"Shuttle Device for Use in a Shared Commercial NMR Instrument Version II "
(December 2004)
by A. G. Redfield, Brandeis University

Return to Index

The Helmholz Coils

[Sections like the following one that refer to "very low fields" or "Helmholz coils" do not describe anything related to MRC. They are new capability used to generate the data in our last paper in PNAS, Dec. 2004]

The Helmholz coils are each about 16 cm in average diameter, about 2 cm thick radially and each 4 cm longitudinally. They were made by Dmitri Ivanov for his quadrupole resonance experiment (J. Magn. Reson. 166:19-27 (2004)), and each has about 345 turns of #16 wire. They were would on a lathe using a wooden coil form 4 cm long and about 12 cm outside diameter, having gaps and screwed together to >20 cm diameter end pieces of wood so that the parts of the form could be pulled off the coil without sliding. The wood was first treated with wax to resist epoxy bonding. Wire came from a spool on an improvised stand next to the lathe, and it was tensioned while winding by a rag held in the hand, or pieces of Delrin clamped lightly on the wire. After each layer was wound epoxy resin was generously smeared on the entire layer, before winding the next. The result is rigid and conducts heat well so that an air fan could increase its power capability, and we did point a fan rather crudely at the coils when Dmitri used them.

The coils are supported by 3 plates outlined on fig. ST-8 (top), separated by spacers which are shown informally in the sketch at the bottom of the same figure. The coils rest on the two lower plates, contacting the four wings on each plate only, for vertical support, and held in place radialy by the four rods. Other than the wings, the coil surfaces are exposed for the sake of cooling. A small muffin fan hung in mid-air helped cooling for Dmitri's setup but we have not used it yet in this device. The rods are ¾ inch in diameter and the central one is threaded at each end for 10-32 screws. The upper rod is drilled out with a ¼ inch hole and the purpose of the upper plate was simply to provide a convenient surface to support the shuttle tube assembly. The coils are held in place by several plastic cabling ties. The supporting plates and rods have pieces of Scotch-brand thermoplastic glass electrical tape attached to them using a soldering iron briefly applied to melt their adhesive, so that the coil does not contact the support structure directly. The inside diameter of the coils was only slightly smaller than the outside diameter of the top aluminum flange of the Oxford 500 magnet, so I only needed to offset the holes in the upper two sets of rod by 1/8 inch to do it as shown. Alternatively separate sets of holes could be used for the four feet. Probably you will redesign the coils so I do not give detailed designs.

The lower support rods (fig. ST-8, between the magnet and the coil) were Delrin drilled out to ¼ at the top, and hollow, with holes large enough for a cap screw head at the top. They fit the outside of the aluminum plate at the top of the Oxford 500 magnet snugly, and thereby align the magnet radialy as well as supporting it vertically, resting on a black plastic (?anodized aluminum?) surface to which the top plate appears to be screwed. Important! The lengths of these rods should probably be determined once the shuttle-tube assembly is complete. Their length should be such that the shuttle tube assembly sits at the correct height for maximum sensitivity of NMR detection (that is, with the sample at the sensitive region of the probe) when the 10-32 screws with larges knurled tops (McMaster Carr) were screwed in with ½ to 1 cm projecting out the bottom of the flange of part (8) ("HEIGHT ADJ SCREW", fig. ST-1). These screws would have to be adjusted if the NMR tube is not epoxied into the NMR tube adapter at the usual heath that you establish. The height adjustment is again discussed in a few pages.

For our setup we have designed the top cap (8) to have the adjustment screws as high as possible, so that the Helmholz coil can be rather high, and so that its center is about 14 cm above the top of the top aluminum plate of the magnet where the field is slightly less than 0.03 Tesla, dictated by the fact that the Helmholz coils and our 50 V 10 A supply could reach this field, and buck it to allow us to get to nearly zero field. [Do you wonder why not zero? If we went close to zero we would have to verify that there is no transverse field component at the magnet center.]

