, or it's vacuum chamber time.   That would definitely call for an independent replication in any case.

If you add a fan and vent tubes and intentional air flow, all critiques will focus on the magnetic character of the fan, the stability of airflow (you need a second set of data to prove the airflow is constant at all times), you need a third set of data to monitor the temperature at multiple locations in the system, you need to verify that there were no vertical or horizontal breaches in your duct-work before and after each run, etc. etc. etc.  You could easily most of your of time defending your air conditioner whether you got positive or negative results.

Simplicity, young weedhopper, simplicity.

Still thinking about horizontal vs. vertical ... if you have a vertical design, you're close to first light.  If you have to go horizontal, all kinds of new issues enter the mix.

You're right Glen. The more complex the easier it is to break or introduce error. Looking to dissipate around 100-200 watts this time, which is about .1 BTU/Sec or 341 BTU/hr. Not a great deal. And in this first light test I'll not be doing it.

I have designs that ramp input power into the drive into the KW ranges and we should consider something like this.


I'm planning a extended 100% high power duty cycle run in the future and (added cat on keyboard, she thought I was finished typing and should pay attention to her) need to consider some form of cooling.

341 BTU at 100 watts.  Basically a tungsten light bulb's worth of heat that has to vanish.   That's a lot.  :)

Attached is a spreadsheet to think about.  It's a "random" test generator.  Remember Tajmar reported oxidation over his runs reducing Q.  You don't want to do 100 runs in resonance and 100 runs in no resonance.  Everyone will point to oxidation effects if you do that.  In fact, you can make a case that oxidation by itself will generate measurable thrust.  (one of my earliest posts).

The attached spreadsheet gives you a way to randomize the trials.  Hit the f9 key to recalculate until you're pretty close to 100 samples of each.  Record the data (copy paste values, you can't save it, it regenerates every time the spreadsheet changes), and run accordingly, then, if you did generate those data sets, and there's something there, you can assert that oxidation was controled for, and perhaps your curves will drift in a conjoint set permitting the measurement of oxidation effects over time and the subtraction of those effects in the final analysis.

Such an approach would honor the standard of "random trials" which would be good stat stuff.

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2367
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 3101
  • Likes Given: 2668
" http://forum.nasaspaceflight.com/index.php?topic=38577.msg1439633#msg1439633




341 BTU at 100 watts.  Basically a tungsten light bulb's worth of heat that has to vanish.   That's a lot.  :)

Attached is a spreadsheet to think about.  It's a "random" test generator.  Remember Tajmar reported oxidation over his runs reducing Q.  You don't want to do 100 runs in resonance and 100 runs in no resonance.  Everyone will point to oxidation effects if you do that.  In fact, you can make a case that oxidation by itself will generate measurable thrust.  (one of my earliest posts).

The attached spreadsheet gives you a way to randomize the trials.  Hit the f9 key to recalculate until you're pretty close to 100 samples of each.  Record the data (copy paste values, you can't save it, it regenerates every time the spreadsheet changes), and run accordingly, then, if you did generate those data sets, and there's something there, you can assert that oxidation was controled for, and perhaps your curves will drift in a conjoint set permitting the measurement of oxidation effects over time and the subtraction of those effects in the final analysis.

Such an approach would honor the standard of "random trials" which would be good stat stuff.
Thanks!
I'm coating the frustum in silver which Tajmar didn't do, had more funds I would do a gold flash over that but silver oxide is still conductive. I didn't miss Tajmar's comment or forget your's, I took that info and made use of it. Thanks again!

Did you notice I built my Faraday cage bigger than needed? In a future test I'm considering doing a good horizontal test, not spinning around but force directed at the other end of the arm horizontally, to late for me to change this one but I did my cage big enough.   

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177
" http://forum.nasaspaceflight.com/index.php?topic=38577.msg1439633#msg1439633




341 BTU at 100 watts.  Basically a tungsten light bulb's worth of heat that has to vanish.   That's a lot.  :)

Attached is a spreadsheet to think about.  It's a "random" test generator.  Remember Tajmar reported oxidation over his runs reducing Q.  You don't want to do 100 runs in resonance and 100 runs in no resonance.  Everyone will point to oxidation effects if you do that.  In fact, you can make a case that oxidation by itself will generate measurable thrust.  (one of my earliest posts).

