I was playing around with magnet arrangements and the following configuration occurred to me:
Usually cyclotrons are designed with one magnet with its S facing towards the dee chamber and another with its N facing, thus forming the required magnetic field across the gap. You'll need a big yoke to bend the whole field around so as to maximise the field (just as Doug C. showed us his work the other day).
In the device I'm struggling to build, my ions may come out of the magnetic field in this configuration but I want to recycle them back in, so I can't tolerate field irregularities past the convex 'edge' of the field.
It dawned on me that the yoke could actually be 'inverted' with respect to the magnets so that the whole of the yoke takes on the flux of the outer magnets and focuses it into the centre of the dees.
This can be done by putting the magnets around the OUTSIDE of the device rather across the centre, and the 'yoke' can now be used to carry the flux into the centre.
With respect to a cyclotron, it would work as in the following images I've prepared to explain this (no hope with words!!).
So now there is a continuous magnetic field that stays oriented in the same direction all the way out to the edge of the field but then doesn't invert yet further out (if compared with a normal poleface-to-poleface arrangement).
The result is a highly compact arrangement whose magnetic yoke can double up as the dees of the cyclotron!!
So that would be it! The dees carry the magnetic flux! There is no leakage through the gap of the dees at all, and little loss to the outside circuit. You could even seal the whole thing up along the gap with a dielectric seat that can hold a vacuum such that the dees a) provide the e-field, b) carry the flux, c) doesn't invert beyond the shoulder of the field AND d) forms the vacuum vessel aswell.
If you don't want to use permanent magnets, no worries, you can replace the magnets as I've shown them with electromagnets. If you vacuum-seal those 'shields' against the dees and seal the whole thing, now the electromagnets are on the outside (air or water cooled) but the flux is actually taken into the vacuum volume by the vacuum container itself.
I think that's pretty neat as it solves 4 packaging problems simultaneously! Has this configuration ever been used before? Will it work!?
best regards,
Chris MB.
A compact DIY cyclotron package solution.
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
A compact DIY cyclotron package solution.
- Attachments
-
-
-
Carl Willis
- Posts: 2841
- Joined: Thu Jul 26, 2001 7:33 pm
- Real name: Carl Willis
- Location: Albuquerque, New Mexico, USA
- Contact:
Re: A compact DIY cyclotron package solution.
Hi Chris,
I like the new approach suggested here. My gut tells me that it's unlikely to be appropriate for the high and uniform fields needed for most cyclotron projects even with modern NdFeB magnets, but it could readily be applied to very low-energy cyclotrons or to the design of microtrons (for electron acceleration).
This is an idea well worth carrying to the next level to see what it is capable of.
Some possible design challenges:
-A coating is needed on the dees that is a good RF conductor--copper, gold, etc rather than just magnet iron. They are the capacitor plates in a very high-Q, high power circuit, so lots of current can flow on them.
-Cooling of the dees to protect the magnets. Nd magnets can't exceed ~100C or they're shot. This design probably requires somewhat large dees (to handle a number of magnets and develop uniform flux in the acceleration gap), necessitating larger RF power for a given dee voltage than otherwise might be called for. So it will be important to consider the cooling needs and make sure the magnets are especially well cooled.
-Force between the dees. Assembly technique. As magnets are added, this can turn into a grenade. It has to be mechanically braced well enough that this does not happen.
Good idea.
-Carl
I like the new approach suggested here. My gut tells me that it's unlikely to be appropriate for the high and uniform fields needed for most cyclotron projects even with modern NdFeB magnets, but it could readily be applied to very low-energy cyclotrons or to the design of microtrons (for electron acceleration).
This is an idea well worth carrying to the next level to see what it is capable of.
Some possible design challenges:
-A coating is needed on the dees that is a good RF conductor--copper, gold, etc rather than just magnet iron. They are the capacitor plates in a very high-Q, high power circuit, so lots of current can flow on them.
-Cooling of the dees to protect the magnets. Nd magnets can't exceed ~100C or they're shot. This design probably requires somewhat large dees (to handle a number of magnets and develop uniform flux in the acceleration gap), necessitating larger RF power for a given dee voltage than otherwise might be called for. So it will be important to consider the cooling needs and make sure the magnets are especially well cooled.
-Force between the dees. Assembly technique. As magnets are added, this can turn into a grenade. It has to be mechanically braced well enough that this does not happen.
Good idea.
-Carl
-
Todd Massure
- Posts: 443
- Joined: Thu Mar 11, 2004 12:38 am
- Real name: Todd Massure
Re: A compact DIY cyclotron package solution.
