FM Concepts posts working video 4/7/2009

[Justalabrat]

New video from DMBoss 

http://www.youtube.com/watch?v=2p9Hjj65Nko

[DMBoss]

New video from DMBoss 

http://www.youtube.com/watch?v=2p9Hjj65Nko

Actually there's 3 videos now, and I've set up a website using the handle of "QDMechanic" as in the YouTube videos.

working feverishly to set up a decent website, as I have a lot to share....  But a temporary page with better details regards the video part 3 is set up on my site for now:

http://qdmechanic.com/

Links to all 3 videos, and higher resolution copies of the stills seen in the 3rd video are posted.  Along with some brief comments on the progress.

There appears to be some gain manifesting at this early stage, when you massage some parameters.  For now after playing in "rough" to explore the parameter changes, I set about a better feedback system for gauging whether a change I make is better or worse.

Accurate speed run down is nice and simple as that yardstick.  And now I've got a much better speed rundown measuring/displaying method.

So I can go back and incrementally change things and record what works best in what direction.

It does not self run "out of the box" but can one expect that from a device which resulted from months or years of trial and error?  But it does have a non-zero net torque in dynamic fashion.  And running it in the prefferred or forward direction has longer run down speed than friction alone does, and running in reverse, has a shorter run down time than friction alone!

This is typical of many "toy" level devices.  I've measured gains from a dozen different ones so far - but the net gain is VERY small in most cases - making them extremely tricky and touchy to get everything "goldilocks" right to get them to self run.

More details on these other things, and this one when I get my site coded and written. And I will be offering some services to this "O/U" community - I have a well set up lab/shop for design, fab, testing and simulation.  Many persons have great or not so great ideas, but  do not have the means to try them or make some of the often custom components.  With my CNC etc, I can whip out parts quickly, and cost effectively. (and also can save heartache and wasted time/effort/money by sharing insights, or helping with simulation or go-nogo reality checks and other free services)

You or someone may want to try such and such idea, but without a lathe, and mill, can't make a decent facsimile of some of the parts.  But with those cost effective parts in hand, they can "tinker" and experiment to their heart's content, and with enough trial and error as a group we can make many of the root princples that produce O/U - manifest!

And there is nothing wrong with "toys" as proofs of concept.  Just because it's not gonna replace a 1000 MW generating plant tomorrow, does not mean it has no utility!  Here's a famous quote that sums it up:

"We are more ready to try the untried when what we do is inconsequential. Hence the fact that many inventions had their birth as toys."     Eric Hoffer (1902 - 1983)

DMBoss

[Omega_0]

DMBoss 

Congrats on getting a gain. And your workmanship is awesome.

How many tries you did with run-downs ? Is this gain an average value and you are getting it consistently or it happened only once ?

[DMBoss]

DMBoss 

Congrats on getting a gain. And your workmanship is awesome.

How many tries you did with run-downs ? Is this gain an average value and you are getting it consistently or it happened only once ?

Omega_0:

Well these kinds of devices and there are many.... involve some VERY small net forces and torques - so cobbling it together is not conducive to getting it to work!  Hence attention to detail and doing a protocol or method, with precision is the only way to be able to repeat it with a reasonable effort.

The root inventors of these things often played by hand with variations for months or years, but then replication is a nightmare - especially if they or us armchair observers do not grasp what really makes it tick!

So I spent a lot of time considering the replication design, with a view to being able to repeatably duplicate it, and be able to make incremental modifications to see what aspects are important and what aren't. 

there are literally hundreds of thousands, or millions of possible combinations of interrelated variables in such a system even though it appears simple!  Because of the non-linearity, and the interrelationships - if you have say 5 factors each with 6 degrees of freedom and all rely upon the other relationships - it is like the odds of a lottery - where factorials are involved. (where 30 elements is then 30x29x28 etc for number of combos - excuse my simplistic math, I'm no math whiz)

So if you just throw a white scarf around your neck and go flying off in the old biplane tradition - you may miss the sweet spots by tens or hundreds of miles!

Anyway that's why some serious attention to detail and some level of precision is needed for these kinds of "toy" level devices.

