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

Return to Index

Part II, Upstream from the Stop Tube

Most of what follows is likely to be modified by your circumstances. In our case the small size and low ceiling of the 500's room dictated that our system be very portable, whereas some labs could have permanent setups at or near the instrument. This is especially likely for shuttlers to be used with physically larger magnets (600 and higher) as will eventually be desirable.

Furthermore, I recently concluded that it is not needed to have the pressure/vacuum introduced at the top of the stop tube, requiring, or nearly so, that a set of heavy valves by lifted up and down each time the stop tube is used. Instead, the air I/O can enter at the side into annular space just above the top of the main glass shuttle tube, and transmitted to the inside of the stop tube by many small transverse holes near its bottom. The upper end of the stop tube might be a solid rod of around 1/4 inch diameter, and would not be completely straightforward to design. This is worth trying because it is annoying and fatiguing to deal with the air I/O as it is now done.

Engineers will hate our electronics system based on monostable (one-shot) timing and be tempted to use a modern system based on an embedded computer/timing system. But the present system is completely adequate. A problem might seem to be that the timing settings cannot be stored and reproduced as are parameters on commercial instruments. Perhaps, but air pressure parameters could not be documented so easily, and because of subtle differences between instruments they could not be easily ported.

For these reasons we will not present such details as we did above.

Section TW TOWER and components on it. The tower is useful probably in any case. We bought a heavy aluminum 84x19" full-size rack made by Bud (they used to make aluminum passenger railroad cars and "Budliners"), strongly recommended, with many 12-24 tapped holes on both sides of its vertical aluminum beams (the electronic supplier that sells it does not sell these screws which are unusual in the US; you get them from McMaster Carr or equivalent.) It has two very heavy (2 1/2 x 6 x 1/4") tranverse aluminum angle base pieces which I mounted upside down and on both sides of the rack's vertical channel pieces (The front of the tower is defined as the side of the rack that will be closest to the magnet). Then to hold the front wheels I cut two pieces of 1 1/2 by 3/16 angle aluminum 25 inches long and fastened them with at least 4 1/4-20 screws to the top (what would have been the bottom) of the base piece, extending out 11 1/4 inches from the front of the rack's vertical channels It thus extends 11 3/4 inches out from the back surface of these channels. The 1 1/2 " vertical surfaces of these angle pieces touch the inner edges of the vertical channels. Two 11" x 1 1/2 x 3/16" aluminum angles 11 long are screwed to the top of the front base piece angle, next to the long 1 1/2 " angles, so that they extend out exactly as far as the longer ones beyond the front. The bottoms of each pair of angles form a 3" wide surface to which I attached, at the most frontward points, stainless steel front casters. To mount the back casters I screwed a pair of 11" long 1 1/2 " aluminum angles to the base piece, next to the longest angle, toward the outside of the rack and ending in back, exactly even with the end of the longest piece. These are also screwwd sideways to the back ends of the long angle pieces To the bottoms, at the back end, of each set of these angles I attached a rear caster. The latter were swivel casters, but the front casters were not but were oriented toward the front; this is desirable because the front of the tower is less stable because the casters have to be close together to fit under the magnet The result is a rack which fits through all doors in the building, rolls freely, does not tip over very easily, and to which all kinds of things can be fastened.

An old piece of heavy plywood was hand-cut to fit in the space between the horizontal angle pieces, to form a wood base, and heavy cider-blocks were placed on it, leaving room for the reservoirs (below), to increase stability. The reservoirs are made inexpensively from polyvinyl chloride (PVC) 3" diameter tubes normally used for sewer piping in houses, and standard fittings, cut with a wood saw and cemented with standard cement for this purpose. Each is about 6 ft high. Heave plywood boards about 4" wide are screwed onto the outsides of the rack, extending backward, and connected at the back with a cross board. This U-shaped assembly is vertically a few inches below the top of the sewar pipes, and its back board is even horizontally with the back of the base of the rack assembly (with the bottom casters). The three reservoirs are attached to this assembly with elastic "bungie" cords available at hardware stores. Small strips of 3/4" plywood nailed into the base wood constrain the bases of the reservoirs, forming a 3" wall at the back and extending a few inches along the side, to constrain the bottoms of the reservoir tubes.

The bottom of each reservoir is closed with a standard cap. At the top, each tube has standard fittings (that can be found only in big stores or catalogs like that of McMaster Carr (with great care)) to reduce the size, and couple to a brass T of nominal 1/2" size (inside diameter ~1/2 "). The long side of each T is vertical, and the upper vertical end is coupled to a large quick-disconnect fitting, to go via a large flexible tube to the valve manifold (below). The side connection of each T goes to a smaller diameter quick disconnect and flexible tube that crosses via an overhead trough to the rolling table. This plumbing will be discussed more later.

A wooden trough that is instantly removable connects the tower to another smaller (49x19") Bud aluminum rack that sits on a relatively inexpensive white plastic and aluminum rolling lab table. The trough is made by nailing (with stainless steel nails) two "one by fours" with their four inch surfaces vertical, to a same-size bottom with the four-inch surface horizontal (and no top cover). The length of the trough is determined by the size of the laboratory and the need to keep powers supplies away from the magnet; ours is about six feet long. It is convenient (for moving in and out) to have the sides longer than the bottom by about 1 foot so that the ends form forks that can snag the wires and tubes that the trough must carry, during installation. The ends of the tops of the side wall pieces of the trough are tastefully cut off with a 45 degree angle cut. The trough is supported at each end by horizontal ~8" light aluminum angle pieces (1x1x1/8) bolted at convenient places on the sides of the two racks, with short (3") pieces of the same angle stock, at their outer ends, to keep the trough in place but allow it to be in stalled and removed easily. The bottom of the trough is thereby about 76" off the floor when it is installed. No wires or tubes run across the floor to the magnet or tower. The trough is vastly superior to commercial metal or plastic wire troughs.

The heavy (1 1/2 by 3/16) pieces of aluminum angle stock that are supplied with the tower's rack, for its top, are not used there. Instead these and lighter 1/8 thick pieces of aluminum angle stock about 20" long are bolted horizontally across the front or back of the tower at various heights to support anything you want. Most important, to support the valve manifold, two of these are mounted across the front of the rack about 3 feet apart vertically, with 7/8' diameter holes in their centers. A long (7-8 foot) 3/4" diameter rod passes vertically through these two holes to support the valve manifold and let us raise and lower it . I made Delrin bearings for the rod out of 1 1/4 x 2 pieces of Delrin 1/2" thick with 3/4 holes that were screwed in above the 7/8" holes in the aluminum with small screws. The 3/4 rod had the valve manifold attached at its upper end, and the entire rod was raised by a rope attached to a short foot at its bottom. The rope went through each wheel of a pair of double pulleys, one attached to the foot of the rod and the other attached high on the frame. In this way we have a 4-times mechanical advantage in lifting the heavy manifold. A nautical cleat fastened to the tower at a convenient level was used to fasten the end of the rope. Thus the valve manifold and rod, total weight around 15 lb, could be easily raised and lowered next to the magnet, and lowered to get through the door.

The valve manifold will be described shortly. It is attached rather loosely to a platform formed from the top of two short horizontal (~6" long) pieces of angle stock running roughly forward to back. The vertical sides of these pieces point downward and are attached to an assembly of aluminum pieces that I will not describe in detail. They form a rectangular box about 5" wide on both dimensions and 6" high. Near the lower end there is a horizontal 1/8 thick aluminum plate with a 3/4 " dia. hole, for the long 3/4 rod, described just above, to pass through loosely. Another similar plate and hole is located inside the box about 5" above the first, for the 3/4" rod to pass though loosely also. The two holes are about 4 1/2 and 10 inches from the top surface of the box, on which the manifold will sit. To get through doors, the 3/4" rod passes through this box, extending about 3" above the top of the box, even with the tops of the valves.

When moving the tower, a commercial collar (whenever possible use these, stainless steel ones, not the cheap set screw type but the ones which squeeze the rod with a single cap screw, it is always cheaper to use them than using some machining) fixes the position of the valve manifold vertically on the rod, and also the rod is limited in how far it can go down by the height of the highest fixed hole (3/4" bearing) fixed on the tower's front, through which the rod passes. The heights of the collar and of the crosspiece that holds the bearing are set so that the 3/4" support rod is about even with the tops of the solenoid valves, as mentioned, and both are just below the top of the lowest door through which the tower has to pass when the rope is released and the manifold is as low as possible.

Once the tower is in the 500's NMR room, we can insert a 6" wood spacer between the collar and the box to eventually get the valves very high. The spacer is made from two 6" long pieces of 3/4" plywood attached to a third narrower one so that they form a groove 6" long, and can surround the 3/4" rod snugly but not tightly. The (vertical) groove between the outside pieces is about 3" deep horizontally, and there are hook-like 1/2" knobs cut on the outer edges of these pieces, at both the upper and lower ends, that tend to keep the spacer in place. When it is, the valve manifold is pushed upward 6" relative to the rod, compared to when the spacer is not there, so that the end of the rod is 3" below the bottom of the valve manifold, just barely going through the top one of the two 3/4" holes in the support box for the manifold.

Section PL, plumbing. This covers all the plumbing from the top of the stop tube back all the way to the sources of pressure and vacuum. Very tedious. The purpose is, first, to provide regulated pressures (low and high) and suction, and then to connect these via electric solenoid valves, in the correct order and time, to the common output connected at the top of the stop tube. At this point you must READ PARTS of the MRC article which I will not repeat here. It even gives the valve manufacturer's part numbers but there was a misprint: the valves are Asco numbers 8262G212 for pressure and 8262G264 (thanks to Boris Itin for finding this). Later I will give an equally tedious description of the electrical system that mainly turns these valves off and on.

