Bicycle Generator Project

[TomW]

Mr. G- or anyone else,
 Could you please help this noob and elaborate a little more on this statement.

"Side note:  The eBay sellers who brag about how many pipe wrenches the OUTSIDE of their PM DC motor can hold up obviously do not know what they are talking about.
G-"

I think I understand, but I have been wrong before

Well, you pay all that money for the flux (magnets)

If ANY flux is exiting the frame / case / motor then that flux is wasted.

It should be directing that flux through the coils for efficient use of the magnetic material.

Just from here.

Tom

[klsmurf]

Thanks Tom, That's what I thought. I'm still happy with it for what I paid. (Bid not buy-it-now) It is all good fun!

[jaysicle]

Because it is already built, physically check to see if any substantial flux is leaking!

See if a large paper clip will hang on the back of the rotor.
If it will not, or just barely holds it up, then there is no flux leakage worth considering.
If a small screwdiver will not stick to the back, not sure there is enough flux leakage to worry about.
G-

Thanks for going to that length to show this.

I did give the paper clip a shot a couple of weeks ago, after bringing the idea up with a friend. 
First a fine toothed jigsaw blade, then a paper clip actually. Both fell off when placed on the back of the assembly.

But because of your response, I just tried the clip in the open ends. There appears to be some leakage at those open sides.
With a clip balanced on the stator edge, about 2" away from magnets it dances quite a bit.

Is that more of a system effect ? I guess in order to find out I'd remove the stator, and hang the clip from a string or balance it on some small thing.

I think that of the 30% likelyhood of trying this bar idea on existing build, it would have to be 100% on the inside, not to effect things too much. Then I could angle the metal bars (fixed by each pole) and create an air draw through the inside, and over the stator, out the open 'ends' - like a poorly designed blower in a car.

That advert with the wrenches does seem odd.

Thanks again G-,
Thanks for having a look TomW,
klsmurf - Congratulations on your build!

[jaysicle]

Update...

Just tried that again - a little more carefully this time.

During the earlier attempt, I made sure (visually) that the clip was not interfering with the rotor, by spinning it slowly around one rotation.

In this second attempt, there was no clip movement. It must have been the untrimmed suface tension lip of the epoxy on the rotors silently rubbing the clip every now and then in the first attempt.

I wil try attempts 3-5 later - but believe that I will have a repeat and similar result to report.
The bar may be on its way down to flywheel weight and cooling effect as gains.

[jaysicle]

A little more resolution...

At the ~1" distance that the outer edge of my rotors are from the outer edges of my magnets, there appears to be no hold to paperclip which is simultaneously subject to gravity, when rested on the stator edge. Slightly different test - I know, but a little more valid than what I was doing, I think. (Not more valid than ghurds posted suggestion of clip on rotor back, but more valid than my last attempts in balancing that clip on the topish of the stator and watching for any kind of pull).

At that same distance, there is also little control of an 'intentional swing' applied to the paper clip when suspended by electronics pliars.

However, when using the same 'intentional swing' method (clip shaken very lightly by hand starting 2" clear progressively drawn closer) as I get closer to the magnet face, the frequency of the swing increases and the amplitude decreases (becomes tighter).  OK - well duh... eventually it rips it off of the pliers and the magnet swallows the clip whole...

But, it seems that I could go about 1/4" in and get something. Trouble is that it might not be enough of something, i.e: enough to tame a shaking paper clip... but not enough to increase the field force in the gap to melt my stator as I ride and see 100% efficiency.
Also, I can only get so close either inside or outside before coils interfere.

I suppose a Gauss meter would tell... (I am interested to find out what the gap strength is as well actually, as this may help keep FEMM honest.  )

[SparWeb]

Hi Jay,
I'm so busy (Life!) that I'm not giving this thread the attention I think it deserves.  Sorry.

I'm sure a box of Timbits will keep in the mail if sent expresspost - but not this week!  The lettercarriers are about to go on strike.

I think you're getting plenty of ideas from the experimenting and things you're trying - it teaches just as much as the math and computer stuff.
The mistake of mine was, I admit, both a waste of time and of money.  I built an axial flux, and instead of making it big enough to have jacking nuts on the plates to keep them apart, I had laser-cut a set of rings to stack between the plates.  So I theorized that they would provide the path for flux that you're talking about, only around the center, not the outside.  They didn't do a damn bit of good.

Another thing I thought of is that there's a forum a bit like this one, though with more crazy inventors on it, based in Australia called the Backshed, and there you can find a few axial-flux alternator projects.  Those guys are copying each other a bit so there are several of these "AxFx" gennies, and they are all fully enclosed.  The upshot is that the magnet rotors have a ring going all around the edges, just like the bridge you've put on your FEMM drawings.