In answer to your questions, yes I've been doing many many run downs.  Initially, and in the preliminary case posted in the video part 3, and on my website the goal was merely to get reliable measuring feedback.  Though as a side bonus I got a reliable non zero result.

Then I went back and refined the measuring protocol to reduce the noise or error margin and make it VERY repeatable and reliable.  Then I could start some "hunting" for what makes more gain.

I've now got the measuring much more refined as you can see detailed in the attached PDF.

My error margins are lower, and no more scope math glitches. (those troubling bumps in the preliminary traces)

Now I can get the kinds of smooth run downs as seen in the attached all the time.  And analysis of many such runs, usually with 5 sub test runs each, the error is well below the non zero result.

Oh and I'm gaining a visceral insight from playing with the bench version along with many many hours of 3D magnetics sims combined with 3D dynamic motion sims that incorporate inertia, mass gravity etc..... of what makes the gain.  (it's not what most assume it is)

Example of why precision is important:  I used "drill rod" for the shaft, which is supposed to be ground to a high tolerance on the OD.  And I "assumed" it was round and true.  I turned the boss/shoulder for the bearing and left the rest of the 1/2" drill rod in the highly polished state.

But my results were all over the place regards non-zero.  So I happened to check the bare shaft, and darn it if the thing was a banana! (not straight or true)

So I machined it, and had to remove 0.003" from the diameter to make it true, round and coaxial with the bearing boss!  Does not sound like much - but now the beastie is repeatable as to a non zero result all the time!

the banana bend in the shaft was attenuating the gain mechanism for certain orientations it was inserted into the stator block.  But now that it's true - the small non zero result happens no matter what orientation it's inserted with!  (note also that in FM's video, his stator vs shaft orientation affected his rotor speed as he turns the stator to various angles in the video...)

DMBoss

[DMBoss]

Testing continues.  A word of caution to any other replication attempts:

BEWARE OF YOUR BEARING CHOICE!

I initially used a full ceramic bearing with my replication.  For the reason that these are the very lowest friction coefficients you can get with radial ball bearings.  However the bearing has several possible influences on the outcome, especially since it is resides in rather high magnetic fields mounted to this rotor, or to FM's original rotor too.

FM used a SHCS (Socket Head Cap Screw) as his rotor shaft.  His was a black oxide finish in the video and his photos, indicating it is plain steel but hardened as SHCS's are.  Thus it is strongly ferromagnetic.  And his unipolar rotor magnets (all axial magnets are oriented the same way up), have their return flux travel down the inside of this ring of axial magnets and any steel or ferromagnetic material in that region gets some serious flux and flux density!

I used a larger piece of steel, machined to accept the bearing and install into the stator base, because my version has slightly larger magnets, and more of them to maintain the same magnet diameter to rotor mounting pattern circumferent ratio.  Therefore my rotor diameter and the pattern diameter are larger - and the gap between the return flux path of the center shaft to the stator magnets and the rotor magnets might be important, so I enlarged the shaft too.

I then decided to try a normal bearing, as all single bearings of different types, grades etc can behave differently in several ways.  I chose my initial bearing size to be standard, so lots of various types and materials were available to interchange with my initial choice.

It is almost impossible to obtain a metallic ball bearing that is not ferromagnetic except for full or hybrid ceramics, which FM did not use.

Now I obtained a stainless ball bearing with a stainless retainer or "cage" for the balls.  Open style as FM used and as my full ceramic bearing is. (you cannot use stock grease or lube in any bearing for this project as the viscous friction is wayyyy too high - you need to either get open bearings and clean then out with a solvent and compressed air and relube with a drop of very light oil, or leave them dry as with ceramics; or pry off any shields to clean and relube metallic ones)

I installed this stainless ball bearing and did speed run down tests as before, and the stainless bearing is absolutely terrible.  This new bearing even when just inserted into the rotor journal, and the shaft installed, locks up or grabs the shaft due the high magnetic fields, and the fact so called stainless bearings are 400 series, which is highly ferromagnetic as are normal chrome steel bearings. (very rarely you can special order bearings in 316 stainless, but they are not easily available - that would be only very slightly magnetic but it's moot as you can't easily get 316)

The ceramic "DRY" rundown has an acceleration of -4 to -5 rpm/second (minus sign indicates deceleration, the higher the magnitude of the number the faster it slows down).  The stainless bearing with the steel shaft, has -20 to -22 rpm/second rundown acceleration going from 200 to 100 rpm as all my rundown tests use.