Hi Chris,
Interesting idea, but my first thought is that the flux would tend to reside almost completely in the circuit set up by magnets and soft iron and wouldn't be forced to jump across the chamber like it does in the figure 8 yolk design with a gap.
I tend to think of the gap in the magnetic circuit like a spark gap in an electrical circuit. Your computer model seems to show flux in the chamber area though, so I'll study it some more.
Thanks for sharing.
Todd
Interesting idea, but my first thought is that the flux would tend to reside almost completely in the circuit set up by magnets and soft iron and wouldn't be forced to jump across the chamber like it does in the figure 8 yolk design with a gap.
I tend to think of the gap in the magnetic circuit like a spark gap in an electrical circuit. Your computer model seems to show flux in the chamber area though, so I'll study it some more.
Thanks for sharing.
Todd
-
Todd Massure
- Posts: 443
- Joined: Thu Mar 11, 2004 12:38 am
- Real name: Todd Massure
Re: A compact DIY cyclotron package solution.
Oh, ok, I get it now. Yeah, seems like it should work.
I've personally abandoned the idea of using the rare earth magnets, just because the cost doesn't seem to work out, oh and they seem hard to work with, oh and with large ones or an assembly of a bunch of them they kind of, ...ok really, scare me.
I might consider using a similar layout with electromagents. It would be nice to be able to find a bunch of off the shelf solenoids or something instead of getting coils custom wound. Plus all the extra surface area on the magnet coils would probably really help with cooling. Plus, who needs a couple of tons of yolk iron to deal with.
Great idea.
Todd
I've personally abandoned the idea of using the rare earth magnets, just because the cost doesn't seem to work out, oh and they seem hard to work with, oh and with large ones or an assembly of a bunch of them they kind of, ...ok really, scare me.
I might consider using a similar layout with electromagents. It would be nice to be able to find a bunch of off the shelf solenoids or something instead of getting coils custom wound. Plus all the extra surface area on the magnet coils would probably really help with cooling. Plus, who needs a couple of tons of yolk iron to deal with.
Great idea.
Todd
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
Thanks for the positive feedback.
Carl Willis wrote:
> -A coating is needed on the dees that is a good RF conductor--copper, gold, etc rather than just magnet iron. They are the capacitor plates in a very high-Q, high power circuit, so lots of current can flow on them.
My build differs from this actual arrangement, but this comment may still apply. Either way, a worth while comment if it gets taken on as a cyclotron design, as I've posted.
> -Cooling of the dees to protect the magnets. Nd magnets can't exceed ~100C or they're shot.
EH grades run to 200C, and I think there are higher grades now even. But that's by-and-by. The point is that the surfaces could now be external if the dees also form the chamber, so could be cooled externally which is a much easier challenge.
>This design probably requires somewhat large dees (to handle a number of magnets and >develop uniform flux in the acceleration gap)
I guess you might be right, though these 2D simulations (Maxwell SV/2D) didn't show up anything noteworth other than uniform fields. I anticipate some pole shape tuning could mostly overcome that concern if it presents itself.
> -Force between the dees. Assembly technique. As magnets are added, this can turn into a grenade. It has to be mechanically braced well enough that this does not happen.
I've always been worried about this (ref my other post on monster neos). I bought a large quantity of smaller magnets and have realised they are no so scary as I first thought. As long as you put them straight onto an unsaturated backplate they do tend to stay there. My initial concerns came from putting them on saturated plates, at which point they love to jump around. This is also true for any question on field uniformality; providing the flux paths don't saturate, all the flux tends to spread itself out in a fully uniform manner as it emerges into the gap.
best regards,
Chris MB.
Carl Willis wrote:
> -A coating is needed on the dees that is a good RF conductor--copper, gold, etc rather than just magnet iron. They are the capacitor plates in a very high-Q, high power circuit, so lots of current can flow on them.
My build differs from this actual arrangement, but this comment may still apply. Either way, a worth while comment if it gets taken on as a cyclotron design, as I've posted.
> -Cooling of the dees to protect the magnets. Nd magnets can't exceed ~100C or they're shot.
EH grades run to 200C, and I think there are higher grades now even. But that's by-and-by. The point is that the surfaces could now be external if the dees also form the chamber, so could be cooled externally which is a much easier challenge.