The attached spreadsheet gives you a way to randomize the trials.  Hit the f9 key to recalculate until you're pretty close to 100 samples of each.  Record the data (copy paste values, you can't save it, it regenerates every time the spreadsheet changes), and run accordingly, then, if you did generate those data sets, and there's something there, you can assert that oxidation was controled for, and perhaps your curves will drift in a conjoint set permitting the measurement of oxidation effects over time and the subtraction of those effects in the final analysis.

Such an approach would honor the standard of "random trials" which would be good stat stuff.
Thanks!
I'm coating the frustum in silver which Tajmar didn't do, had more funds I would do a gold flash over that but silver oxide is still conductive. I didn't miss Tajmar's comment or forget your's, I took that info and made use of it. Thanks again!

Did you notice I built my Faraday cage bigger than needed? In a future test I'm considering doing a good horizontal test, not spinning around but force directed at the other end of the arm horizontally, to late for me to change this one but I did my cage big enough.   

IMHO controling for oxidation is good, and I'm sure when you look in any 19th century mirror, you know that your coating will oxidize.   :)  But you're right, silver oxide should behave better than copper oxide, although it's not as pretty after a century or two.

One other stupid thought, is there any way at all to measure the Q value at the beginning and end of each run, or during the run?  That data in itself could be quite useful, especially if the hypothesis relates Q to lift.

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2367
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 3101
  • Likes Given: 2668
" http://forum.nasaspaceflight.com/index.php?topic=38577.msg1439633#msg1439633




341 BTU at 100 watts.  Basically a tungsten light bulb's worth of heat that has to vanish.   That's a lot.  :)

Attached is a spreadsheet to think about.  It's a "random" test generator.  Remember Tajmar reported oxidation over his runs reducing Q.  You don't want to do 100 runs in resonance and 100 runs in no resonance.  Everyone will point to oxidation effects if you do that.  In fact, you can make a case that oxidation by itself will generate measurable thrust.  (one of my earliest posts).

The attached spreadsheet gives you a way to randomize the trials.  Hit the f9 key to recalculate until you're pretty close to 100 samples of each.  Record the data (copy paste values, you can't save it, it regenerates every time the spreadsheet changes), and run accordingly, then, if you did generate those data sets, and there's something there, you can assert that oxidation was controled for, and perhaps your curves will drift in a conjoint set permitting the measurement of oxidation effects over time and the subtraction of those effects in the final analysis.

Such an approach would honor the standard of "random trials" which would be good stat stuff.
Thanks!
I'm coating the frustum in silver which Tajmar didn't do, had more funds I would do a gold flash over that but silver oxide is still conductive. I didn't miss Tajmar's comment or forget your's, I took that info and made use of it. Thanks again!

Did you notice I built my Faraday cage bigger than needed? In a future test I'm considering doing a good horizontal test, not spinning around but force directed at the other end of the arm horizontally, to late for me to change this one but I did my cage big enough.   

IMHO controling for oxidation is good, and I'm sure when you look in any 19th century mirror, you know that your coating will oxidize.   :)  But you're right, silver oxide should behave better than copper oxide, although it's not as pretty after a century or two.

One other stupid thought, is there any way at all to measure the Q value at the beginning and end of each run, or during the run?  That data in itself could be quite useful, especially if the hypothesis relates Q to lift.
Got me there, a girl likes shiny things. ;)

Actively monitor Q would be better and having datalogging of Q vs thrusts would be one of my goals. Group ideas? I have some but there are not as good as some of the Microwave cavity geeks here.

Last night I was trolling Ebay and Amazon and a couple other places for a digital scale that was cheap that I could do just that. Didn't find much but it's on my list to dig more.

Shell
« Last Edit: 10/27/2015 01:56 PM by SeeShells »

Offline wallofwolfstreet

  • Full Member
  • *
  • Posts: 165
  • Liked: 169
  • Likes Given: 436
Now you have a comparison that can be made.  If the two curves differ outside the range of their respective error bars, then something needs to be explained.  If the only change is the position of a dialectric and its impact on resonance, then either its position and the corresponding change on the center of mass has to be reviewed, or, resonance creates more heat than no resonance , or it's vacuum chamber time.   That would definitely call for an independent replication in any case.