So a SS bearing is 4-5 times higher friction than is a ceramic and this is a killer for the teensy net torque this system might produce!

So be very careful in your bearing choice!

I subsequently replicated my steel shaft in brass and tried that.  Now the SS bearing gets into the 8 to 10 rpm/s deceleration from 200 to 100 rpm.  Still double the friction of the ceramic, but not as bad as with a steel shaft!

I will add that my full ceramic is  ABEC5, and the stainless one is ABEC1 rating.  This is a standard for bearing fit and tolerance - the higher the ABEC number the closer tolerance and smoother and truer surfaces of the balls and races.

So the SS bearing has a looser tolerance.  Now I surmise the reason this SS bearing performs so poorly in the magnetic field, is the ball retainer, or cage, is also stainless steel and magnetic.  This makes clamps or brakes on each of the balls in a high magnetic field!

I just looked closely at FM's photos and while he has a steel or stainless ball bearing, his retainer is brass or bronze, which is not magnetic as the stainless retainer on my second bearing is.  So this clamping or braking effect I mention won't happen with brass or bronze as retainer!

So pick your bearings carefully!  get non magnetic retainers (choices are brass, bronze or some plastics!)

However on a different note from observations of the stainless bearing and the brass shaft - the darn stainless bearing behaves more favorably in participating in the gain mechanism!  I am not sure why yet, but it can get better delta's for now.  Even though it's absolute friction is way higher, it behaves better.

I don't know if this is because of the fit of the balls and races (the full ceramic has higher tolerances being ABEC5 instead of the ABEC1 of the stainless bearing), or if it is related to the number and size ratios of the balls vs races.  The two are different.  The full ceramic has 8 or 9 small balls, and the stainless only has 5 or 6 balls larger in size.  Both have the same ID and OD, so it's only the balls that are different in size, but this matters in several ways it appears.

And in the vein of updating the "hunt" progress for getting more gain...  I modified some other factors when I made the brass shaft and this enhanced the gain.  I got one set of conditions to improve to about a 30% delta between FWD and REV, compared to only about 10% in previous reports.

And I got one case of getting really close!  You know when you have a small gas engine that won't start?  Like a lawnmower that's been sitting for too long without being run.  You pull and pull and pull and try the choke etc, remove spark plug and clean it, and check for spark ....  Then all of a sudden you do something and it sputters and almost catches on the next pull?

That happened to me yesterday, with the stainless bearing in place and the new shaft system changes  I spun it up by hand after hitting the record button on the data acquisition system for speed sensing.  And the beastie took about 3 times longer to run down from 200 to 100 than any previous attempt with the lousy bearing!  Almost as long a run down as with the ceramic bearing and no rotor magnets!  There were some randomized "boost" humps in the speed rundown, not unlike if that engine is firing every 3rd or 4th cycle, but at random number of dead cycles in between!

I was so excited I went and wrote over that recording of this anomalous test run.  And of course I was not able to get back there (Murphy's Law has a special subset for this work ).....  But that's OK, it means I am getting close to the right set of sweet spots!  But said Murphy's Laws for O/U might result in not seeing that happen again for some time, or it might get sorted out next week - this anomaly is good, but not certain of early success.  (I'd rather count on repeatable increments to the small gain than bet everything on finding that happy accident on demand)

And there are some kind of resonant elements involved in this working - that is almost certain from the bench observations.  A "tuning" of sorts has to be just right for this to "catch" like that dead lawnmower example.

Some of that is mass magnitude and distribution in the rotor body, including the magnets, and it's also related to the bearing, shaft and stator mount hole for that shaft - each of which has several effects on all the others.

I know I've got too much mass in my rotor, expected that.  But it is not difficult to machine off some mass so the magnets and outer rim form a flange and material is removed between this flange and the center bearing journal.

Also my cg is in the wrong place in the vertical sense....  I have underslung the rotor mass from the bearing's horizontal plane - and FM has his mass slightly above the bearing plane.

DMBoss