The solid wires of each coil are connected to separate short 2-wire cables that are tied down with tape to the support assembly. Then these cables go to a temporary screw terminal barrier strip where the are connected in series, and from there via a few feet of multiple stranded wire cable to a 24 pin AMP "circular plastic connector" type 23-24 which hangs on the cable (more trouble to order than to make. Do not bother to get a crimping tool, solder the wires.). The solid wires must not flex in ordinary use! This makes a compact unit that can be lifted on and off the magnet when moving in or out. This magnet produces more than 320 G using a 50 v 10 amp Kepco supply which is left turned on, and whose output connection is on and off with a solid state relay. A reversed power diode across the coils protects everything. See Fig. EC. This is the fringe field 13.5 cm above the magnet. The gradient from the main magnet is about 10 G/cm there. For Dmitri's work we could get the field homogeneous to about 1 Gauss over 1 cm by running unequal currents through the two coils. We did this by connecting two cheap Tektronix supplies carrying opposing currents to the two coils. You can connect several current- regulating supplies to the same place but you have to put in series resistors or series fixed voltage supplies in series in order that the supplies can operate on a positive voltage within their range. We will do this if we want to measure R2 or do polarization transfer at low field. However, it is unlikely that we will do this except perhaps for R2 of ³¹P. We probably cannot get much above 1000 Gauss with this magnet, and in most cases T1 is too short at this field to allow us to do any experiments.

It would be nice to have a temperature sensor that will turn off these coils. Otherwise in case of an electronics failure the 500 magnet might be damaged when the magnet goes up in flames.

The extender tube (fig. ST-5A) can be put on the top when we need it, to get to very low fields by raising the sample to 14 cm above the top of the magnet and using the field of the coils to buck the remaining ~300 gauss there to nearly zero. Because it is hard to machine with precision we made it slightly larger in diameter than the glass. Probably 0.804 inside diameter would work better to avoid problems. It is very straightforward, mating at the bottom part (8) in a precision manner, and shaped at the top like the top of part (7), so that the stop-tube-clamp can be put at the top of the extender tube instead of on the top of part (7). It is awkward to have this tube attached if it is not needed, and I designed it with trapped screws where it attaches to the top of the shuttle tube, which can easily be tightened by hand (Fig. ST-5B). Only the lower half of the inside of the brass tube has to be well machined because the upper half does not actually contact the shuttle. It would be hard to machine an extender tube much longer than this one unless you have an unusual lathe, and instead a glass tube coupled with plastic caps as described above would be needed.

The stop tube, fig. ST-6, is very simple. We have 3 of different length, overall 20, 38, 81 cm. This is a good assortment to cover the full range of fields. The shortest is for use at very low field with the extender tube. A short shoulder around the outside of the lower end piece (fig. ST-5A) with small clearance to the glass centers the stop tube so that the end of the shuttle tube that holds the O-rings will not hit the shoulder of the stop tube that contacts the O-rings when the shuttle tube hits the stop. This shoulder should be short, as shown, so that the glass cannot be easily broken when the stop tube is not lifted out absolutely vertically.

The upper end of the stop tube is simple cut off with a small bevel, at the top. It is coupled to a rubber tube that goes to the solenoid manifold by a vacuum quick-disconnect as described in a later section. It happens that the tube slides into this fitting too far, so we have a stainless steel collar semi-permanently attached about ¾ inch from the top of each tube to prevent this from happening. This has to be put on the end after the clamp (just below) is put on the tube.

The clamp assembly, fig.ST-7, holds the stop tube in place to determine the lower field-value. There is an identical one on each stop tube. It gave a lot of trouble because it must be easy to move the stop tube when the clamp is released, but it must hold it tightly enough to resist the thousands of poundings during cycling. No matter what the design, it is good to put a piece of paper tape on the stop tube just above the clamp when you start a new run so that you can look in on the spectrometer every so often and see if the tube is moving upward during the run, at least if you are not confident about this point.