Resonance does create more heat than no resonance though.  This is very well known and nothing new; it has to do with impedance, power factor, etc.  The short of it is that in the real world, all resonant systems will reach a maximum stored energy where they dissipate energy at the same rate they take energy in, because no system is perfectly resonant in the sense of zero losses.  At resonance, more energy is delivered to the system than if it were off resonance.  Hence more power is dissipated and therefore the system gets hotter.   

So unfortunately, characterizing thermal lift by screwing around with resonance won't work, because thermal lift is itself intimately tied in with resonance.  Change resonance, change thermal lift.   

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177
Now you have a comparison that can be made.  If the two curves differ outside the range of their respective error bars, then something needs to be explained.  If the only change is the position of a dialectric and its impact on resonance, then either its position and the corresponding change on the center of mass has to be reviewed, or, resonance creates more heat than no resonance , or it's vacuum chamber time.   That would definitely call for an independent replication in any case.

Resonance does create more heat than no resonance though.  This is very well known and nothing new; it has to do with impedance, power factor, etc.  The short of it is that in the real world, all resonant systems will reach a maximum stored energy where they dissipate energy at the same rate they take energy in, because no system is perfectly resonant in the sense of zero losses.  At resonance, more energy is delivered to the system than if it were off resonance.  Hence more power is dissipated and therefore the system gets hotter.   

So unfortunately, characterizing thermal lift by screwing around with resonance won't work, because thermal lift is itself intimately tied in with resonance.  Change resonance, change thermal lift.   

Now that's a bummer.

It would therefore seem to me that runs would therefore have to be done pointing the gizmo so thrust was the opposite direction of the lift.  You would not have a measurement of actual "anti-lift" but if the thrust overwhealmed the additional thermal, then you still might see something.

Is there a means of calculating the thermal delta with and without resonance? 

Alternatively ... flipping to the horizontal mode ...

There has to be some clean way outside of a vacuum chamber...

Offline wallofwolfstreet

  • Full Member
  • *
  • Posts: 165
  • Liked: 169
  • Likes Given: 436
Now that's a bummer.

It would therefore seem to me that runs would therefore have to be done pointing the gizmo so thrust was the opposite direction of the lift.  You would not have a measurement of actual "anti-lift" but if the thrust overwhealmed the additional thermal, then you still might see something.

Is there a means of calculating the thermal delta with and without resonance? 

Alternatively ... flipping to the horizontal mode ...

There has to be some clean way outside of a vacuum chamber...

Quote
Is there a means of calculating the thermal delta with and without resonance?

Yes, the equations are basic undergraduate electronics and some slightly more complicated heat transfer.  Getting accurate quantities for the variables isn't easy however, and validating the accuracy would require experimental runs regardless, so this procedure doesn't really save any work. 

Quote
Alternatively ... flipping to the horizontal mode ...

Probably the best bet (in conjunction with the following).

Quote
There has to be some clean way outside of a vacuum chamber...

There is.  Enclose the whole thing in a hermetically sealed chamber (eliminates hot air jets) of high rigidity (minimizes buoyancy) with high heat capacity (minimizes convection off chamber walls) and high heat transfer coefficient (minimizes anisotropy of convection off of chamber walls).

Ideally use some form of mixing within the chamber to ensure relatively even heat transfer across the inner surface of the chamber.  This further improves the uniformity of convection off the chamber walls.  With uniform convection off the chamber walls, thrust can be determined by flipping the orientation (up to down) of the chamber and emdrive with it.  This uses the same technique which was (incorrectly) applied to Iulian's up/down data.  A properly sealed chamber of this type has only thermal convection effects that are orientation independent, something that a frustum never will.

Other posters have already advised for this design, and didn't Shawyer even use it?  That may be one of the few things he actually got right. 

In general, "normalizing" thermal effects won't work.  There are many uncontrolled factors that feed into convection force: ambient pressure, humidity, ambient temperature (since it affects viscosity and subsequently Reynolds number), even things like surface roughness come into play by way of Reynolds number.  Trying to normalize that out will always be suspect unless fairly rigorous controls are in place, probably above what a DIY could employ.

