How Amateurs Can Build a Simple Magnetic-Resonance Spectrometer

in the nineteen fifties

I've seen kids do this at science fairs. All you need to do CW NMR is a fairly homogenous magnet, a couple of coils, and a little RF stuff. Pulsed isn't much harder.

I'd bet you could do something with some iron pole-pieces and a bunch of the little bulletin-board supermagnet cylinder things. Move the magnets around to tweak field uniformity. A microwave oven magnetron could be used, too, with a little pole-piece shimming maybe.

Water makes a huge NMR signal. The Qs are extraordinary, 1e6 to 1e9, and depend mostly on how homogenous the field is across the sample volume. Higher frequencies, and corresponding higher field strengths, make better signals, but even the earth's field does usable NMR.

Amateur MRI should be feasible, too.

Scientific American used to be great.

John

in the nineteen fifties

with an fpga adc and dac you have what is needed, but getting it all to work will take some effort. but it is definatly doable by a (very) determined amateur. I think biggest problem is getting a homogeneous magnet, even 100's of ppm magnets are kilo bucks

magnet shimming is like a black art, and you'll probably need a working system to actually measure your progress

yeh, signal goes up with something like the field to the power of 2.5 but as you say the H signal is huge

-Lasse

On a sunny day (Sat, 05 Feb 2011 12:28:28 -0800) it happened John Larkin wrote in :

in the nineteen fifties

Yes, that is exactly what the article uses.

I think with modern electronics, some linear current drive (feedback), you could perhaps make some really usable stuff. Yes, Sci Am I used to read it if I could get hold of it. Then later it sort of detoriated in a way, maybe it fell to safety, political .. no risk to kids that sort of thing. They had very interesting and inspiring special issues too back then.

back in the nineteen fifties

I saw a kid do it at a science fair in roughly 1964, with tube electronics. He put orthogonal coils around a test tube of water, fed one from an RF oscillator, ran the other to a scope; Tek loaned him a

545 and I was jealous. He use an old magnetron magnet, working in the low MHz, similar to the SciAm thing but using two coils and a pure DC field.

A small sample volume and a bit of shimming will get you a signal.

Sure, once you get a signal, tweak and learn.

In liquids it is. For some reason it doesn't work as well in solids. They like to spin solids, too, to homogenize the field... air turbines at insane speeds.

John

in the nineteen fifties

Modern NMR systems use superconductive magnets with passive and active shims. The passive shims are small superconductive coils inside the magnet that do coarse shimming and run in persistance mode. The active "room temperature" shims are a maze of coils that trim the field uniformity. The basic ones are polynomials on the Z axis, the magnet bore. Z1 slopes the field linearly as Z changes. Z2 bends the field intensity as z-squared. Than there are all sorts of funny and sideways ones. I saw one system that had 40 shim coils, and it was a nightmare to tune.

The biggest magnets are over 20 feet high, run over 200K gauss, and cost megabucks. That hit around 1 GHz hydrogen line frequencies.

I build NMR gradient drivers, which make Z1 (linear gradient) pulses. People write scripts with funny names that look like musical scores, combinations of RF waveforms and gradients. They do chemical and quantum mechanical tricks with all that. NMR is a pretty powerful analytical chemistry tool. Agilent recently purchased Varian Inc, probably to get the NMR stuff.

John

back in the nineteen fifties

yeh for CW it is basically just a radio transmitter and reciever I've only done pulsed NMR, it would be a bit more work to get that to work, but with todays electronics none of it is not overly complicated, biggest hurdle is still get a good magnet

getting the reciever to work in insanely small signals microseconds after it has seen at least parts of a several (soemtime 100s) watt transmit pulse takes a bit of work

sure a small sample volume will make it easier, but it will also give you less signal

it's only several KHz :) I can understand why moving the sample around in the field will expose the molecules to an average of the field, but as far as I understand theres more to it than that

working with this stuff, what surprises me is that much of it was invented in the late 50's

-Lasse

THIS kid will hold the crown for home brewing for a long time...

Mark

back in the nineteen fifties

Except for dealing with the pulse, (diodes or whatever) I think pulsed is easier, you can still see 'something' even when it's way out of tune. CW is harder, you have to get near resonance and get the matching close.

mineral oil gives a nice proton resonance signal for ~1/2 Tesla magnetic fields.

'sigh' George H.

rom back in the nineteen fifties

...

T1 the spin lattice relaxation time goes down to.... (micro seconds??) in solids. I'm not sure why that is? I do know that to get any interaction the magnetic moment needs to see a magnetic field changing at a rate equal to it's resonance frequency. In a solid there must be more... Perhaps due to the vibrations in a solid?

George H.

We (myself and two other chem students) built one with slightly more modern equipment back in about '73. We had a lot of help from the local ham radio group. I built a gas chromatograph and a mass spectrometer from Sci-Am. articles. I entered the mass spec in the regional science fair. Won several ribbons and certificates. Hardest parts were making McLeod gauge and mercury vapor diffusion pump from glass. On a good day the vacuum system would pull a millitorr, maybe less.

All of the Amateur Scientest articles are available > Somebody has been posting terrific old Scientific Amercian articles from =

back in the nineteen fifties

..

back in the nineteen fifties

Yeah, pulsed shouldn't be that hard. The ringdown, the FID thing, is impressive.

John

rom back in the nineteen fifties

...

Yeah and once you've got pulses, you can do pulse sequences. And see spin echo's which are cool! A 90 degree pulse, delay tau, a 180 degree pulse.. and 2*tau seconds later there is an echo... If Jan is going to try and build one, I'd reccommend starting with a two crossed- coil design. Pulse on one coil and pick-up on the other... coils at

90 degrees to each other so there is less of the RF pulse that gets into the reciever.

George H.

On a sunny day (Mon, 7 Feb 2011 13:54:20 -0800 (PST)) it happened George Herold wrote in :

Ha, I was not at that point, but it seems a clever idea to cross the coils 90 degrees. It is very busy here at the moment and I do not have a lot of time now for tinkering. And there is an other big project that I have not even really started on. But, sure, God Willing (they say that ) i would like to try this. I need the magnetron magnets for some other experiment with superconductors, and I already have those. Next Christmas?

2012 sounds cool too.

s 90 degrees.

r tinkering.

rs,

OK. One 'gotcha' to watch out for if you use water for a sample, is that it can have really long T1 times... (The time it takes the spins to align with the local DC magnetic field.) like seconds! This sets the repeatition rate at which you can do the measurements. So you need to add a bit of magnetic impurity to speed things up. Copper sulfate works well.

George H.

from back in the nineteen fifties

Ca...

s s .

it gets a bit messy making 90 degree coils with a permanent magnet, because of the direction of the field. a bit easier with a superconducting magnet where you already have to make saddle coils if you want easy access to the sample

how much isolation can you really expect from doing 90 coils? there's huge difference between the tx and rx signal

a single coil, a quarterwave stub and some diodes works resonably well to isolate the transmitter and reciever, when you find an LNA that doesn't take forever to wake up after it has seen the transmit pulse

-Lasse

es from back in the nineteen fifties

s+Ca...

ges

is

ff.

ut

Sure, the two coils have to be perpendicular to the DC magnetic field. Fortunately we have three dimensions. :)

a bit easier with a superconducting

I've used super conducting magnets, but have never done NMR/ESR in them. The source of the field doesn't make that much difference. Well OK currents can be lots more stable than magnetic materials.

There's always leakage sure. I was thinking about this, coming home. The big disadvantage of the crossed coils is that you need a lot more power in the pulse coil. (You want to keep the pickup coil as close as possible to the sample.) Our first NMR had crossed coils and a three stage RF amp that I helped design... I don't remember the power, the only complaint calls we get is when people over drive it and set- off the thermal shutdown. (No one reads the manual.)

Anyway I was talking with Norman, (who is the genius behind all our pulsed NMR stuff.) and he was happy to report that the new single coil design needs a single (or was it two?) gain stage.

George H.

On a sunny day (Mon, 7 Feb 2011 14:41:23 -0800 (PST)) it happened " snipped-for-privacy@fonz.dk" wrote in :

I think any little bit helps, I would expect that one can easily adjust for zero coupling by driving one and looking at the output of the other.

In the original pdf the frequency is about 6 MHz (water). Because the magnetic field strength of those magnetron magnets is low. A quarter wave stub at that frequency will be rather long?

cles from back in the nineteen fifties

urs+Ca...

sages

MR is

tuff.

out

he

s

ee

is

d-

t

the rf field should be 90 degrees to the static field so my point was the with a permanent magnet the static field it perpendicular to the "hole" in the magnet, so the coil can be a solenoid around a sample tube. Making one more coil 90 degrees to that coil you need to make a saddle coil and make sure it is aligned with the poles of the magnet

in a super conducting magnet the field is parallel to the "hole"

yeh, the strong permanent magnets are something like 2000ppm/k

-Lasse

it'll be a little over 8 meters of rg58, so not impossible, you could also get the same with two caps and a coil if you don't want the long cable

-Lasse