Author Topic: 3HP conversion (Read 5111 times)

dinges · « **on:** June 18, 2006, 01:14:56 AM »

Dear diary.

I checked whether the bearings in the 3HP would be big enough to handle the axial load, and they are, barely. So they will be replaced with contact angle bearings, which can handle much greater axial loads (in one direction). The original 6206 can handle about 1000N axially dynamically, whilst the 7206 should be triple that. The dimensions are exactly the same. (check e.g. www.skf.com)

Did a lot of thinking on what magnets to use and how much to skew. Will go with many smaller magnets (2x6 magnets per pole; 4 poles; total 48 magnets of 15x8mm; this gives me an output, according to Zubbly's law, of 620W; not much, compared to the original 2.2kW of the motor). But still a lot of power. The motor weighs 35kg at the moment, probably the same after conversion. A lot of weight, IMO. More than an axial flux. Then again, it is quite waterproof (enclosed) and very solid, thus strong.

Skewing angle of the rotor is a lot more complicated than explained elsewhere. Basically one should copy the same angle from the original rotor, and you should be fine. If you simply skew according to the rule '360deg/#slots', i.e. 10 deg in my case, one should realize that a rotation of 10deg in one plane can NOT be simply transferred to another plane. In my particular case, 10deg rotation (one stator rib) translates to 3.5 degrees on the rotor. Check out my previous diary post for an explanation.

Note, however, that all this is still theory. Hope I'm not mistaken, don't want to spend a lot of work & money to end up with something that coggs terribly. Don't think it will, but still. Only real results count. The safest bet would be to skew more, like Zubbly, but this leads to voltage cancellation in the coils.

The magnets will be loctited in place, though the attraction to the steel core of the rotor should be enough to keep them in place. It probably will be very hard to extract a magnet after a try-fitting in its hole...

Below is the current design for the rotor; nothing is final yet, so there may be changes.

Next step: ordering the magnets and building of the rotor. Drawings are more or less finished, only have to make one calculation (of rotor to axle), the amount of negative play to end up with a strong enough crimp.

I'm having fun!

WXYZCIENCE · « **Reply #1 on:** June 17, 2006, 08:38:54 PM »

Peter, I agree with changing up the bearings on your 3 hp. Better to feel right about it from the start. I will start again on my 5hp once I get a little more experience.(Watching intently

) I have switched to a 1&1/2 and have skewed the stator. Working on the rotor now. A mini of the one Zubbly is working on. Joe

dinges · « **Reply #2 on:** June 17, 2006, 10:43:24 PM »

Another try w.r.t. skewing the magnets.

In the picture below we see the rotor; the blue lines are a projection of the stator ribs/slots. As can be easily seen, the left-most magnet is over an adjacent stator rib from the right-most magnet.

And, what's more strange, at first sight: the magnets are not skewed by 10 degrees but by 3.5 degrees !!

BTW, thanks Sparweb, for the reply to my initial diary. It confirms what I was thinking. I hope Zubbly understands my explanation. I'm curious whether, when he talks of an angle of 10 deg. skewing, he skews 10deg where I skew 3.5 deg; because, strange as it may seem at first, in my drawing below, the magnets are skewed by 10 degrees and by 3.5 degrees, depending on the projection of the rotor! Maybe I've misunderstood Zubbly here, but I don't think this point can be made too clear for other readers on this board.

So, Zubbly, please (dis)confirm: the angle in my drawing that is 3.5 deg, do you make THAT ONE 10 degrees? Because if so, then your magnets would not span 1 but several stator slots.

Also check this reply by Sparweb:

http://www.fieldlines.com/comments/2006/6/15/19126/8190/46#46

and this one by me:

http://www.fieldlines.com/comments/2006/6/15/19126/8190/44#44

Hope these drawings clarify matters.

Sorry if I keep going on w.r.t. this issue, but I think it's important to be sure we understand eachother.

Flux · « **Reply #3 on:** June 18, 2006, 12:36:09 AM »

Peter

As far as I can see, the skew angle depends on the number of slots and the core length, you seem to have it right for your case.

I can only comment on some alternators I made years ago with Alcomax magnets and with steel pole shoes. These had very defined edges compared to neos and it is obvious that with neo things are not as critical.

With the steel poles, without skew you needed a pipe wrench to turn it and it turned like a stepper motor. Any skew helped to some extent, but there was one angle where cog virtually vanished, it was very critical and was not the theoretical angle, but close.

I think with neo the field has "soft" edges. The worst case I think will be with curved neos that fit the stator curvature. With block magnets or lots of smaller ones the field edge is not well defined and in some cases it may even work with any value of skew angle.