Offline aero

  • Senior Member
  • *****
  • Posts: 2996
  • 92129
  • Liked: 797
  • Likes Given: 282
I think bagging and venting is a reasonable idea. You will want to take a lesson from wind tunnel design and reverse the fan to draw the warm air out. That will eliminate fan induced air turbulence from the system and you can measure the air temperature easily in the exit tube. (as well as inlet).

And since warm air wants to rise, perhaps the exit should be the top pipe so that the rising warm air naturally exits instead of reversing flow direction to the bottom before exiting. This again reduces air turbulence.

Drawing the air out instead of forcing it in should keep the insulation drawn up snugly to the chicken wire structure and so maintain the balloon volume. Air jetting from the frustum gaps into the bag shouldn't cause detectable effects as doing so would violate the same COM law that we have discussed forever.

The hot air inside the frustum itself will cause a lift force but the only way to eliminate that is to eliminate the air, or to create a perfectly rigid, perfectly sealed frustum and neither of those are likely.

Retired, working interesting problems

Offline Notsosureofit

  • Full Member
  • ****
  • Posts: 668
  • Liked: 719
  • Likes Given: 1407
https://www.dropbox.com/s/a0px866l2gl50jd/2015-10-27%2012.31.45.jpg?dl=0

Finally have a uhv chamber available for the forseeable future.

If someone could post the picture for me, that would be great !

This chamber could handle a 20" diameter remote system, etc, there are some feedthrus available as well....

Offline aero

  • Senior Member
  • *****
  • Posts: 2996
  • 92129
  • Liked: 797
  • Likes Given: 282
While we're thinking about it folks, something that has not been addressed is the ambient conditions of Shell's lab.

As I understand it, the lab is at about 8,400 feet elevation so the atmosphere won't be anything like sea level conditions. I think the air at that elevation is commonly quite dry and I know that lift will be quite reduced from those at sea level. So while Shell is not using a vacuum chamber, the elevation of her lab does take her a step in that direction.
Retired, working interesting problems

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177


There is.  Enclose the whole thing in a hermetically sealed chamber (eliminates hot air jets) of high rigidity (minimizes buoyancy) with high heat capacity (minimizes convection off chamber walls) and high heat transfer coefficient (minimizes anisotropy of convection off of chamber walls).

Ideally use some form of mixing within the chamber to ensure relatively even heat transfer across the inner surface of the chamber.  This further improves the uniformity of convection off the chamber walls.  With uniform convection off the chamber walls, thrust can be determined by flipping the orientation (up to down) of the chamber and emdrive with it.  This uses the same technique which was (incorrectly) applied to Iulian's up/down data.  A properly sealed chamber of this type has only thermal convection effects that are orientation independent, something that a frustum never will.

I see the advantages of a hermetically sealed chamber, however, running power into the chamber would require extreme care to retain the seal.  At that point, you've essentially built a vacuum chamber except you don't require the 14 psi survival for the chamber.  This is a high cost for the DIY community.

Greg Egan posted some calculations in John Baez' blog about the predicted convective thrust.  Expanding on that, if all cumulative thermal effects can be a-prior modeled and calculated, at least to within a specific order of magnitude, then what a DIY can do is establish a hypothetical floor for the minimum required thrust to be observable.  i.e. if Egan's calculation stated that 28 watts will generate 127 micro-newtons of convective thrust, if the observed were 1,270 micro-newtons then a closer look would be warranted?

Absent the costs of doing hermetic seal / aka wimpy vacuum chamber, you could take an alternative approach.

Does the following approach stick to the wall or dribble down like a 3 month old pizza?
I suggested earlier a 2 sample group, no-resonance, resonance.  You promptly pointed out that thermal changes so that's not sufficient.  Change it to 4 groups, up resonance, down resonance, up no resonance, down no resonance.  If the error bars in the up-down don't overlap with resonance, but do overlap with no resonance, would that not in principle be the start of an experimental design with some teeth?

Offline aero

  • Senior Member
  • *****
  • Posts: 2996
  • 92129
  • Liked: 797
  • Likes Given: 282
https://www.dropbox.com/s/a0px866l2gl50jd/2015-10-27%2012.31.45.jpg?dl=0

Finally have a uhv chamber available for the forseeable future.

If someone could post the picture for me, that would be great !