BUT

That's not why they put it there (I don't think) and even though they do it, I doubt they get much benefit in the way you propose.

Basically, if the rotor backing plates that are used are of sufficient thickness, that they can pass 100% of the magnetic flux from pole to pole, then there is nothing to be gained by adding another part.  Actually that's a whopping over-simplification.  Ghurd demonstrated it best by showing the improvement you get just by putting ANY backing behind the magnet.  He's also made a better point and it shows that when there is enough backing plate, the returns diminish by thickening it any more.

What you're really trying to increase is the total flux across the gap.  Nothing else really matters.  Are you using the FEMM line-integral?  I see a few representations of coils in your attached graphics, but not the characteristice red line that shows an integration.  When I use the integral, it spits out a nice graph of Teslas crossing the line, and the "sum" spits out the total flux in Webers.  I should mention it's in the "FEMM-View" post-processor.

Yes/No/Whadda you talking about?

[jaysicle]

Thank-you for the tip about the line-integrals. I got more or less the same (0.429T) vs. 0.434T as at test point.
FEMM plots/ graph views are interesting as well in that regard - again, Thank-you.

I appreciate that you have a lot on your plate, and that taking (any) time to post on my project / hypothesis is kind.

Here are the meter numbers....


Transverse Probe...

 



Transverse Probe Mid Air away from machine... and kind of out of focus...





At rotor back (1% stray)





At rotor edge (1% stray)




Stray rotor edge at close distance (2.1% stray) (bloody close to coils... oh well.... only 150 Gauss)





Actual gap flux density... 6900 to 7200 Gauss, good stuff.

[SparWeb]

Holy flux-lines batman,

0.7T is pretty high!  How does that compare to the model?  Does the drawing that gives you 0.43T roughly compare to the meter giving you 0.7?
If not, I'd blame the model first.

Now the fun challenge is to find the problems with the model, fix them, and then confirm that the model is valid if it works for other tests and measurements!

How far do you want to go?  I only get vicarious jollies since you're the one with the cool instrument!   

[ghurd]

Holy flux-lines batman,

I only get vicarious jollies since you're the one with the cool instrument!   

Looks like the paperclip test is outdated!
G-

[jaysicle]

Holy flux-lines batman,

0.7T is pretty high!  How does that compare to the model?  Does the drawing that gives you 0.43T roughly compare to the meter giving you 0.7?
If not, I'd blame the model first.

Now the fun challenge is to find the problems with the model, fix them, and then confirm that the model is valid if it works for other tests and measurements!

How far do you want to go?  I only get vicarious jollies since you're the one with the cool instrument!    

Yeah, it is funny that you write that... Turned computer on to try this stuff out - I checked FL on the way...

My first (project) thought today was that I would have another look at the model with the 0.434 and check that it is accurate-ish. My next thought was that maybe looking at the 0.509 model that was drawn with the "rolled-out" view that you and RP have led me to try would be a better / more realistic place to start. Then the thoughts and 'next steps' kind of flooded in including re-building the stator without the plexi-glass, and simulating that after 'correcting' the current model.  

I just got off of the bike actually, watched a re-run of M*A*S*H during the ride. The whole time I was thinking about looking into the model vs. real difference. That and finding out what happens with the possible hepatitis outbreak starting with Father Mulcahy.  

[dave ames]

Nice meter!  

We have been trying to get a handle on those FEMM simulating tools for some time now and it always ends in tears ....I'm blaming the slow pc for the most part, but have noticed the bio version is not running as fast these days as either.

interesting,.. the calculated .43 & .7T meter reading difference..  could be 43,000 "flux lines" per square inch calculated with the FEMM program?  that seems to fit?

been having some comparison fun with this tool.
http://www.intemag.com/calculators/Flux_Density_Disc_Yoke.html

great stuff,
cheers, dave

Edit: a better link? http://www.intemag.com/glossary.html

[jaysicle]

Yes, I have found that even though my machine is pretty quick, if I turn off smoothing, and a couple of other things through preferences- the math seems to get resolved quicker and it just 'likes' that. I won't pretend to know how the .exe actually uses my machine to get what I am asking for done... but I offer this observation.
------------
I hope that the following is useful in a few ways - 1) to anyone new reading; 2) to the guys that have most recently posted; and 3) to me when I look back. No money was spent, but some of the forum guys did give thought and advice so there is a time component which I am very appreciative of/for, but I can't say (and hope it is not perceived) that it was wasted and here is why...
 