Once again it seems that hours of mathematical analysis will at best only give you a starting point, but with your magnets it is likely to be good enough.

I proved these angles with the steel poles by skewing the core pack, which was far easier than making lots of pole shoes with different skew angle.

Again the issue with neo may not be the same as with steel pole shoes, but as long as cog was not severe I never found it to be the issue that determined start up.

With even a fair bit of cog, the thing would rock itself off the cog point and start rotating in modest wind. It took a lot more wind to overcome the iron loss and pull out of stall and come up to speed.

Flux

vawtman · « **Reply #4 on:** June 18, 2006, 06:23:56 AM »

Peter how does this magnet configuration match up with the coil span?

If you skew the mags and its to narrow and doesnt cut in and out itll probably cog pretty bad.

Go to Zubblys anotherpower photo gallery i think its page 7 and hell show you the relationship of the two.

Hang in there

zubbly · « **Reply #5 on:** June 18, 2006, 09:16:04 AM »

hi again Peter,

it boils down to this. the drop string check i have mentioned should land one slot over. you can view it as 10 degrees or 3.5 degrees, or any way you wish, as long as the string check is observed.

a few other things can possibly come into play. the width of the mags you use and how they cross the stator tooth barrier related to their width. in reality, you could be off a degree or 2 either way and still work out fine. but i just suggest to shoot for the one stator tooth offset as a standard starting point. this is "not" an exact science as there are way too many variables that can come into play. i am just trying to keep instructions as simple as possible. if you wish to take analysis to the extreme thats great as some may appreciate a more in depth explanation to relate to.

just one other thing i wish to mention. when plotting the exact location and centre distances between all the mags in one pole, think of it as one magnet. keep the mags as close together as possible but with still maintaining structural integrity. assuming you use multiple round mags and depending on what material you make a magnet cage from, on aluminum i try to keep a wall thickness between mags (at the turned down rotor surface) to approx 1 mm of aluminum between mags. you may find thatincreasing it a bit may be better to help form the ideal mag pole width. this is part of the reason why i sometimes use 1/2 inch diam mags or 3/4 inch wide mags.

zubbly

dinges · « **Reply #6 on:** June 18, 2006, 10:06:50 AM »

Thanks Zubbly,

In that case I think I misunderstood you. You do skew one slot, not 10 degree; then I made the 'error' of taking an angle in one plane and transferring it to another

It's just that in your pictures it looked like you skewed very much more than the 10 degrees/one slot.

As to the width (and the volume) I have limited choice of suitable magnets, 12.7x6 and 15x8. I'd rather have 12.7x8 or 12.7x10 (your 1/2"x3/8") for more magnetic volume, but alas. Also because I'd rather have less spacing between the magnets and a bit more span on the rotor. It's just that it doesn't work out very well with these magnet sizes. Will check if I can fit 20x10 magnets in, this may be the last option.

As to the span of the coils: I don't think it's 12-10-8 wound; at least, inside the coil there are 5 free slots, not the 6... Strange. Are there other systems of winding too than 12-10-8? If so, I hope it won't have an effect on converting this motor.

Glad the misunderstanding is solved

Peter.

Countryboy · « **Reply #7 on:** June 19, 2006, 08:18:49 AM »

Hello Dinges,

I think I know what is confusing you.

One stator tooth creates a skew angle of 10 degrees. The line of that angle has no width. We see that angle as a 2 dimensional image.

Magnets have width, which adds a 3rd dimension.

Try to imagine your placement of 12 magnets as one big magnet. In your drawing above, your 3.5 degree angle does not measure to the top of the magnet.

Measure the angle of skew by measuring to the magnet in the upper right hand corner. That should be closer to the 10 degrees of skew.

When you measure the angle, measure to the center points of each individual round magnet.

Does this make sense, or do I need to explain further? (Your picture showed 13 round magnets, with 7 in the upper row. I scribbled out the 13th magnet for clarity.)

Countryboy · « **Reply #8 on:** June 19, 2006, 08:27:26 AM »

Try to think of the group of 12 little magnets as one big magnet. Measure skew of the magnet group as a whole. You were measuring skew by thinking of one row of little magnets as the whole magnet. Having 2 rows of little magnets changes the angle you need to measure. You need to measure to the top row of little magnets to get an accurate skew measurement.

SparWeb · « **Reply #9 on:** June 19, 2006, 12:48:41 PM »

Oof - angular contact bearings? Don't you have to precisely pre-load those when you mount them? Usually installed in pairs, too - do you have the space?