This chamber could handle a 20" diameter remote system, etc, there are some feedthrus available as well....
Retired, working interesting problems

Offline SeeShells

  • Senior Member
  • *****
  • Posts: 2367
  • Every action there's a reaction we try to grasp.
  • United States
  • Liked: 3101
  • Likes Given: 2668
While we're thinking about it folks, something that has not been addressed is the ambient conditions of Shell's lab.

As I understand it, the lab is at about 8,400 feet elevation so the atmosphere won't be anything like sea level conditions. I think the air at that elevation is commonly quite dry and I know that lift will be quite reduced from those at sea level. So while Shell is not using a vacuum chamber, the elevation of her lab does take her a step in that direction.
Low O2 content as well.

Looks like my insulation is going to be late for the lab, it was coming from TX but now IN.. sigh.  This weekend looks to be a good time to install.

Shell
« Last Edit: 10/27/2015 04:21 PM by SeeShells »

Offline wallofwolfstreet

  • Full Member
  • *
  • Posts: 165
  • Liked: 169
  • Likes Given: 436
I see the advantages of a hermetically sealed chamber, however, running power into the chamber would require extreme care to retain the seal.  At that point, you've essentially built a vacuum chamber except you don't require the 14 psi survival for the chamber.  This is a high cost for the DIY community.

I see what you're thinking, but it really isn't the case.  A sealed chamber, and a vacuum rated sealed chamber are orders of magnitude different beasties in the cost and effort they take.  I've made perfectly valid hermetically sealed chambers with a 3D printer, acetone to cure the surface and plumbers silicone to seal gaps for wiring.  These would have been useless as vacuum chambers, but they worked fine with the pressure differentials I needed them to.  Making something airtight is only an issue with significant pressure differentials.  The slight pressure differential that would arise from air heating (with a high heat capacity and conductive chamber) over a short run wouldn't require anything special.       

Quote
Greg Egan posted some calculations in John Baez' blog about the predicted convective thrust.  Expanding on that, if all cumulative thermal effects can be a-prior modeled and calculated, at least to within a specific order of magnitude, then what a DIY can do is establish a hypothetical floor for the minimum required thrust to be observable.  i.e. if Egan's calculation stated that 28 watts will generate 127 micro-newtons of convective thrust, if the observed were 1,270 micro-newtons then a closer look would be warranted?

Yeah, but now the workload has shifted from designing a low systematics experimental rig to designing a high thrust emdrive.  You have to choose which one you think is harder/more likely.  I'd go with the rig myself.

Quote
Absent the costs of doing hermetic seal / aka wimpy vacuum chamber, you could take an alternative approach.

Does the following approach stick to the wall or dribble down like a 3 month old pizza?
I suggested earlier a 2 sample group, no-resonance, resonance.  You promptly pointed out that thermal changes so that's not sufficient.  Change it to 4 groups, up resonance, down resonance, up no resonance, down no resonance.  If the error bars in the up-down don't overlap with resonance, but do overlap with no resonance, would that not in principle be the start of an experimental design with some teeth?

That's a clever idea, but I don't think it's valid because convective force is both orientation dependent and power input dependent (ie. resonance dependent), so I would expect that (up, resonance) is different from (down, resonance) in a way which itself differs from the difference between (up, no resonance) and (down, no resonance).

Trust me though that with reasonable pressure differentials, air-tight seals aren't as hard as you'd think they'd be.  You can quantify that pressure differential with basic thermo and heat transfer equations.  If it's only a few kPa, hermetically sealing isn't a big issue.   
« Last Edit: 10/27/2015 05:05 PM by wallofwolfstreet »

Offline rfmwguy

  • EmDrive Builder (retired)
  • Senior Member
  • *****
  • Posts: 2166
  • Liked: 2684
  • Likes Given: 1124
My money is on simplicity and Uncle Glenn for extracting deltas out of thermal lift during mag on to mag off cycles ;)

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177
I see the advantages of a hermetically sealed chamber, however, running power into the chamber would require extreme care to retain the seal.  At that point, you've essentially built a vacuum chamber except you don't require the 14 psi survival for the chamber.  This is a high cost for the DIY community.