>This design probably requires somewhat large dees (to handle a number of magnets and >develop uniform flux in the acceleration gap)
I guess you might be right, though these 2D simulations (Maxwell SV/2D) didn't show up anything noteworth other than uniform fields. I anticipate some pole shape tuning could mostly overcome that concern if it presents itself.
> -Force between the dees. Assembly technique. As magnets are added, this can turn into a grenade. It has to be mechanically braced well enough that this does not happen.
I've always been worried about this (ref my other post on monster neos). I bought a large quantity of smaller magnets and have realised they are no so scary as I first thought. As long as you put them straight onto an unsaturated backplate they do tend to stay there. My initial concerns came from putting them on saturated plates, at which point they love to jump around. This is also true for any question on field uniformality; providing the flux paths don't saturate, all the flux tends to spread itself out in a fully uniform manner as it emerges into the gap.
best regards,
Chris MB.
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
So I have an update on this, having found a little time today to knock up some metal plates to test the essence of the idea. I am not well equipped with tools nor have facilities, so it was all I could do to make these 4 150mm x 150mm x 2.5mm steel plates in the time I had. That being said, Maxwell SV doesn't suggest there is much to gain, in terms of saturation, in having any more thickness.
Pic 1 – general view of the thing I've made – steel top and bottom plates with neodymiums between them. Each is 22mm diameter and 20mm high and they are stacked on each other making 40mm gap. There are two such pairs at each corner, making a total of 16 of the 22mm x 20mm magnets.
Pic 2 – a close-up. The studding is [non-magnetic] stainless steel. (These didn't actually seem necessary in construction, see below.)
Pic 3 – a lash-up quickie magnetometer using a A1302 Hall effect sensor (http://www.rapidonline.com/netalogue/specs/82-1020.pdf). These are around 1.3mV/G sensitivity. Power supply to the part set so as to read 2.50V with no field.
Pic 4 – ~1” outside the assmebly, 2.26V == 185G
Pic 5 – at edge of assembly, 1.41V == 840G
Pic 6 – ~1” inside assembly, 1.29V == 930G
Pic 7 – at centre, 1.27V == 950G
Pic 8 – on top, 2.74V == 185G
Pic 9 – I disassembled it afterwards.
Pic 10 – a simulation, as per a diagonal cross-section, using Maxwell SV.
Figures are approximate, of course, the device's tolerance being “1mV/G to 1.6mV/G”, but it gives some indication of the cut-off at the edge of the assembly and uniformity within. I still wonder if a bigger thickness of plates would help keep yet more of the field in, though Maxwell SV doesn't indicate any big advantage.
Assembly was a piece of cake. Supported on the studs, so that the plates were loose but retained, as I brought a stack of two of the neos up to the plates, they kinda 'snuck' into the gap without any fuss like they wanted to be there! Nothing dramatic, just a sharp, easily anticipated tug as they get close - as you'd expect from these magnets. Same with the next, etc.. Once inside they could be pushed around with a bit of force, but without too much problem. I could push the pairs right up close to each other without them repelling (as you can see in pics). By the time all 8 pairs were in place the whole assembly was quite rigid. There was no need to tighten the bolts up. In fact, the plates' holes had gone slightly off-square with each other and the bolts wouldn't actually tighten up easily! I had to knock the assembly on the floor several times to nudge the plates over to get the holes lined up! So the studs seemed to be redundant, though they usefully held the plates in alignment as the magnets were introduced. (I did clamp them down then, just to make sure the magnets didn't move.) Disassembly was a breeze – just push the pairs around to give some space to get a good hold, then just yank the pair straight out, in one quick movement.
I'm pleased with it. OK, 0.1T isn't going to get you into high energy cyclotron territory, but that much field with plenty enough space for a 50mm radius could, e.g., give you >1keV protons. Given the relatively small volume of magnetic material to permeate such a large volume, I think ~0.1T is quite a sound result in this case. With some pole-pieces, as suggested in the pic of the first post of the thread, then higher fields would be expected. I think the sharp cut-off at the edge of the assembly, and the field uniformity, suggests the approach is working 'as planned'.
Pic 1 – general view of the thing I've made – steel top and bottom plates with neodymiums between them. Each is 22mm diameter and 20mm high and they are stacked on each other making 40mm gap. There are two such pairs at each corner, making a total of 16 of the 22mm x 20mm magnets.
Pic 2 – a close-up. The studding is [non-magnetic] stainless steel. (These didn't actually seem necessary in construction, see below.)
Pic 3 – a lash-up quickie magnetometer using a A1302 Hall effect sensor (http://www.rapidonline.com/netalogue/specs/82-1020.pdf). These are around 1.3mV/G sensitivity. Power supply to the part set so as to read 2.50V with no field.
