If one tinsy crack developes the rust will make more and more.

Dan is in a dry area.  I am not, but I don't have any salt air.

Lots of PMAs in my very damp, sometimes weeping or flowing, cellar. Many with uncoated magnets or partially peeled chrome hard disk magnets.
They were glued down with slow cure, water proof, grocery store purchased, 2 part epoxy, then heavily coated with left over epoxy, something like painted with paste.

Most were not subjected to rapid or extreme temperature differentials.  Some were, but not many cycles.
Most are induction conversions or similar.
I put a lot of effort into totally coating the magnets as they were placed.
The worst ones have been down there for over 3 years.
No visible rust as far as I can tell.

A couple have epoxy separating from the rotor ends, but I suspect it due to improper rotor surface preparation.
Some without protection, or damaged protection, are not usable. And I believe they have not been in "The Dungeon", as my wife calls it, as long as a few of the older PMAs.

These are pretty typical for what I do.  A necessary evil to understand what I am saying.
It is offensive. Sorry.  :/
http://www.otherpower.com/images/scimages/2050/3phHD.jpg
http://www.otherpower.com/images/scimages/2050/FanArm.jpg

Hugh, look for patterns and see if the patterns lead you anywhere.
Were magnets on both rotors damaged, or was the damage limited to only one magnet rotor?
Did the damaged magnets all have the same pole facing the stator?

I would also try passing a current through the stator.  Lay a magnet on the stator, and slide the magnet around on the stator, while holding a multimeter lead on the magnet.  That way, you can check to see if the stator is arcing anywhere.  You may want to wet down the stator to see if that affects anything.

ok, my thoughts since you asked  :)

first, i see on every magnet, radial scrub marks. i think eventually all the magnets will fail the same way. with the unit assemble with only one magnet plate on, i suggest rigging a dial indicator to determine the trueness of the mag plate. very slow rotation by hand will allow you to see just how much the plate runs out. also, with loaded torque, it may be bending more and allowing mags to scrub the surface of the stator. a possible larger air gap may be needed to ensure no rubbing during operation. a few extra turns in the coils may help offset the drop in voltage from an increased air gap.

the growling that you hear may be typical of strong electrical noise associated with having many many poles in the machine. the growl does not affect performance, but just makes noise. a possible cure for this, as is commonly done in high pole induction motors that also share the same problem, is to connect each phase "skip pole".

normaly in the phase, one coil is connected to the next coil (adjacent pole) and your jumpers from coil to coil are connected from an end of one coil to the beginning of the next coil.

in skip pole, you do not connect to the next coil in the phase, but skip that coil and connect to the next one. your coils then connect from the end of the coil, to the end of the coil, and from the beginning of the coil, to the beginning of the coil.

this method completely eliminates the growling in high pole induction motors, but am not sure if it will cure the noise in a coreless machine.  just thought i would mention it as a possible cause and cure.

Electroysis I do not think so.  As I understand electrolysis this is something that occurs under water when you have two different metals touching - such as copper and steel.  The baser one of the two metals corrodes.  In this case yes, the steel disks are corroding, but not seriously, and the real problem I have is with the cu-ni-cu plating on the magnets becoming ruptured, letting in moisture that ruins the magnets.

Rubbing yes I do think this is very noticable on this alternator and it brings me back to the thing about such a lot of swarf around the edges of the magnets too.  These are factors I do not usually see, or only to a small extent.  Maybe this machine was dirtier than usual.  The bearings do seem a bit loose and some rubbing is certainly possible.  Some of those marks will have been made during assembly, and the air gap is my usual (about 1.5 - 2 mm is my minimum gap).  The stator does show some signs of abrasion too, but on one side it has not reached the copper nor was there any sign of arcing on that side.

At first I only noticed the problem on one side, and the stator was by then significantly damaged in one spot so that the copper was bare and looked like it had been sparking.  I did think for a while that the sparking was the cause of the problem.  However when we had fixed all that and started to put it back together was when I noticed bulges on the other rotor - the back one.  Here there was no visible damage to the polyester coating other than some superficial rubbing marks but four of the magnets were starting to bulge noticably.  It's like the damage starts from within and moves outward.  But it does not start at the steel disk - always on an outer edge of a magnet block.

My concern is that I still do not understand what damaged the plating on the edges of some of the magnets.  Not direct impact for sure (unless it occurred during assembly before the polyester covered them).  Maybe some sort of arcing via swarf build up at the magnet edges.  But I am surprised that the stator did not show more blackening if this were the case.  On one side the stator looks quite normal.  Some scuffing, sure but nothing to indicate arcing and sparking.

I like to embed magnets in resin and glass fibre for a couple of reasons.  One is to protect against scuffing and knocks during assembly and when stuff gets into the air gap.  In this case it seems to have failed.  Another reason is to securely support the magnets agains centrifugal force. Maybe this is unnecessary paranoia.  If I were to use epoxy then a third reason would be to keep moisture away from the magnets too.  Maybe as Dan says I should just use thicker magnets and a wider air gap and skip the covering of resin - just glue them to the disk.  On that subject, by the way I believe that the best adhesive for this is BONDMASTER F246 (a surface-activated structural acrylic adhesive).  I would use stuff like this except that I am drawn to low tech solutions where possible.  Maybe this is nonsense when using Neo magnets since they are already pretty high tech, but I like to use materials that are easy to find all over the world when I can.

Thanks for all the discussion and I hope to hear from others in due course if anyone has more ideas.  Flux does not seem to be out of bed yet?

The corner plating failing might be attributed to disimilar heat expansion-contraction cycles, where once the bond has failed on the largest flat planes (helped by moisture intrusion?) the resin encapsulent heave literally lifts off a microflake of the corner plating that has remained well bonded - the (or all) corners having been minutely damaged during batch process handling on an automated assembly or counting machine, etc. and so vulnerable to our encapsulent shifting.

So - aside from using OEM epoxy coated neos to begin with - how about armoring the nickel-plate with a epoxy coat "dip" before encapsulating?

I make reference to it cracking here

It is too thin and when we set clearences in the shed all is OK. But when we place the blades on the unit as it is on the tower to do a final tracking check, 90 degree different it rubs, but up in the air all is OK

So we will be lowering it to reinforcing this plate, is this a possible cause of your problem.

 It seems to me that this problem comes from a series of elements.
  First, the concentric scratches that Dan, Flux and others have noted, appear not to be the root of the failure..if so, the rust/corrosion would have initially manifested itself at those scrape areas, not at the outermost ends as it appears to be happening. The scrapes still look ok, right?     As a side note, could those scrapes be caused by the stator "growing" with heat expansion under heavy generating load, and closing your 1.5 to 2mm gap?

 Second, and since I only have one image to make this observation, this may be completely off base, but the photo shows most of the affected magnets that were cut out, on one concentrated area of the rotor...does this area correspond to a heavy spot ie. when this machine comes to rest in no wind (does that ever happen in Scoraig?), would the worst corroded area of the rotors be pointing towards the ground, effectivley making this the "moisture dripping low spot? The last area to get dry?

 Third,as everyone else has noted about polyester resin not being waterproof, it will also almost always contain moisture even after curing. Its a sponge. Now combine the sponge effect, add a little salt spray (you have that) which gets absorbed into your sponge, throw in a little nickel, chromium, and mild steel, some expansion and contraction from heat...and voi-la, you've got a perfect incubator for corrosion.

 I used to take care of an old Formula 1 car from the sixties, and many parts were nickel plated chromium steel. It would rust right in front of your eyes practically. I spent many, many hours cleaning corrosion from those parts, over and over again. And I didn't have the problem of a sponge holding salt water against those parts.   Nickel plating is porous!

 I finally got smart about it. I had them powdercoated. No more corrosion!

 Take this with a grain of salt. Like I said, I'm very new to this. I've never even held a Neo in my hand!  I'm especially grateful to even be able to post here, with all of the brilliant minds. Thank you all for your expertise, and sharing that knowledge!

disaray1

Although I cannot comment on whats going on with the mags, I can comment on the rotors and the possible scrapes and growling.

I am in the process of building a large rotor machine myself, One of the things that I found when testing or setting up the rotors was that the pull of the high powered neos was so great, that it would actually bow the half inch steel a few thousandths. Then I took another look at your web site on the nirvana and it looked like you were using jacking screws closer to the outer part of the rotors.

When testing a test coil, I found that by putting a load on the coil (just for the fun of it), armed with two dial indicators one on either side of rotors directly across from the coil it still would cause the rotors to move a few thousandths in and out.

Then I tried the jacking screws only drill for one of the rotors, it was much better
as far as movement, but could hear a slight noise. so I drilled both rotors for the screws, and now have no movement or noise at all. I think the screws were vibrating on the back rotor.

Just a thought.

All the magnets I just bought have burrs on them that were left by the grinding belt that was used to champher the blocks when they came off the surface grinder.... I cannot see any signs that these blocks were put in a deburring machine before they were plated.

Could it be possible that some of the tiny knife edged burrs are being broken off and leaving holes in the plating?

Could it be that the salt water that is finding its way in the pocket with the magnet, is accelerating the oxidation process?

It would not be the first wind turbine that might have been visited by our feathered friends ...