I see what you're thinking, but it really isn't the case.  A sealed chamber, and a vacuum rated sealed chamber are orders of magnitude different beasties in the cost and effort they take.  I've made perfectly valid hermetically sealed chambers with a 3D printer, acetone to cure the surface and plumbers silicone to seal gaps for wiring.  Making something airtight is only an issue with significant pressure differentials.  The slight pressure differential that would arise from air heating (with a high heat capacity and conductive chamber) over a short run wouldn't require anything special.       

Quote
Greg Egan posted some calculations in John Baez' blog about the predicted convective thrust.  Expanding on that, if all cumulative thermal effects can be a-prior modeled and calculated, at least to within a specific order of magnitude, then what a DIY can do is establish a hypothetical floor for the minimum required thrust to be observable.  i.e. if Egan's calculation stated that 28 watts will generate 127 micro-newtons of convective thrust, if the observed were 1,270 micro-newtons then a closer look would be warranted?

Yeah, but now the workload has shifted from designing a low systematics experimental rig to designing a high thrust emdrive.  You have to choose which one you think is harder/more likely.  I'd go with the rig myself.

Quote
Absent the costs of doing hermetic seal / aka wimpy vacuum chamber, you could take an alternative approach.

Does the following approach stick to the wall or dribble down like a 3 month old pizza?
I suggested earlier a 2 sample group, no-resonance, resonance.  You promptly pointed out that thermal changes so that's not sufficient.  Change it to 4 groups, up resonance, down resonance, up no resonance, down no resonance.  If the error bars in the up-down don't overlap with resonance, but do overlap with no resonance, would that not in principle be the start of an experimental design with some teeth?

That's a clever idea, but I don't think it's valid because convective force is both orientation dependent and power input dependent (ie. resonance dependent), so I would expect that (up, resonance) is different from (down, resonance) in a way which itself differs from the difference between (up, no resonance) and (down, no resonance).

Trust me though that with reasonable pressure differentials, air-tight seals aren't as hard as you'd think they'd be.  You can quantify that pressure differential with basic thermo and heat transfer equations.  If it's only a few kPa, hermetically sealing isn't a big issue.


I agree that asymetries could kill you, but assuming the asymetric effects are small, I just finished a simulation of what it could look like.

Attached is a spreadsheet.  I can't upload excel macros to this site so to see it work, you'll have to include the VBA in Sheet2 and load that into the VBA code for the Button.

Here's the question I asked myself.

Assume thermal lift is a #, i.e. 300 units of something.
Assume the device lift is a #, i.e. 30 units of something.
Assume the variability is a #, i.e. 10% of the corresponding unit

Over 200 runs, randomly selecting resonance, no resonance, up, down, what would data look like, and how easy would it be to see a signal?

You can vary any of the fields in Yellow, Thermal Lift, thrust, variability to see whether you could detect a signal through the noise.  Won't work unless you know how to add the VBA in sheet 2 to the button. (Hint: Developer tools, right click on button, NEW, I think)

It looks to me, your asymetry issue asside, and that can be added to the simulation, that it should be possible to detect a signal for thrust at 10% of the thermal value, if you can keep your noise variability at +- 10 %.  At +-15% it gets iffy. At +- 25% it gets really iff, and above that, it's indistiguishable from random.

Hence, I contend that a clean signal, good measurement, and lots of replications should be able to find a signal without investing in a hermetic chamber.

You're good at busting my bubble so I await your next effort.  :)

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177
p.s. if the no resonance thermal is less thant the resonance, then the signal to noise level improves rather than degrades.

I'll upload a spreadsheet in a bit that allows you to insert asymetries as well.

edit:  Spreadsheet now includes ability to set thermal lift for no resonance, up thermal asymetry, down thermal asymetry.

Load the same macro as before.
« Last Edit: 10/27/2015 05:48 PM by glennfish »

Offline wallofwolfstreet

  • Full Member
  • *
  • Posts: 165
  • Liked: 169
  • Likes Given: 436

I don't really use excel that much, I use Matlab for stuff like this.  I won't have the time to try out your program until later. 

If I understand you correctly, if you assume negligible asymmetry, you end up with:

thermal lift= X +/- (variability)*X
device lift= Y +/- (variability)*Y

which is just a rephrasing of:
Quote
Assume thermal lift is a #, i.e. 300 units of something.
Assume the device lift is a #, i.e. 30 units of something.
Assume the variability is a #, i.e. 10% of the corresponding unit

In that case, then why bother with resonance vs. no resonance?  The "Iulian method" would work just fine since we (incorrectly) assumed no asymmetry.