Pic 4 – ~1” outside the assmebly, 2.26V == 185G
Pic 5 – at edge of assembly, 1.41V == 840G
Pic 6 – ~1” inside assembly, 1.29V == 930G
Pic 7 – at centre, 1.27V == 950G
Pic 8 – on top, 2.74V == 185G
Pic 9 – I disassembled it afterwards.
Pic 10 – a simulation, as per a diagonal cross-section, using Maxwell SV.
Figures are approximate, of course, the device's tolerance being “1mV/G to 1.6mV/G”, but it gives some indication of the cut-off at the edge of the assembly and uniformity within. I still wonder if a bigger thickness of plates would help keep yet more of the field in, though Maxwell SV doesn't indicate any big advantage.
Assembly was a piece of cake. Supported on the studs, so that the plates were loose but retained, as I brought a stack of two of the neos up to the plates, they kinda 'snuck' into the gap without any fuss like they wanted to be there! Nothing dramatic, just a sharp, easily anticipated tug as they get close - as you'd expect from these magnets. Same with the next, etc.. Once inside they could be pushed around with a bit of force, but without too much problem. I could push the pairs right up close to each other without them repelling (as you can see in pics). By the time all 8 pairs were in place the whole assembly was quite rigid. There was no need to tighten the bolts up. In fact, the plates' holes had gone slightly off-square with each other and the bolts wouldn't actually tighten up easily! I had to knock the assembly on the floor several times to nudge the plates over to get the holes lined up! So the studs seemed to be redundant, though they usefully held the plates in alignment as the magnets were introduced. (I did clamp them down then, just to make sure the magnets didn't move.) Disassembly was a breeze – just push the pairs around to give some space to get a good hold, then just yank the pair straight out, in one quick movement.
I'm pleased with it. OK, 0.1T isn't going to get you into high energy cyclotron territory, but that much field with plenty enough space for a 50mm radius could, e.g., give you >1keV protons. Given the relatively small volume of magnetic material to permeate such a large volume, I think ~0.1T is quite a sound result in this case. With some pole-pieces, as suggested in the pic of the first post of the thread, then higher fields would be expected. I think the sharp cut-off at the edge of the assembly, and the field uniformity, suggests the approach is working 'as planned'.
- Attachments
-
-
-
-
-
-
-
-
-
-
-
Dan Tibbets
- Posts: 578
- Joined: Thu Apr 17, 2008 1:29 am
- Real name:
Re: A compact DIY cyclotron package solution.
~ 1000 Gauss fields with perminate magnets- I didn't know they were that strong.
Dan Tibbets
Dan Tibbets
-
Todd Massure
- Posts: 443
- Joined: Thu Mar 11, 2004 12:38 am
- Real name: Todd Massure
Re: A compact DIY cyclotron package solution.
Thanks for the report Chris. Very interesting.
You probably know this already, but it's my understanding that there's nothing to be gained from stacking the neodymium magnets - in other words putting them in series doesn't add to the total flux in the magnetic circuit.
I would be curious what the field would be with all of the magnets making direct contact with the metal plates. Of course the smaller gap would change things a lot too.
Anyway, once again, nice work.
Todd
You probably know this already, but it's my understanding that there's nothing to be gained from stacking the neodymium magnets - in other words putting them in series doesn't add to the total flux in the magnetic circuit.
I would be curious what the field would be with all of the magnets making direct contact with the metal plates. Of course the smaller gap would change things a lot too.
Anyway, once again, nice work.
Todd
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
Maxwell SV suggests there is little difference to the internal field, however many you stack, it's just a bigger width across which that field is established. More room to 'do stuff' inside that volume! Three magnets might come apart as I push them around, and one wouldn't've given me enough space to get my hands in to pull 'em out! Maybe I'll try to build with just one and re-measure.
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
At some small risk to life and limb (namely, two small blood-blisters later) I re-did the above with 1 layer of magnets. As presumed – very fiddly! Two seems a good compromise!!
Pic 1 – at edge == 1090G
Pic 2 - at centre == 1120G
so a little more, much as Maxwell SV indicated - Pic 5
Pic 3 – I added two more of the magnets either side, so from 8 magnets to 12.
Pic 4 – now indicating == 1700G
Now I have to try to take them apart... there is quite some field strength around the edge to battle with... (edit - got them apart, not as easy/safe as with the two-stack. I noticed one felt like it might have been reversed when it came out, which would've acted to oppose the others)
Pic 1 – at edge == 1090G
Pic 2 - at centre == 1120G
so a little more, much as Maxwell SV indicated - Pic 5
Pic 3 – I added two more of the magnets either side, so from 8 magnets to 12.
Pic 4 – now indicating == 1700G
Now I have to try to take them apart... there is quite some field strength around the edge to battle with... (edit - got them apart, not as easy/safe as with the two-stack. I noticed one felt like it might have been reversed when it came out, which would've acted to oppose the others)
- Attachments
-
-
-
-
-
-
JohnCuthbert
- Posts: 339
- Joined: Mon Aug 19, 2002 4:30 pm
- Real name:
Re: A compact DIY cyclotron package solution.
Very impressive.
I don't know about building a cyclorton, but you might have the start of a DIY NMR spectrometer there.
I don't know about building a cyclorton, but you might have the start of a DIY NMR spectrometer there.
-
Todd Massure
- Posts: 443
- Joined: Thu Mar 11, 2004 12:38 am
- Real name: Todd Massure
Re: A compact DIY cyclotron package solution.
Chris,
It seems to me that without the studs it might be possible to leave the magnets in place on one plate and slide the other plate on and off for access as needed. From your experience does that seem possible?
Todd
It seems to me that without the studs it might be possible to leave the magnets in place on one plate and slide the other plate on and off for access as needed. From your experience does that seem possible?
Todd
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
No. Not at all. You mount the magnets with the same pole all facing one way. So once you've got a concentration of them, then they'll twist and jump right off the spot given the slightest inclination so as to jump onto the magnet next to it, if you only have one plate. With both plates facing each other, the first one comes in and snaps the plates into a position which holds them apart at a given distance, so the next ones then slot in without being able to twist and jump over. You loose the flux path to the centre if you try without the two plates together, which means the magnets no longer 'feel a contentment' to stay where they are.
Also, even if you could hold them down somehow (I was originally thinking to drill snug-fitting holes in a piece of teflon) with that many magnets the next plate would get so far then I reckon you'd not be able to budge it any further.
Also, even if you could hold them down somehow (I was originally thinking to drill snug-fitting holes in a piece of teflon) with that many magnets the next plate would get so far then I reckon you'd not be able to budge it any further.
-
Todd Massure
- Posts: 443
- Joined: Thu Mar 11, 2004 12:38 am
- Real name: Todd Massure
Re: A compact DIY cyclotron package solution.
If one was willing to be locked into a permanent configuration, maybe it could first be assembled as you have, then, by sliding them around some, add a thin layer of epoxy between the magnets and one of the plates. After the epoxy cures, then perhaps the other plate could be slid on and off (?)
Todd
Todd
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
I still don't think you'll be able to move that plate easily. With that set of magnets, the holes didn't line up on the plates and I had to hit the whole assembly like a hammer on the floor just to budge them a millimetre or two at a time. It was really very rigid. If you did that with a plate sticking out after getting stuck on a couple of the magnets at the edge, it'd probably flip off, or something else dramatic. Besides, you suggested a removable plate for access - wouldn't epoxy-ing everything make access less flexible? I guess you could try all options, depends on the application. This should suit my purposes.
-
John Futter
- Posts: 1850
- Joined: Wed Apr 21, 2004 10:29 pm
- Real name: John Futter
- Contact:
Re: A compact DIY cyclotron package solution.
Chris
If you put the nuts on the inside you can use them as a jack to slightly seperate the plates to make moving the blighters easier.
also an aluminium plate with drilled holes to locate the magnets fixed to the steel pole pieces -plastic /brass 316 Stainless are also suitable
If you put the nuts on the inside you can use them as a jack to slightly seperate the plates to make moving the blighters easier.
also an aluminium plate with drilled holes to locate the magnets fixed to the steel pole pieces -plastic /brass 316 Stainless are also suitable
-
Chris Bradley
- Posts: 2930
- Joined: Fri May 02, 2008 7:05 am
- Real name:
Re: A compact DIY cyclotron package solution.
John Futter wrote:
> If you put the nuts on the inside you can use them as a jack to slightly seperate the plates to make moving the blighters easier.
I am not sure it would, and I would not recommend anyone tries. (You can if you like, of course....) These came out very easy, the one-stack a little more fiddly only because that was just a single finger's width, hence limiting force you can put on them. (I got the blood blisters from re-packing them after removal, not from assembly/disassembly!!) Once you hold one and just pull in one, clean big movement to fully clear the edge, there is no problem at all. If the plates were separated by more than the height of the magnet then I could imagine they could easily twist and jump as they clear the edge.
Reporting magnetic field and uniformity is one aspect of what I've written. I'd say of equal substance is this ease of assembly - providing the plates are somewhat retained by a separation around the height of the magnets, all goes very smoothly. There is so much force in these things that to get 16 of them all together in one assembly with this much ease came as a surprise. If you've ever tried to put just a half-dozen on one plate and move them around/close, you get a feel and a big respect for how much force/damage these can apply.
> also an aluminium plate with drilled holes to locate the magnets fixed to the steel pole pieces -plastic /brass 316 Stainless are also suitable
That was my original intent, to locate the magnets into a little recess. But this would make the assembly process I've described not possible. These magnets stay where they are yet are straightforward to push around while in the assembly. They become 'very well behaved', considering. Therefore, the purpose of the studs is merely to provide just a little extra clamping force and a little extra confidence that they'll stay put, though they did anyway without the studs clamped up.
I absolutely do not believe you could safely assemble the part by hand, as pictured above, by lowering a plate down on top of the magnets, nor sliding across a plate (As Todd has suggested). One or other would be required if you create a recess for them to fit in.
I guess the magnets feel a good flux circuit into the plates because those plates have a relatively much larger area to cross back to the opposite pole of the magnet, and so they have no incentive to do anything else but hang on to the plate they are nearest. If you do anything that means they no longer have a plate in direct contact, then they'll probably find the pole of another magnet more attractive and wiill jump and twist if they can.
If anyone wants to try a different assembly technique, then I'll be interested to hear. Just think safety and wear eye-protection of course. These things can literally start 'twitching' and a ms later have jumped a metre whilst building up a shattering momentum in the process [and the 'shattering' is either magnet or bone, just depends what's in the way].
> If you put the nuts on the inside you can use them as a jack to slightly seperate the plates to make moving the blighters easier.
I am not sure it would, and I would not recommend anyone tries. (You can if you like, of course....) These came out very easy, the one-stack a little more fiddly only because that was just a single finger's width, hence limiting force you can put on them. (I got the blood blisters from re-packing them after removal, not from assembly/disassembly!!) Once you hold one and just pull in one, clean big movement to fully clear the edge, there is no problem at all. If the plates were separated by more than the height of the magnet then I could imagine they could easily twist and jump as they clear the edge.
Reporting magnetic field and uniformity is one aspect of what I've written. I'd say of equal substance is this ease of assembly - providing the plates are somewhat retained by a separation around the height of the magnets, all goes very smoothly. There is so much force in these things that to get 16 of them all together in one assembly with this much ease came as a surprise. If you've ever tried to put just a half-dozen on one plate and move them around/close, you get a feel and a big respect for how much force/damage these can apply.
> also an aluminium plate with drilled holes to locate the magnets fixed to the steel pole pieces -plastic /brass 316 Stainless are also suitable
That was my original intent, to locate the magnets into a little recess. But this would make the assembly process I've described not possible. These magnets stay where they are yet are straightforward to push around while in the assembly. They become 'very well behaved', considering. Therefore, the purpose of the studs is merely to provide just a little extra clamping force and a little extra confidence that they'll stay put, though they did anyway without the studs clamped up.
I absolutely do not believe you could safely assemble the part by hand, as pictured above, by lowering a plate down on top of the magnets, nor sliding across a plate (As Todd has suggested). One or other would be required if you create a recess for them to fit in.
I guess the magnets feel a good flux circuit into the plates because those plates have a relatively much larger area to cross back to the opposite pole of the magnet, and so they have no incentive to do anything else but hang on to the plate they are nearest. If you do anything that means they no longer have a plate in direct contact, then they'll probably find the pole of another magnet more attractive and wiill jump and twist if they can.
If anyone wants to try a different assembly technique, then I'll be interested to hear. Just think safety and wear eye-protection of course. These things can literally start 'twitching' and a ms later have jumped a metre whilst building up a shattering momentum in the process [and the 'shattering' is either magnet or bone, just depends what's in the way].
Re: A compact DIY cyclotron package solution.
just fyi on teflon under high stress it tends to flow out of the way if you put 2 magnets one on each side the teflon will slowly squeeze out of the way. Know someone who worked on the starfigter jet in the airforce and they had issues under high g of the wires migrating out of thier teflon sheath and shorting to each other but they would spring back after and made them very hard to find till they had to cut them apart and found the thinning of the walls.