General remarks: it is good not to have leaks in the system before the solenoid valves (though very small ones can be tolerated). Downstream (toward the magnet) of these valves small leaks are not at all a problem because they are not likely to be large compared to the flow through the shuttle tube (however, note that this latter flow is nearly shut off as soon as the shuttle is at either end of its travel), and the valves are open usually only for a small fraction of the cycle.

The valve manufacturer recommends that the valves be mounted with their magnetic axes vertical, to aid in clearing particle dirt, but this is not a big problem because the air is well filtered, as is the vacuum (joke). I have had problems after modifications or glass breakage. In the latter case the shuttle itself often sticks in the glass tube and it has to be taken out and washed with water and dried with paper towels pushed through the shuttle tube, after any NMR tube brakage. We recently had one valve stick for no obvious reason (it was the last one to open, to let the pressure equilibrate to ambient pressure) but this fixed itself as soon as I had time to identify the valve, and to determine that it was not an electric problem. I followed this recommendation, and also arranged the valves so their bottoms were also on the same plane so they could sit on the platform formed from two pieces of angle stock, mentioned above. It is also desirable to make the manifold as compact as easily possible, and clustered around the 3/4" aluminum tube that supports it (not in a straight line, for example) and with space for this rod to pass through between the valves, vertically, close to their center of gravity. It is also a good idea to be able to add more valves later, easily. It was hard work to design the manifold. In general it was a poor idea to use plastic plumbing which breaks, so brass was used everywhere except as noted.

The entire manifold can be disconnected mechanically and electrically in a few minutes, and lifted off for inspection and maintenance. Unfortunately a strong person has to get on a ladder to do so. Getting the manifold back on is especially strenuous. The valves seldom are expected to need replacement (although I have one spare of each type), but their internal parts have to be inspected for wear probably ever half million cycles and this is easy if they are vertical. The Asco Company seems to think you will know how to take them apart. You have to slide off the top nameplate horizontally with the help of a screwdriver, and you can eventually figure this out. Then the solenoid coil slides up off the top of the valve. A spring then has to be removed to get access to the large hex nut that has to be unscrewed. I invested in a large wrench of the "box" type that encloses the nut like a socket wrench, but it is a single assembly. I took apart each valve, and put it together again, before assembling them, so that they would be easy to take apart later. Then you can take out the guts. BE SURE NOT TO GET CONFUSED ABOUT HOW PARTS GO TOGETHER, ESPECIALLY FOR THE NORMALLY OPE VALVES, AND INCLUDING THE NAME PLATES. ONLY DISASSEMBLE ONE VALVE AT A TIME.

Asco does not tell you what to look for when inspecting their valves, so I can't either. I have not had to replace one yet. I have a repair kit for both of the types of valves I used.

We have an electronic counter (Veeder-Root) that has a battery that will last 7 years and counts the number of cycles, to help us decide when to opent the valves up for inspection.

Ordering the parts for the manifold (from McMaster-Carr) was a major job. Read what follows carefully when you do are ready to build it, and not before, you will forget what you read.

I do not use Teflon tape for plumbing even though it looks good and does not soil your hands; I use standard plumber's compound to save time and frustration. Some of the details may be important, and changing them could complicate easy construction of the manifold.

Tour through the plumbing: I now describe the gas plumbing in tedious detail, going upstream, starting from the top of the stop tube. The stop tube itself ends at the top simply cut off on a lathe and then beveled slightly on the outside ("chamfered") to facilitate connecting the to the manifold. Upstream of the stop tube, everything remains connected together forever. The first thing upstream is a "vacuum quick disconnect" that adapts any 5/8 tube via a compressed O-ring to a brass solderable tube. They can be bought cheaply from vacuum supply houses. It is adapted to nominal (US size) 1/2 plumbing by drilling through a commercial brass plug of this size and soldering it in. This is screwed into one side arm (oriented vertically) of an aluminum 1/2" US nominal Tee (aluminum to save weight, this is a big Tee). The opposite vertical side arm has a similar plug screwed into it to which is adapted a silicon pressure gauge (Honeywell XCA505DN, marketed by Newark Electronics, 5 volt supply, +- 5 pounds per square inch). This is adapted to the brass 1/2 nominal plug with a plate that pushes it onto a questionable O-ring seal with the gauge's surface. These gauges contain parts that break if you drop the gauge. After you receive and study the gauge you will figure out how to seal it. It is a permanent part of the manifold assembly so it will not be likely to be dropped.

All plumbing of what follows is 3/8" US nominal plumbing and 3/8" (~11 mm) inside dia. rubber, copper, or plastic tubing except the solenoid valves which are only available as 1/4" nominal, and the 3 flexible tubes that are about 60 cm long and connect the movable manifold to the fixed tops of the reservoirs. The long (~8ft) plastic or rubber (for vacuum) tubes that go from the reservoirs to the table, via the trough, are also 3/8" inside diameter reinforced Tygon or thick rubber, not because pumping speed is critical for them but to avoid kinks that could pinch off the air flow.

Returning to our tour, the side arm of the aluminum Tee is adapted connected with a male (M) to female (F) aluminum elbow, (known as a "street elbow"). This elbow is important: it does not affect the speed of gas flow much but it allows a higher degree of flexibility of the connection between the tower and the stop tube. (Note also that the entire manifold can swivel about the 3/4" aluminum rod, on around a vertical axis, which we arrange for the same reason.)

The other connection of the 1/2" aluminum elbow is adapted with standard brass fittings to a 6" length of rubber tubing whose other end is adapted to the central Tee of the manifold (which is female gender). This is the side-arm hole of a FFF Tee of the type which looks as if it were milled from a square block. (Tees have three connected holes and are sold in three forms, FFF (all female) and FFM (or is it MFF?) and FMF which have one out of the three connections being male and with FFM having this male being one of the two end connections, and FMF in the middle.)

The overall shape of the manifold is more or less like a two-pronged fork whose handle is the rubber hose already described, oriented towards the magnet, then a perpendicular cross piece made of three block-type Tee's, and two arms pointing away from the magnet. (see fig. PA). The six valves are connected, three on each side, to the outer sides or ends of the arms. The cross piece consists of a central FFF Tee, already mentioned, screwed directly into FFM Tees via their M connections, and all in the same orientation. The two arms each consist essentially of a single standard (non-block) FFF Tee with its long part pointed back away from the magnet coupled to the middle F connection of one of the two cross-piece Tee's. All five Tee's are oriented so that the plane defined by the 3 connections is horizontal. The side-arm Tees and the two cross-piece Tees are not directly coupled, they are connected by a short (but not he shortest available, you need some space here) M-M connector (a "stud"). Now if you put it together, there are 6 female connections always connected to the rubber tube via the Tees: two at each end of the arms, and four pointing out though the relay. The current is off on any relay when its control line from the timing card is up (+5) so the strobe for sideways away from the manifold. The solenoid valves will be screwed into these holes via short MM 1/4 nominal to 3/8 nominal adapters. But wait, don't put it together now, first you have to screw the valves onto the T's, at least the side ones. And before you do that you have to read the next three paragraphs.

First you have to carefully read what I wrote in MRC to determine which wire driving the valves go to + voltage and which to -voltage!!! These wires are both red and you can't tell which is which, each valve has to be tested!!!. (A green wire on each valve is supposed to go to ground for safety reasons but is not so important for these DC voltage valves). IMPORTANT!! LABEL the wires after determining the one that is +.

Second you have to read the obscure paper sent with the valve to understand which way the gas is supposed to flow in the valve, indicated only by letter on the two ports. This may not be so important if you use the valves I suggest that are rated for considerably higher pressure than the pressure you are likely to use; but you never know. Obviously, for the three normally closed valves used for pressure the flow direction will be toward the rubber hose; but for the other three, that is two normally open for vacuum at the start of a cycle, and one normally closed for atmosphere relief at the end, the flow is away from the rubber tube. Mark these valves before assembling the manifold.

OK, after several hour you got it together. Plan now how to attach it loosely to the aluminum platform mentioned above, with a transverse scraps of aluminum angle stock and two long screws. Perhaps don't build the platform until now, after further thought. Before, this plan and install the plumbing to/from the three reservoirs. The valve that releases to atmosphere at the end has nothing connected to its female connection hole, not even a plug. The low pressure valve is connected to medium wall 1/2" inside diameter tygon tubing with standard fittings and this and the remaining two large flexible tubes are clamped semi-permanently with screw-type hose clamps. The remaining two lines from the reservoirs, vacuum and higher pressure, each go to a pair of electrically and functionally parallel solenoid valves. I used copper tubing to connect the upstream ends to the hoses, to get a pleasing compact connection rather than a multitude of bulky Tee's, at the cost of a slightly complex connection procedure (and a similar reverse disconnection, if the valve bodies ever have to be disconnected which is unlikely). First I assemble copper adapter elbows, 1/2" nominal pipe thread to 1/2" nominal copper "sweat" (solder) connections by soldering 4 of them onto 4 pieces carefully cut to be equal in length, about 3" long. These were then screwed into the four valves (with pipe compound) on the side that was to be further from the rubber tube to the magnet. Then, with the four tubes pointing downward, the four solenoid valves were coupled to the Tee's as shown in fig. PA. Finally the Tees were all connected together as already described. Then I soldered the lower ends of the two pairs of the nearby copper tubes together and each pair to a short horizontal output tube, using two short pairs of copper tubing and two each of an elbow and a Tee for the entire solder part of the job. The long flexible tubes that go down to high-pressure and vacuum reservoirs are clamped directly onto the two open ends of the short copper tubes with screw-on tubing clamps.

These three large diameter Tygon or rubber tubes that go to the reservoir have quick-disconnect pressure connectors (McMaster Carr) (not to be confused with the vacuum quick disconnect used above at the top of the stop tube). They could be eliminated in favor of screw-type hose clamps, but use of quick-disconnects encourages removal of the manifold for inspection. (Here and below, "quick-disconnect pressure connectors" refers to a type of connector that usually has a one-way valve inside one side which is normally the high-pressure female side; but in our case we used so-called fast-flow types in which this valve is not included. For the particular variety we ordered, the female's were brass but the males were brass-plated magnetic steel, and we had to re-order stainless steel ones.) The quick disconnects used here were 1/2" nominal size. For these connecters in general we used an orientation for gas flow from male to female; as a result only one of the two pressure pairs of connectors has to be color-coded to assure proper connection because a vacuum line cannot be connected to a pressure line. Each of these three connections are standard, using hose clamps and barbed tubing connections where needed. The corresponding smaller lines to the relay rack on the table, via the trough, are nominal 3/8" size and entirely similar otherwise.

The vacuum reservoir also has a small diameter side-connection to a 4 foot small-diameter hose, going to the sample sucker. (An annoying feature of our present system is that the vacuum lines must all be connected and the electronics plugged in to use the sucker. This could to a considerable extent be alleviated with a small reservoir in this line having a one-way valve at its downstream end).

The three long supply hoses are semi-permanently connected at the table, using hose-clamps to the three vacuum/pressure gauges (see MRC for the types). Pressure of about 20 PSI comes from a regulator on the lab wall, fed from the spectrometer's carefully filtered 80 PSI air, with a gate valve in front. A small tube connected reversibly to a taper fitting at the wall regulator goes to the higher 0-10 PSI regulator for the higher pressure, and feeds the lower pressure 0-5 PSI regulator, as well as the line to the tower. Manual pressure-relief valves at the output of both regulators are useful for setting the pressures. All these plumbing and the vacuum connections are straight-forward. Unfortunately all vacuum gauges sold in the US seem to cover only 0-30 PSI vacuum and are marked in inches of mercury. These valves and gauges are mounted on recycled pieces of rack-panel, on the upper part of the relay rack that sits on the rolling table.

The vacuum originates in an oil-free pump (type given in MRC). This is a vane pump and can be used either for low vacuum or pressure, apparently. It has been trouble free but I have a repair kit anyway. It is in a sound-reducing box in the lab down the hall where I do off-line testing. The box is made of 3/4 roofing plywood to whose inside is cemented the fake anechoic chamber lining made of cheap foam (left over from the treatment of our 500 MHz NMR room). This is not a simple box but its internal top extends not quite all the way across, and an outer top covers all but a small section on the other end, and is about 4" above the inner top, so they form together a duct. This duct is lined with the anechoic foam. The box has a cooling fan mounted on a hole in the lower part of one wall to draw air into the top duct, very important, and should have a temperature turn-off switch to avoid a fire in case the fan stops and the pump motor blows up.

We had a vacuum line installed for this pump down the hall to the NMR room, with gate valves at each end. The vacuum is sensed by a standard electromechanical regulator at the pump, which turns the pump on and off to keep the vacuum around 1/2 atm. This switch tends to stick in the on-position so we installed a solid-state relay to run the pump, controlled by the regulator switch. We also have a pressure line installed, to ship the dried air pressure back from the NMR room to the development lab. It is not connected yet so we use standard nitrogen gas tanks to supply the brief tests we usually do in this room.

SECTION EL, electronics. Tower. The only electronics in it is a +5 volt regulator chip that converts the 12 volt relay supply voltage to 5 volts for the silicon pressure sensor . All lines go to two ten-position terminal strips which also hold the 5 v. regulator. These terminal strips are bolted to the vertical arn of a piece of 1x1 aluminum angle stock, that also holds down the valves on their platform by means of two screws through holes in the platform. The 12 wires from the relays are screwed to these strips and then all the relay + wires are connected together to the input 12 volt supply. (see Fig. EA), lower section. The two pairs of minus wires from the valves that operate in parallel, for the vacuum and the high pressure, are screwed into the same terminals of the blocks. Back-biased power diodes are installed here between the four - control wires and the +12 volts, with the cathode side to +12, to protect the solid-state relays. All wires but the output of the pressure gauge go through a 6 conductor cable about 70 cm long, to an archaic 8-pin connector (Cinch Jones) which will be replaced by a type that is easier to connect and disconnect. (We will standardize on a low cost 24 pin round plastic connector made by AMP, CPC series.) The signal from the silicon pressure gauge also goes through the same connector but is otherwise carried by a coax cable taped to the 6-wire cable to reduce noise. The upper end of the coax and 6-wire cables are firmly taped to tubes of the manifold, to avoid breakage.The 8-pin connector is disconnected when the tower is removed from the NMR room. The pressure signal is again routed through a small coax line, taped onto a six-wire cable after this connector. These lines then all go via the trough to the table, where they are semi-permanently connected to a circuit board.

Table. Solenoid valve control. Most of the electronics in general is on circuit boards that live in a Vector type CCA13C/90 cage. The circuit boards are mostly standard Vector type 3677 proto-boards with 0.1" hole spacing and convenient layout of connectors and pads. The cage has its own small +5, +-15 V supply module. Interconnections are via soldered wires between board connectors or 26-wire ribbon via standard IDC connectors or barrier strips, to the cage, to a few BNC cable connectors, or to the wires mentioned above.

The four minus end wires from the valves, and the ground and +15 go to a connector-less circuit board via a barrier strip, that slides into this cage and normally lives there (fig. EA, top section) This card has four nearly identical solid-state relays on it controlled via a ribbon cable from the timing card in a conventional way. The +inputs from the solid-state relays all go to a +5 supply voltage, and the -inputs go to the ribbon.

The timing control card (fig. EB) runs these four control lines and another that turns the Helmholtz coil on & off that is used for very low field operation (below). These are controlled by 7407 open collector drivers. The inputs of this card are only 3 wires: most important, a positive gate from the Varian console that coincides the "raisetime" of the sequence. It comes from spare (coax) output 1 on the back of the Varian console to the end of a long coax cable that dangles off the magnet in back. To this is connected another long coax line that travels via the trough to a coax connector on the back of the cage. The other input lines to this card select either this strobe, or an internal test strobe generated on the card. There are also three potentiometers on the front panel connected to the timing card via the ribbon connector, one of which determines the length of this test strobe, and another which determines the time between strobes. A third one controlls the time of the high pressure . These are connected via the same (and only) ribbon cable as the solenoid control lines. There are also two wires that transmit the position of the "Varian-stop-test switch on the front panel.

I will not describe how the 555 chip that generates the strobe works (bottom left of Fig. EB), you will have to get help from a local electronics technician, nor how the "one shots" (officially called monstable multivibrators) type 74221 work. Basically the latter convert a TTL negative or positive edge into a pulse whose length is determined by a soildered-in capacitor and a resistor. They are about the simplest thing you find in digital elactronics. The circuit diagram EB uses conventional symbols for some logic gates, and box symbols for the timer and one-shot units. The one-shots are in dual packages and each section is represented by a single box. Generally the pin numbers are written near the wire-lines and often the function is written just inside the box near the wire; if not, the positions of the wires are similar for similar units. Letters like (HY) near each box or gate indicates to me where the unit is located, as row H, column Y on the circuit board, to help me trouble-shoot. Generally I leave out of the diagram a 0.01 microfarad capacitor which I ALWAYS solder across the 5 volt power line to ground., right at every chip. These cards are intended for standard logic with three long lands that are for ground, and four others for +5 V. Lines going to square boxes with letters in them or numbers up to 22 denote edge connections, I use A for +5, and B and Z for ground. It is good practice to connect the distant ends of ground and + 5 together with cross-wires, and connect them with .01 microfarad capacitors, and ground the grounds B and Z to the cage, and I always do. Lines going to circles with numbers inside them denote connections to the ribbon connector.

The 555 chip generates pulses at a rate determined by a variable resistor on the front panel and the falling edge of this triggers the one-shot (EX), whose length is determined by another variable resistor. The output of this is the test stroe. It, and the Varian strobe input via a BNC connector go to a 74253 data selector (lower right part of figure) that selects which strobe to use as selected by the panel switch.

The selected +strobe (Varian strobe or on-board generated test strobe) is convert via logic to a strobe which drives the vacuum control line and turns on the vacuum (upper left, to ribbon line (8). Note however that the vacuum relays are normally open so this strobe. In fact only these vacuum, normally open, solenoid relays, of the four, have current running through them most of the time, to keep them closed (because they are "normally open"; normal means, in this case, with no current running through them strobe must turn off the current these relays is up during the selected). MRC explains why I do this.

A minor point is that this current heats up these relays, which could shorten the life of their coils or internal parts. For this reason I inserted a 3 ohm resistor between the "plus" line of the solid state relay R₁ and the + 12V supply, to decrease the steady current through these solenoid valves; the valves only need about 1/2 of their stated current to stay closed once they are closed. But to assure that the valves actually close, it is desirable to apply the full voltage to them just at the beginning of closed period, and doing so results in faster closing at this time which is the most crucial of the sequence. I do this by use of the 0.0048 farad capacitor connected from the plus terminal of the solid-state relay R₁ to ground, so that the capacitor charges during the time (~0,1 sec or more) that the valve is open (and the solid state relay is off, and the vacuum is raising the sample or keeping it up), and then when the solid-state relay closes full voltage (12 volts) is applied to the solenoid, and then decays to about 8 volts in about xx milliseconds. The resistor and capacitor are mounted on a spare location in back of one of the pressure regulators, because the capacitor is too large to fit in the Vector cage.

Now going forward in time, the negative edge of the strobe goes to the "A" input of the next one-shot and turns it on for a time controlled by a front panel potentiometer mentioned above. This applies a negative strobe to the high-pressure control line and turns on the high pressure for only a short time (~15 msec) depending on the fron t panel potentiometer setting.

The end of this strobe turns on another one shot that controls the low pressure time, set by a screwdriver-adjusted potentiometer typically to about 0.3 sec. It has to be longer than the "doptime" (see MRC).

Finally the end of this strobe turns on the last solenoid relay that opens the manifold to atmosphere for about 0.3 sec. Three very similar steps.

This board also generates two other important signals:

1.The "rear scope trigger" goes to an edge terminal (11) and then to a BNC coax connector on the rear, to trigger the storage oscilloscope sweep. It is identical to whichever strobe is being used, the internal test strobe ore the Varian-generated strobe. A TTL amplifier isolates this output. See below under "pressure monitor". This signal also goes to the Veeder Root counter that tells the number of cycles that the system has done.

2. The Helmholz coil used for very low field measurements, and located on top of the magnet, can get hot if it is run all the time. So we turn it off during most of the time that the selected strobe is off (down) and the sample is at the center of the magnet. This can be done by using the selected strobe directly (passing through the TTL inverter at (FY) and edge pin (22) to a coax connector on the rear of the Vector cage because the Helmholz circuit is turned on by a negative strobe. However, there is a delay between the time the vacuum is turned off, and the same time that the high pressure is actually applied to the sample because of delays in turning on the solenoids, then a delay in actually opening the solenoid valves, and finally a delay in filling the shuttle tube with high pressure. We do not want the field to start turning off before the sample starts to leave the upper position at the stop tube. So the time at which the Helmholz coil is turned off is delayed by "stretching" the strobe that controls the coil, by about 0.1 sec. One-shot (HY) does this; it is triggered by the falling edge of the selected strobe and turns on for aboput 0.1 sec. Its positive output is "Nor'd" by use of the second inverter connected to the output to lengthen the output at the rear connector. It is necessary to check with an oscilloscope to see if the delay in turn-off is as long as needed, and also that the current through the Helmholz coil is on by the time that the sample is at the stop tube.

Helmholz coil. fig. EC shows the entire electronics of the coil. Basically, a solid state relay turns it on under control of the output just described. It is in series with the connection of the Helmholz coil to a large power supply which is set to regulate current. The size of the current is determined by the Varian computer under control of the NMR operator.

Staring at the right side of fig. EC, 8 lines of a binary representation of the user-entered parameter "cur" are presented by use of simple commands (see Varian user programming manual, which does not tell you that the most significant bit of the number that appears is on pin 1 and not pin 8 as I initially assume!). This is a number from 1 to 255 (if cur is zero the program is written to turn off the shuttle strobe entirely, and a high field R1 run results). The current in the Helmholz coil, if it is on, is supposed to be proportional to this number, and the system is adjusted so that if cur is nearly 255, the Helmholz coil exactly cancels the main magnet's fringe field. We determine exactly what this number is by using a Hall magnetic field probe and seeing at what number of "cur" the field in the center of the coil is zero. The number cur is fed directly into the 8 most significant digit inputs of a digital-to-analog converter, while the four least significant inputs are set to zero. The output of the converter is then proportional to Cur and equals about 1 volt when cur = 255. This voltage is fed via a cable to the input of the power supply that controls its output current when it is in the voltage-to-current operating mode. The power supply is an excellent 0-10A 0-50V Kepco supply. We mounted a toggle switch on the top back of the supply to switch from the external-voltage-control, constant current mode, to the normal mode using the front panel controls. THE FRONT PANEL CONTROLS HAVE TO BE TURNEDTO THEIR MAXIMUM VALUES when the supply is externally controlled (at least for this power supply). More precisely, the current control should be all the way on (clockwise) but the voltage control should be set very slightly below 50 v or the supply mat turns itself off and will have to be turned on ). You will have to consult the manual of the particular power supply that you have, to learn how to wire this constant current mode; and be sure that you can get or buy a power supply that regulates current; high speed is not needed.

In theory the output circuit (left side of Fig. EC) does not have to be grounded anywhere but it probably works best if the + output of the supply is grounded, as it is. A back-biased diode is connected across the Helmholz coil which conducts no current and normally does nothing, but when the magnet is switched off suddenly by the solid state relay, a larger reverse voltage could appear on it and destroy the relay. (This diode may not be needed, but it is hard to get practical information on these relays. Aside from this, the main thing is to connect the input and output voltages with the right polarity This must be a DC-to DC relay, be sure not to get one for AC output. Also, one with plenty of voltage and current capability, and mount it on a heat sink or panel). A series "shunt" resistor connected to a digital voltmeter ("DVM") monitors the current through the coil and is not used much in running; it was mainly used initially, to verify that the current output was proportional to the parameter "cur".

The control voltage (off, or zero volts, to turn on the current) comes from a coax connector on the back of the Vector cage, as described already at the end of the last section. The pull-up voltage is supplied by a little modular supply of the type used to supply portable electronic consumer devices. I had some trouble with this and had to put a resistor to the + voltage in order to turn off the current completely; perhaps the 7407 driver does not turn off completely.

Slightly exotic circuitry and mechanical design is used to have the negative output of the A/D converter floated at the ground of the Kepco supply, to reduce "ground loop" problems. This is suggested by the "earth" symbols, a horizontal line with cross hatches /// below it. Consult an engineer who understands about ground loops.

Yet a third smaller Bud aluminum rack sits on a dolly made of heavy plywood and casters, and holds the Kepco supply and a medium-size rack panel on which is mounted the other things shown in fig. EC. This rack is not wheeled into the lab except if very low field measurements are planned.

Pressure monitor. The time-course of the pressure during the field cycle is monitored using an excellent and inexpensive 60 MHz Tektronix digital storage oscilloscope. Its horizontal sweep is triggered from the selected strobe ("rear scope trigger", usually its trailing edge so we can see the key part of the sequence at the end of the sample's visit to low field. We have described how the output of the silicon pressure sensor is shipped via the trough to the table's relay rack. This signal could be fed directly into the oscilloscope's vertical input, but this is inconvenient when looking at small pressure changes because zero voltage directly from the silicon pressure gauge is indicated by +3.5 volts. This is avoided by connecting three forward-biased small diodes to this point, whose other (base) end goes via a 22K resistor to -15 volts supplied by the Vector cage's supply. The a BNC connector on a panel that goes to the oscilloscope vertical input is connected between the diodes and the resistor to give about zero volts with zero pressure.

Power supply. The +-15v, +5 V supply is a small unit inside the Vector cage. A 12 V 5A open supply sits of the bottom shelf of the lab cart on the end furthest from the 500 magnet. This shelf carries the Helmholz coil in transit also. The middle shelf carries the shuttle tube and all other long pieces and miscellaneous junk. The trough and a professional-grade 6 foot ladder have to be hand carried. The voltage of 12 volts was selected for invalid reasons; 24 volts is probably better; 24 volt valves are more standard and easier to get.

Fringe-field calibration. See MRC. Fig. ED shows the operational amplifier circuit used. It is powered by two 9 volt batteries in the same box. Probably the coil wire and number of turns are not critical. It was wound onto a small brass spool formed by soldering 3 annular brass cylinders, all 1" diameter with a hole through which passed a 3/8" dia.brass tube about 1.3 meters long. Two of the discs were soldered at the bottom about 1 cm apart to form a spool for the wire; the third was about halfway up the rod as a guide. The outside of the coil of wire is about 20 mm dia and it is probably #30 wire.

After tests using stir bars, we (Dan Miner and I) inserted the coil in the magnet (with the Varian upper stack removed and no probe) and the reset button was pushed to zero the output voltage, and subsequently the integrators drift control was adjusted continuously. Then the coil was moved to different heights at different speeds and a good speed was found, fast enough to allow us to estimate the voltage change, but not too fast to give non-reproducible results. The latter may occur if the induced voltage is so high that the op-amp output is overloaded. Then we recorded a series of voltages with displacements from the center of the magnet going up, and also from the top of the magnet (at the upper edge of the uppermost flange that is permanently part of the magnet) and especially between these points. The latter voltage should correspond very accurately to 11.75 T, to allow us to convert the voltages to gauss intervals. Dan Miner then took the deflections and fitted them with a spline-fitting program to generate the graphs (Fig. EF and Fig. EG) which we use to set the height of the upper stop (see part I).

This calibration has been used for our three papers published so far, and I am not worried about the calibration, because the relaxation rates do not vary strongly except at fields below about 2 Tesla, where errors in the calibration may be small. However, as this is being written I am trying to check this calibration by field-cycling NMR at about 4 Tesla, and at lower fields by use of Hall magnetometers calibrated using a teaching NMR setup (Teachspin).

Future sections. I did not write a sections on the modifications we make to standard Varian library sequences. There is some information in an earlier report. But hey, no one else has this machine so there's no hurry, and it is easy. I will also write more on measurement of the fringe field when I have done it better.

I have been trying variations of the upper plug, making it nearly half as long, and eliminating the groove at the top. It is now made with only a cap at the bottom (where the sample is). It is still hollow but we no longer make it with such a thin wall (drilling the inside with a smaller drill), and this makes it easier to make. We no longer rely on the fact that the plug will float; the wire at the top makes this unnecessary, although it is good to have the plug as light as possible. The end sections are made to fit the NMR tubes as closely as possible and are slightly longer. (Frank can machine the clearance with better than 1/2 mil clearance, that's nearly 100 mkicrons.) The clearance in the middle is also a little less. The flats at the bottom are still used. Probably we will go back to a longer plug, or some compromise length. All this is to reduce sloshing and bubbles still more. All the above is being tried with 5 mm tubes. I am also experimenting with 5-minute epoxy for cementing in the plug, and also for cementing the tube in the adapter. This may allow a lead of only 4 hours between sealing and running, rather than 18 hours. This epoxy turns out to be less viscous as well as setting fast. The theory is that if the epoxy and sample meet only in the close clearance at the top, they will not mix much before the epoxy sets in 5 minutes. When I have more experience I will revise the relevant sections above.