If we include asymmetry, which we must given that the emdrive is fundamentally asymmetric, and employ your up/down and resonance/no resonance method, the model becomes (before adding in variability):

Xres,up = thermal lift with resonance, up orientation
Xres,down = thermal lift with resonance, down orientation
Xno res,up = thermal lift with no resonance, up orientation
Xno res,down =thermal lift with no resonance, down orientation
Y = device lift

So you have 4 measured data sets based on the experimenter's control of (resonance, up)....(no resonance, down).
where

(resonance, up) = Xres,up + Y
...
(no resonance, down) = Xno res,down + Y

Is there a way to uniquely solve for 5 variables from 4 data sets?  Unfortunately not.

In general, without assuming away asymmetry, you can't pull out a thrust signal.   

So:
Quote
Hence, I contend that a clean signal, good measurement, and lots of replications should be able to find a signal without investing in a hermetic chamber
only works if we assume negligible asymmetry (so you're right in your specific case, but not in general). 
« Last Edit: 10/27/2015 06:22 PM by wallofwolfstreet »

Offline glennfish

  • Full Member
  • ****
  • Posts: 424
  • Liked: 283
  • Likes Given: 177

I don't really use excel that much, I use Matlab for stuff like this.  I won't have the time to try out your program until later. 

If I understand you correctly, if you assume negligible asymmetry, you end up with:

thermal lift= X +/- (variability)*X
device lift= Y +/- (variability)*Y

which is just a rephrasing of:
Quote
Assume thermal lift is a #, i.e. 300 units of something.
Assume the device lift is a #, i.e. 30 units of something.
Assume the variability is a #, i.e. 10% of the corresponding unit

In that case, then why bother with resonance vs. no resonance?  The "Iulian method" would work just fine since we (incorrectly) assumed no asymmetry.

If we include asymmetry, which we must given that the emdrive is fundamentally asymmetric, and employ your up/down and resonance/no resonance method, the model becomes (before adding in variability):

Xres,up = thermal lift with resonance, up orientation
Xres,down = thermal lift with resonance, down orientation
Xno res,up = thermal lift with no resonance, up orientation
Xno res,down =thermal lift with no resonance, down orientation
Y = device lift

So you have 4 measured data sets based on the experimenters control of (resonance, up)....(no resonance, down).
where

(resonance, up) = Xres,up + Y
...
(no resonance, down) = Xno res,down + Y

Is there a way to uniquely solve for 5 variables from 4 data sets?  Unfortunately not.

So in general, without assuming away asymmetry, you ca't pull out a thrust signal.   

So
Quote
Hence, I contend that a clean signal, good measurement, and lots of replications should be able to find a signal without investing in a hermetic chamber
only works if we assume negligible asymmetry (so you're right in your specific case, but not in general).

Now you done tossed something onto the table that I hadn't thought of.

I would assume that your Y is zero without resonance, but it may not be, so I have to add in a non-resonant value for Y.

Asymmetry was in the latest simulator upload.

As for why bother to include the no-resonance?  Well, there are a lot of folks here who believe it's a meaningful thing to have, so I'd include those tests just to be politically correct.

In any event, it looks to me like it is possible to see a signal as small as 1/10th the thermal, in spite of all the above, provided the noise can be kept low.

Edit:  Now has the following input values
1.  Thermal Lift Resonance
2.  Thermal Lift No Resonance
3.  Asymmetric lift UP
4.  Asymmetric Lift DOWN
5.  Device Lift Resonance
6.  Device Lift No-Resonance
7.  Variability from perfect value (%)

Still looks like unless some of the values are really whacked, See-Shells could See-Shomething if her signal is clean, and the thrust is > 10% of the thermal effects. 

VBA has to be manually added to the button from Sheet 2.  No changes in VBA

edit:
If this simulation is even vaguely close to reality, even Beagleworks should have been able to see a signal with enough trials....

edit:

no downloads?  Think I'll spend the next two days memorizing Endymion. 
« Last Edit: 10/28/2015 01:04 AM by glennfish »

Tags: