NMR Magnet yoke design

[ChristofferB]

Do you think this is a yoke?!

Hi all! Once again I'm trying to get some progress on my home built tabletop NMR spectrometer, and I figured the magnet's a good place to start, since everything else is kindof dimensioned after magnet strength.

I've settled on getting two puck shaped neodymium magnets, around 15 mm height 30 mm dia., and and mounting them in a U shaped yoke with a 10mm gap for the probe.

This will give me 0.85 T in the gap, which would have 13C resonate around 9.1 MHz, that's all good.

But how should the yoke be made? Soft iron, of course, but in practice;

 - how should it be sized?
 - can the magnet diameter be larger than the width?
 - get some flatbar and bend it to a U or weld 3 pieces in a square-ish U?
 - how could the magnets be attached to the U? epoxy maybe? They'll be pulling away from the yoke, towards each other with some significant force..
 - Where do you even GET soft iron?
--OR would the easiest be to cut out a chunk of a stripped toroidal transformer's core to make a rounded U?

thanks in advance! I really want to get this right, because a good, homogenous magnet would be a good basis for a lot of development and experimentation on the 'tronix side.

I found a pretty nice gap field strength calculator: https://www.kjmagnetics.com/gap.calculator.asp, might be useful for some

Thanks for the attention!

[jbb]

Not an MRI expert in any way, but I'd suggest you think about laminations.  The will have lower eddy current loss when you start throwing AC magnetic fields around (e.g. gradient coils for imaging). Eddy currents would distort the AC field components and impair imaging.

You might be able to find a big set of laminate iron U cores.  These would have 2 gaps; maybe use 1 gap for the sample and the second gap for the magnet stack?  Nano crystalline / amorphous iron would be a good candidate (but might be very expensive in large sizes).

[ChristofferB]

evb149: Thanks for all the suggestions! Getting something completely custom made would of course be the ideal, but not super price-friendly. I wouldn't mind if the project was reproducable, either, so something a bit more common would be nice. I could always go down to my local machining factory and ask if they could help, but the more I can do myself the better.

The old non-superconducting magnets do come up occasionally, but they're so heavy and expensive to ship It pretty much has to be VERY close to me to be possible. Maybe I should put an ad up at my university, that'd be a blast:

WANTED: NMR MACHINE OR MAGNET

Preferably free, please contact me on...

Turns out I had a U shaped piece of soft iron from an old hand-cranked telephone dynamo. It was magnetized, but I've heat treated it (750 *C for ~10 min), and I may try that. I can always acetone off the magnets. It's way wimpier than what I'd like, though. only 8mm thick and 20mm wide. Could put two side-by-side...


Jbb: That's true, re-using a transformer core was my first thought too. Could it be circular? I have a giant toroidal transformer that I could plunder..

Only issue is you'd be pressed to find a U with the correct width for your magnets. as I understand it, it's VERY important for homogeneity that the magnets are parallel. And in my experience, hand filing laminated cores tend to split the shims outwards.

By the way, here's the design I'm sortof imitating. I'm not using electromagnets as my main field source, though, as that would require building a custom current-controlled psu.
http://www.its.caltech.edu/~derose/labs/exp5.html
 

[T3sl4co1l]

HRS (hot rolled steel), or annealed mild steel (of whatever sort), will do an okay enough job for pole pieces.  The saturation may not be very high, but this should still be okay at 0.85T.

Make sure they are thick enough.  Minimize gaps in the path.  If you're going to weld it, use a mild steel (or maybe something with a little nickel, but for sure nothing hard) filler, and anneal it when you're done (blast it up to red-orange hot, then let cool slowly).

Magnets can be glued in place; you may need to arrange some sort of jig to position them.  They should stick to the pole pieces stronger than each other, at least, so you shouldn't have to worry about that.

Permanent magnets aren't a very good choice because they have a strong tempco, hysteresis and aging.  You'll also need pole pieces on top of them (the working face) to get a uniform field.  You'll still have to shim the field afterwards.

PM at least saves you a lot of field windings.  You still want them, to trim drift -- and you only need maybe 0.1T worth of trim if that, not the whole 0.85T, so it helps a lot.

Then all the shimming...

Tim

[ChristofferB]

That's good to know, that the pole pieces are less critical like that.

I'm aware that technically, electromagnets would offer better characteristics overall, but as mentioned before, there's just so much more work/machinery involved, so
"magnet V.1" will be mainly permanent magnet based.

The pole piece on top confuses me. I've seen this gap arrangement have cone-shaped pole pieces on top of the magnets, but could it just be thin mild steel sheet?




Adding some small coils to trim drift might be a good idea, maybe this could just be a DC bias on the field modulation coils?

I'm thinking to position the magnets I'd clasp them together around a center block that's exactly (-some µm) the gap diameter, apply epoxy to both magnets and U, and slide i in, and then with a wooden mallet or some such thing gently tap out the centerpiece. All good projects requires the delicate hit with a mallet.

This may seem basic, but how exactly would you shim a magnet?

[ChristofferB]

That sounds logical, I'll have to read up on shimming, I think.

My goal is to be able to do spectroscopy, yeah. To my knowledge that has never been done by a DIY instrument before, atleast not widely documented if so.

I'm going to start with 13C because there's a wider frequency range between resonances, and thus you can make do with lower resolution. Commercial analytical NMR machines typically run at 100 MHz for 13C, and 400 MHz for 1H - and the range is much larger on 13C NMR:

Acetone 1H NMR:


Acetone 13C NMR:


So I just figured it was a good start. On the other hand, the abundance of 13C is much lower than 1H, so 1H might actually provide more intensity.

My plan is to do old-fashioned CW NMR, and then AF modulate the magnetic field, to sweep over different resonances.

True, tabletop models do exist commercially. if you make the magnet gap tiny enough, you can obtain great fields in tiny magnets, I even think they have a solid state NMR ic now that does all the sensing and magnet stuff..

[jbb]

Jbb: That's true, re-using a transformer core was my first thought too. Could it be circular? I have a giant toroidal transformer that I could plunder..

Only issue is you'd be pressed to find a U with the correct width for your magnets. as I understand it, it's VERY important for homogeneity that the magnets are parallel. And in my experience, hand filing laminated cores tend to split the shims outwards.

I guess you could cut a toroid.  But you'll probably get a non-uniform field.

Regarding electromagnets vs permanent magnets: I expect permanent magnets are easier to get started with (no heating issues or high current drive required), but maybe you could plan to add a trimming coil so that you can tweak the field a little bit.

If you ever want to go full electromagnetic, the current sources are a bit difficult to design because you'll need very good stability (i.e. current held constant over time & temperature), low noise and low ripple.  This can absolutely be done, but you don't want to leap into addressing all the NMR challenges on your first go.

[ChristofferB]

In light of what T3sl4co1l said, I don't think I'll attempt to salvage a transformer core. getting some 30x30mm square mild steel, or even better wrought iron and then get a friend to weld the U is probably as easy, and will give better control over dimensions, etc.

Last design quarrel I have is the pole pieces ontop of the magnets in the gap.

you don't want to leap into addressing all the NMR challenges on your first go.

That's exactly what I'm thinking. I haven't even begun designing the electronic side of things, but I'm figuring, if I can get the magnet 'sorta right' it's a pretty good platform to mess around with.

By the way, if I'm welding from square stock, I can pretty much do any shape I want. Below is a cyclotron magnet, I suppose design requirements of homogeneity are simila. Would it be better to make a complete square O of yoke, and then have the magnets in the middle?

[johansen]

no welding required if you make it like it is in the above photo, but you will have to have the metal machined flat. band saw cut may be good enough if its square.

you may have to glue the magnets to the steel.

[ChristofferB]

Hey you're right! It'll support itself. Of course, it'll have to be lying down, the sample needs to be vertical. You could even clamp it together on the outside for added stability, I guess.

I think we're talking angle grinder, bench grinder, and hand filing/lapping for the machining side of things. it'll be a lot of work, but I hope it'll turn out decent.

Would one get anything out of adding iron powder to the epoxy?

[johansen]

you should be able to get the surfaces flat to .001" with just a dial test indicator and a flat plate and a file. you don't need to buy a surface plate, because they don't need to be flat, they just need to fit together, and you can use aluminum foil as a feeler gauge, and dykem or layout fluid to help the process.

anyhow ask around, you can probably find someone who has a small milling machine or a lathe. if their milling machine is trammed correctly you should be able to get less than .0003" thick air gap between the pieces of metal.

you don't need a mill if their lathe can hold the long side of the bar and face the entire surface.