Have redrawn and I think I have the gap covered... (pun wasn't intended, but now that I've typed it...  )

Original views...
0.434T = Test Point at skin of imaginary coil in my original layout. (Single Magnet Pair with Steel backplates)
0.550T = Test point at mag face in my original layout. (Single Magnet Pair with Steel backplates)

Redraw here,

0.545 = Test point skin of imaginary coil in "rolled out" version that RP and SparWeb imparted with mag pair on either side of test point
0.671 = Test point mag face in "rolled out" version that RP and SparWeb imparted with mag pair on either side of test point

I should explain that the last set of numbers and the redraw are after looking at my build a little closer. Please excuse this, I built the generator portion in 2006, and likely stuck with more rounded dimensions to recall in my brain... The plexi is more like 3/32" than 1/8", and the actual air gap on either side of that plexi is more like 3/32" than 1/8" as originally drawn. So I built a little better in reality than what was depicted even in the original 0.434T model that I drew for that reality... by 1/8" or so in those regards.

With SparWebs' Line Integrals... and the more accurate redraw - FEMM out across magnet face is...
Average B.n = 0.648629 Tesla

This matches the average value of gap measurements that I took with the meter. ~.670 or so... Because I had (and fully intended to) set the meter on DC Peak, it likely saw some fringe effects while coming into position and ended up with 0.721T.  

None of this effects the 1:1 (apples to apples) of my original bar speculation, and the resultant meter readings confirming that there just isn't anything to be had in that way... but should be recorded just the same for accuracy.

[SparWeb]

Ain't is pretty when physics and reality agree with each other?
Have you already jumped to the next step, or are you working your way through Maxwell's equations just one at a time?   
You have everything you need to know now, to find the open-circuit voltage and cut-in speed of the alternator.

[jaysicle]

Ain't is pretty when physics and reality agree with each other?

Yes, they are both rare and special times.  

------

Did not enter open EMF numbers earlier in the post... just Power plots vs. RPM.
Now that it has come up, here they are...

EMF at usable 30km/hr rate IRP = ~22.5VDC
(22.5VDC EMF - 14VDC at Batt) = 8.5VDC
8.5VDC / 1.65ohm resistance = 72W

The EMF was 53VDC wired in Star, but the internal resistance was high, and I had to choose between 75W for 15 minutes at an uncomfortable ("uphill") 22km/hr cadance, or rewire in IRP and get ~72W for 45min at much more comfortable 28km/hr. (Thanks again G-) ....  

Empirical cut-in is about 300RPM when clamped to the battery bank.

--------

In terms of going forward... it would be beneficial to mathematically close the loop(s) - and continue to check them vs. empirical.

It will be downright necessary for the stator re-build (if I decide to go down that road), where I think I can squeeze a little under 1/4" out of the gap. That may require changes in wire gauge, number of windings and so on to get to the "feel" I need for a consistent ride, with equal or hopefully greater power output than existing build. Relationships would need to be fully understood to get that right. I have a feeling nothing is free and if I just tightened the gap by the available 25-30%, leaving all else the same - I would need to gear up (driven) some to balance that.
 

[jaysicle]

Update on ghurd controller improvements...

Have been meaning to get back to this for a bit now, and today I re-worked the dump load resistors in the contoller.

I've added some extra aluminum heat sink material under the dump load resistors, with a decent amount of heat sink compound between the case and the aluminum strip, and between the aluminum strip and each resistor. The 1.67ohm has been replaced with a 2.49ohm so that they are all the same value now.



I have tested the controller out again, dumping for about 30 seconds or so - and all of the resistors heat up at the same rate. They also feel cooler to the touch than in previous trials with roughly same dump duration.  Nice to get the heat shrink tubing on the diode runs as well to clean that up a bit.

[GoVertical]

Hi, I have search for the internal resistance of the stator you are using and have not been able to find it. What is the internal resistance of your stator?  I was able to run some basic tests on the PMA I have been working on and I have been using your test data as a comparison, it has really helped me better understand many of the relationships. Your project has been very helpful and inspiring. Great work.       

[jaysicle]

Hi GoVertical,

They are 20g coils, 100turns each, 0.6ohms per coil, so 1.8ohms per phase, (eventually wired IRP).

I apologize for not getting back to you sooner. Thank-you for the kind words.
Your fabricating and drive to see this through is equally inspiring, - and I am not just blowing smoke.
You are obviously one solid spirit.

When you have a handle on the relationships, please share them if with us...
I think I have got NxFxfx2 cornered... we'll see though - I have been wrong before, and likely will be wrong again.