I thought you were going to make your own mounting plate for the bearing - would that allow you the extra 10mm you need for a 6306 bearing?

dinges · « **Reply #10 on:** June 19, 2006, 01:26:57 PM »

Are we talking about the same bearings? The angular contact ones that I mean have exactly the same physical size as the 6206. They are normally used in pairs, but since I will only have to take up axial force in one direction, I think one would be enough. Just checked, and apparently they need a minimal axial load. Hmm... Hopefully any windpressure from the prop will provide that load. Will have to check though.

Thanks for the heads-up. I was basing my plan on a text-book (Roloff-Matek), and the angular contact bearings they have there look more like a combination of a normal roller bearing (one one side) and an angular contact bearing on the other side. In other words, the ang.contact bearings from Roloff-Matek seem to be able to handle (small) axial loads in the reverse direction too (they operate in that mode like a normal ball-bearing). The SKF catalog however, states differently...

If worst comes to worst, the 6206 should be able to handle the axial load too.

Decisions decisions...

dinges · « **Reply #11 on:** June 19, 2006, 01:39:59 PM »

Just checked; at 2000RPM, this bearing should see a 22N minimum axial load. Think that at this RPM the wind will provide that load

At lower RPMs this required minimum load becomes less. Also, when the axial load is lower than this minimum value, load bearing capacity of the bearing is reduced. However, when no axial force is present (no wind), then the bearing has no load to bear; it's basically freewheeling with little or no load, so who cares that the max. load bearing capacity is reduced in this situation?

But, for ease of mind, I think I'll install a flat spring (the kind you sometimes see with bearings)under the 'normal' 6206 ball bearing to provide some minimal axial load.

If I'm missing something here, however, don't hesitate to point it out to me! I'm not a bearing expert. And thanks for the head-up!

Peter.

dinges · « **Reply #12 on:** June 19, 2006, 03:00:34 PM »

Here's the proposed new situation w.r.t. bearings. On the backside is a 7206 contact angle bearing. On the right a plain ball bearing, 6206, with a cup spring mounted to provide the small required axial load for the contact angle bearing. The 7206 is supposed to take full axial load. The retainer ring, which used to be mounted on the front, is deleted. It's no longer needed since the 7206 won't be mounted in the front scale but in the back scale.

Feel free to have at it, I'm open for criticism.

dinges · « **Reply #13 on:** June 19, 2006, 03:10:22 PM »

Thanks for the reply.

You were right, that leftmost magnet shouldn't be there, had added it as an aid while drawing. Because the leftmost and rightmost magnet should not be exactly at the same position (but over different stator slots); only if you add an EXTRA magnet, than that magnet should have the same position as the first one. If the first & last magnet occupy the same position (over a different slot) one would still have slight cogging, I think. But maybe I'm nitpicking

As far as angle goes, I disagree; see rest of my explanations. However, I think that the full width of the magnets should fit between the two coil legs. So, if you measure the width of the coil (say, 50mm), then the skewed magnets (full span, tip to tip) should fit inside, I think. But again, think I'm nitpicking here.

To be honest, like with axial fluxes, you have to really screw up to end up with something that does not generate.

Peter.

RP · « **Reply #14 on:** June 19, 2006, 08:20:53 PM »

By "flat spring" do you mean a belleville washer or wavy washer?

SparWeb · « **Reply #15 on:** June 20, 2006, 04:47:48 PM »

Can you get angular contact bearings that are sealed? If not, how will lubricant stay inside? How will water or dirt stay out? Isn't this type of bearing more expensive than radial ball bearings?

I would still consider using a 6306 radial ball bearing on the front face, since you are re-mounting a bearing anyway, but I won't nag on the topic again.

I'll report my experiences with bearings, in case it is helpful.

When building a VAWT, I designed it to be cantilevered at the base. The VAWT was 1.2 meters tall, the spacing of the bearings at the base was about 90 cm. I used tapered cone bearings so that I could very finely adjust the pre-loading on the bearings. Too little, and the axle could jiggle up-and-down. Too much, and the bearing friction rose drastically. The VAWT was sensitive to this friction (being a poor starter) so I was happy with this arrangement. Pre-loading made a big difference in running torque in the bearings. Keep in mind that the VAWT has a large axle, so the bearings are huge. Bearings that size in a radial bearing format would weigh 10 pounds each, and have a large running torque of their own, so I opted for a lighter type of bearing (tapered cone) and the friction was dramatically reduced. Trying to seal up the bearings would have been an enormous effort, so I resorted to using a "slinger" and changed the grease every few weeks. Fine for an experimental rig, but not for a permanent installation.

dinges · « **Reply #16 on:** June 20, 2006, 07:49:14 PM »

Thanks for the comments Steven. As usual there's a bit more to take into account than I had expected in the beginning. Feel free to 'nag' all you want, btw.

(I'd call it 'advising'; hey, if you're clever, call it 'consultancy'; that way you could send a huge bill

You're right, the 7206 (contact angle bearing) can't be had with shielding. At least not in the single-row version from SKF.

Just bought the bearings; the 7206 is triple the price of the 6206, but I think I'll switch back to the original configuration of 2x 6206. The 7206 will be returned and I'll get another 6206.

The main issue I have with that 7206 is keeping dirt out & grease in. Regular maintenance is slightly complicated with a windmill

The issue I have with the 6206 is that it isn't the best suited one for axial loads.

But the 6206 should be able to handle the load. I've been a bit conservative in several safety factors and the result was right at the edge of what the bearing (6206) could stand.

The way I see it now, that 7206 needs lots more pampering and maintenance than that 6206; in which case it might be wiser to simply replace the 6206 every few years rather than re-greasing the 7206 every few months. In both cases the turbine would have to be grounded.

Decisions decisions...

dinges · « **Reply #17 on:** June 20, 2006, 09:58:48 PM »

Haven't read it yet, but SKF seems to have a few articles about bearings in windturbines.

http://www.skf.com/portal/skf/home/solutions?lang=en&contentId=288086&category=appreq&la
stContentId=287721

dinges · « **Reply #18 on:** June 20, 2006, 10:08:26 PM »

Hmmm... Apparently FAG has shielded contact angle bearings. The 7206 they offer is available with double seal. It's different from SKF's 7206 in that it is axially loadable in two directions (though one direction can take much more load).

It seems better suited for my application than SKF's 7206. Interesting. Will check whether I can get FAG's 7206. If not, it's going to be the 6206.

http://medias.ina.de/medias/en!hp.ec.br.pr/72..-B-2RS_FAG*7206-B-2RS-TVP_FAG;aS7l6cqhPXyf

Everything you ever wanted to know about bearings. But were afraid to ask.

Peter.

dinges · « **Reply #19 on:** June 20, 2006, 10:59:36 PM »

Yeah, I think I'm worrying too much. Just checked the 6206 bearing again, but this time with an online calculator from FAG. With a 3m prop and furling at 15m/s, the axial load will be maximum 1000N. I've calculated with 2000N axial load and 1500RPM, for 100% loadscheme. In that case, the bearing would survive 4800 hours (Lh10). 10% of bearings will go defect at 4800 hrs. Expected lifetime is 75.000 hours.

And this genny is never going to see 2000N load & 1500 RPM continuously for those 4800/75000 hours.

Guess you guys were right. Glad to have been able to confirm it with some calculations

http://medias.ina.de/medias/calc/data/output/1150865799328.htm

I admit it, I'm overcautious. Blame it on Murphy

Peter.

dinges · « **Reply #20 on:** June 20, 2006, 11:29:48 PM »

Not sure. In Dutch it's called a 'golfring' (wavy ring).

dinges · « **Reply #21 on:** June 21, 2006, 12:48:49 PM »

Zubbly, I'd be interested in your opinion once more...

I intend to mount 56 magnets of 15x8mm; this gives a total of 4.8 cubic inch --> about 720W power (if 150W/cubic inch holds true for me). However, since this is a 2.2kW machine, this seems little power, to me. What do you think?

And I'm wondering: should I fill all empty space on the rotor with magnets (like you used to do in your first conversions), or should I go with my above arrangement, with spacing between the poles? Esp. if the poles are to fit 'inside' a coil. Or do you get cancellation inside a

Something I've noticed, judging from the pictures: rotordiameter in my motor seems to be smaller than yours: the diameter of your 1.5/2HP rotor is about the same as my 3HP... Probably has to do with different voltage in US vs. Europe, I guess.

Smaller rotordiameter means less space for magnets too...

Peter.

dinges · « **Reply #22 on:** July 09, 2006, 03:53:25 PM »

This evening some work has been done (finally) on the rotor for the 3HP conversion. A totally new steel rotor has been made (out of one piece of C45; old axle is discarded). An aluminium sleeve had been machined. It was Loctited in place. However, .02mm positive play (on diameter...) is too little for Loctite. As if we shouldn't have known. Not even 40.000kg (40 ton) of pressure could get the last 28mm (abt. 1") on. So, excess aluminium sleeve has been cut off. A new ring will have to be machined for the other end and Loctited in place.

Too bad it will have to be patched up, when it could have been perfect.

When that's done, pockets for the magnets can be machined and Loctited in place. That'll be for another time though.

Pictures are below. I'll let them speak for themselves, without too much commentary. Have been reduced to 40% kB-size, but will resize next time to 20%. Not much detail seems to get lost.

The pix:

News:

Author Topic: 3HP conversion (Read 5111 times)