We list and explain the components of the clamp starting from the bottom, in the order in which they have to be placed on the stop tube. The first component is, when in place, just above part part (8), fig. ST-4G (and above the plastic cap (9), fig ST-4H, which it MUST not touch), is a ring (Delrin or aluminum) about 1.2 " OD, 5/8 " ID, ½" long (not shown)) that sits in the small well at the to of part (8) and centers the stop tube. It is machined to slide easily on the stop tube and fit easily into the well. It is essential that this end of the stop tube be moderately centered so that the shuttle's upper end will enter the stop tube smoothly. After this is an optional 1/16 thick O-ring 5/8 or ¾ inside diameter, as a cushion. Next is the part (A) fig. ST-7 which is in fact two nearly identical pieces of aluminum or brass that fit around the 5/8" stop tube with the central hole illustrated made large enough to slide on the tube without much resistance, as already mentioned. The gap between pieces allows the clamp to be closed tightly on the tube. One side of the clamp is closed by a 10-32 screw with a tight lock nut and a spacer chosen to hold the clamp slightly open (all not shown). The spacer consists of washers or thin shim stock, and they are threaded onto the screw after it is inserted through the un-threaded hole of one side of a clamp, and before the screw is threaded into the other side of the clamp. The screw on the other side is closed with the wrench (S2) that can be tightened without great effort. It has to be long enough to clear the various components nearby. Next on top is another O-ring for a cushion, and finally a top plate (B) that holds two knurled screws (S1) in two small tubes that extend below it. The part (B) is put in with these tubes pointing downward and is oriented as shown in fig. ST-7 so that the flat plat is in contact with the O-ring. The length of the small tubes is such that the lower ends of the tubes almost touch the upper surface of part (8), and can be screwed into holes tapped into that plate to secure the clamp system. The idea is that these screws can be tightened to slightly compress the O-rings and firmly capture the clamp, without preventing the clamp from being loosened. The screws S1 and plate assembly trap the screws so that they do not get lost. Finally, a small slot was cut in one of the clamp pieces to make it easier to close while permitting a looser fit when unclamped.

The sample-sucker for pulling out samples is shown on fig. ST-8. It consists of a plexiglass fitting cemented to a standard US ¾ size sweat-solder Tee. To the side arm of the Tee is permanently attached a small valve, and a rubber tube is permanently attached from the other side of the valve to the vacuum reservoir on the Tower (below). The other arm of the Tee is not used and you can probably use an elbow instead of a Tee. The valve is normally closed. To remove a sample the stop tube has to be removed by unscrewing the knurled-knob screws that go through part (B) page ST 7, and pulling the stop tube out being careful not to break the shuttle tube. Then the bottom of the sucker assembly is pressed into the 1.2" diameter shallow well normally occupied by the lowest part of the clamp assembly, in the top of the shuttle assembly (see part (8) fig. ST-4G), and the valve is opened. The sample rises and you may see the top of the shuttle enter the plexiglass piece. You then gently lift the sucker, with the sample stuck to its bottom (leaving the valve open, of course). Once it is raised about one inch you grab the shuttle with your fingers and lift it and the sucker out, and eventually close the valve on the sucker. That's it.

Installation. Generally I install the shuttle tube assembly before the probe and, obviously, after the Helmholz coils are in place. Then insert the probe, as usual taking things very slowly. For the 10 mm probe there is no problem. For the 5 mm probe you have to worry about a vestigial ridge that previously aligned the probe's gradient connector. [these lines were re-routed by Varian to come out the bottom of the probe on a connector, at no cost, after the probe was delivered. Be sure to ask them to do this, apparently it is easy except there was some indication that they did not do a great job of by passing these leads against RF interference. Almost certainly they could omit the ridge I am talking about, it appears to be made of epoxy]. For use with this probe I put in a matching groove (on part (10), fig. ST 4A, and by rotating the entire shuttle tube assembly the ridge and shoulder can be lined up. Varian's ridge can also fit into one of the slots in part (1) that is put in to let air pass. In either case you have to be very careful not to force the shuttle assembly to rotate and break something or deform the thin upper aluminum tube on the probe. [Better get Varian to leave out the ridge! Bruker does not have such a problem].

Remember, do not loosen the probe at the bottom the way you normally do for a Varian probe. Set the probe height up once, then do not change it (assuming you have your own probes). By not having the shuttle tube and probe in direct vertical contact, as Varian does, I hoped to avoid transmission of vibration to the probe when the shuttle lands at the bottom.

Adjustment. The only adjustment is to have the height of the shuttle tube such that the sample is centered in the sensitive region of the probe, when it is not being sucked up. By sealing the NMR tube in in a reproducible way I hope to avoid making this adjustment all the time, but it will be necessary initially and occasionally thereafter. The height is changed by screwing in/out the three knurled screws at the extreme top. Adust one, then adjust the other two to equalize the force on them, changing height about 2-3 mm at a time. Keep track of what you do by measuring the distance between the top surface of the Helmholz coil assembly and the bottom of the flange that holds these screws. Lock on the sample, optimize the shim carefully including the phase, keep track of the lock gain, and keep to lock power as low as possible while the lock gain is near maximum. Note the lock signal size, change the height, optimize shim settings and lock phase, try again. Very tedious, find the highest lock signal. (It might be better to use the proton or phosphorous signal instead but we have not done so.)

If you keep track of the height as described below, then if you do not get the sample position just right, as determined by measuring the distance from the sample center to the bottom of the nmr tube adapter, you can compensate for this error by changing the height of the tube at the top. Be sure to remember that you did so, and change it back for the next sample. Then the tedious adjustment can be avoided or at least speeded up. (This works for us).

You may find that the vertical screws that go through the top of part (8), to adjust the height of the shuttle tube as just mentioned, end up at an inconvenient heigh, so that the screws have to be screwed all the way in, or all the way out, or nearly so. (In the former case spare blocks of metal can be used as extra spacers). If so, you can change the wat you cut the NMR tubes to length and cement them together, or else shorten or lengthen (re-make) the rods RCB. Sorry.

Clamping the stop tube at the correct height. The following applies when not using the Helmholz coil to get to very low field. Before you run, measure the overall lengths of the three stop tubes or refer to your notes if you have already done so. Pick a likely stop tube and call the overall length S. Before dropping each sample+shuttle in, measure the length N from the liquid sample center to the top of the upper O rings on the shuttle. Measure the height C of the Helmholz coil assembly from the magnet top flange to the support assebmbly top (note that this is 9mm less than the actual height of the coil support in our design because the legs doe not sit on this flange.

For each R₁ run, first decide what Tesla you want to get. Refer to the graph that tells you what the depth D is below the top of the top plate of the magnet, to get this field. You want to calculate the height H to set the top of the stop tube above the Helholz coil assembly top. By drawing a diagram on a piece of paper you will find the relation D=N+S-C-H, or H=(N+S-C)-D. So you calculate (N+S-C) when you put the sample in, and write it down, and subtract D to get the height H for each Tesla value. Now with the meter stick and a drafting square you try to clamp it at this height, as tightly as possible. The put a piece of paper adhesive tape on the stop tube to easily see, during a run, if it is moving as a result of the shuttling. If the height H is calculated to be negative, or positive but too small to get two cm or so that is needed for the clamp assembly, then you have to use another stop tube of a different length, or you may not be able to reach the desired field. If H is calculated to be too high for convenience, or the stop tube will not pull out that far, you may have to use a longer or shorter one.

Running at very low field (below about 300 G.) Use the shortest tube, which will permanently live on the upper end of the brass extender tube. It has a commercial stainless collar for a clamp. It is supposed to be set so that the sample is sucked up to be centered half way between the two helmholz coils. We set this by eye, in the magnet, by unplugging the electronics from the wall power (which opens the normally-open vacuum solenoid valves) and connecting the vacuum line to the wall, and opening the suction supply valve on the wall (and setting the vacuum regulator, if necessary to a couple of inches vacuum). If the sample is not in the right position half way between the Helmholz coils, the clamp has to be removed from the brass extender tube, and the clamp loosened with a hex wrench and reset to make the position correct.

To set the field, a parameter "cur" is set in the sequence FCLtonep, for example, to a number 1 to 255 according to a table that I supply, of cur vs field. Not that setting cur to zero turns off the field cycling entirely. You can do experiments with several values of cur, and one at 11.7 T, in que'd experiments.

This completes description of the parts that were made by a professional machinist. Most of the rest was assembled by A. R. with only hand tools and a drill press.