The black magics of energizing a Magnet

The energization of an SC Magnet, or “running it up” in service engineer’s parlance, has been always considered as something with an high percentage of black magic involved. Service engineers themselves did nothing to attenuate this impression; they instead increased it, telling numberless tales of inexplicable behavior of magnets, and adopting exotic operating methods, as a magnet was better a young girl to be convinced to behave properly rather than a metal can with some wire inside.

As for myself, here I admit it: I have always been somewhat afraid of magnets; the very first magnet I ran up was a Varian R2D2 300 MHz mono filament, and I had a very short (4 hours) training from a serviceman in hurry. The power stick get frozen in the plug and it took me up to 4 a.m. (yes, deep in the night) and a cylinder of Helium gas to get it free. The magnet quenched 3 times during run up, I was really frustrated and was considering to quit NMR altogether if my future business would be to work on such a thing. But, with time, I took my revenge: this very same magnet has been cut open and I took part in the operation [link all’articolo].

Years passed by, I grew old and gained some experience; then very recently I had to re-energize a 300 MHz Varian R2D2 again, but multifilament this time, and I knew was a good guy from a previous run up. I took my time, did every step with calm and adopting better methods. The result was probably the very best Varian R2D2 magnet I ever ran up, and final shimming to get good lineshape took minutes instead than hours.

The entire process to energize an SC Magnet is the sum of 5 steps:
1) Precool the cryogenics chambers to liquid Nitrogen (LN) temperature
2) Cool the inner chamber and the SC coils to liquid Helium (LHe) temperature
3) Charge the coil to the nominal current to produce the required magnetic field
4) Fine adjust the current to locate NMR signal at operating frequency
5) Fine adjust the SC “shim” coils to obtain required lineshape
Every step requires the proper procedure and some degree of attention; the first steps will be shortly explained, with some more details on the last two ones.

1) Precool the cryogenics chambers to liquid Nitrogen (LN) temperature
It’s the longest and more boring step; first the LN chamber is cooled and filled with LN, and this will also precool the heath shield between LN and L He chamber, and the L He chamber itself. While this goes on, the LHe chamber must be constantly flushed with He gas to keep any air or moisture out of the chamber, as they will freeze and solidify as the temperature goes down to -269 °C.
At beginning, all the magnet structure is at room temperature and all LN introduced will immediately evaporate with violence; hence, care should be taken to prevent too much pressure to grow up in the inner chambers, since this would very much slow down the process, and also create the possibility to damage the thin walls of the chambers if pressure rises too much.
When the LN chamber is full, the LHe chamber is also filled with LN to precool it, and this will greatly reduce the amount of LHe required for final cooldown.
Once both chamber are full of LN, the magnet is left to cool down for at least 3 hours, or much better overnight.
Note that the bigger is the magnet, the longer is the time required to precool, and high field magnets require a somewhat different procedure.

2) Cool the inner chamber and the SC coils to liquid Helium (L He) temperature
First, the LN must be completely removed from the LHe chamber, since it will freeze and solidify when LHe is introduced; some solid Nitrogen over the SC coils will almost for sure produce a quench at some point during run-up, and this step requires close attention. Usually the “pump and flush” technique is used, i.e. the LHe chamber is slightly vacuum pumped to evaporate the LN present, then flushed with He gas; 2 or 3 cycles of this process will remove any LN still present.
LHe is now introduced in the inner chamber, as done in a normal refilling, but again paying attention not to build up too much pressure in the chamber.

Clouds in the room


During this step, a thick cloud of vapor often grows up in the ceiling of the room, where the He cold gas concentrates, and the operator must be very careful that the oxygen percentage in the room does not decrease below the safety level.
At the very end, also L He chamber is full of L He. The whole thing takes usually a day; time can be shortened, but some price in higher LN and LHe consumption must be paid for.
Magnet must now be allowed to cool down for several hours, since the SC coils must be at LHe temperature before any current is applied, and I’m used to wait overnight; again, high-field magnets need longer time

3) Charge the coil to the nominal current to produce the required magnetic field

Let me introduce now  the Power Stick and Magnet Power Supply (MPS), very good friends of mine.

Magnet Power Supply


They enter the game now, and they purpose is to charge the coil with the required current. As easily noted, in case of R2D2 magnets they are old-time pieces of hardware, and the MPS is 100% analog electronics, fully controlled thru knows and switches. But I’m an old-timer too, and feel fine with it.


The frozen connector during run-up




The magnet has a very high inductance, several Henrys, therefore a long time constant, and cannot be charged quickly; self-inductance between the various coils (main, compensating, shim etc), very close each other, is also high, and this produces spurious induction currents in the coils that must be periodically discharged or “dumped”. In top of it, the wires and the connector of the power stick are not made of SC wire, hence they heat up quite a bit when a current of about 40A flows, and evaporate LHe quickly; since the coil must always be completely submerged by LHe, a LHe refilling must be periodically performed.


In practice, the energization or run up, what you like most, is performed in several steps; every step is smaller and shorter of the previous one, so that they require about the same time, and the final target current is obtained with a kind of adiabatic approximation. At every step, the shim coils are “dumped” and the LHe is refilled to full magnet. All this takes time, and the process requires about a day of work.
Sometimes, another small problem shows up to further slow down: a quench. But let’s talk about quench later.
Eventually, you reach the nominal current, the one carefully noted in the magnet’s factory data; so that, you turn on the spectrometer, insert a water sample in the probe, and look for a signal. But there is very little hope to see a signal.
Every magnet has is own current for resonance field, and this was first found out in the factory where the magnet was assembled; but don’t hope to find the resonance at the very same value: the magnetic environment of the room where it sits now is completely different from the one in the factory, the power supply is not the same one, the moon is in the wrong time, and the engineer is nervous. No chance!

Photo courtesy Prof. V. Vrcek, Univ. of Zagreb


A typical energization scenery: power supply on a (wooden!) chair, the Varian R2D2 magnet with protective cover and liquid Helium dewar at hand;  the drawing board on the back is very useful to write down the current at different  steps





Let’s go the the next step, a quite interesting one.

4) Fine adjust the current to locate NMR signal at operating frequency

The traditional method is to setup the spectrometer for 1H observe with a water (H2O!) sample in the probe, use wide observe spectral width, connect an oscilloscope to the output of the observe receiver and slowly (slowly!) ramp the current applied to the magnet looking at the scope. Soon or later, a FID will hopefully appear.

But time goes, the current is already higher than the value noted in the factory data and no signal is in sight; in the meantime, helium does evaporate, time for refill approaches, and the engineer get nervous. Therefore, I apply a somewat different method.

a) Use a 50/50 H2O/D2O sample and setup one experiment for 1H observe and another for 2H observe;

b) Move the Observe BNC cable from observe to lock channel on the Probe; this uses the lock circuits, tuned on 2H frequency, for 2H observe without any need to retune the probe.

c) Close the magnet in persistent mode (“Park”) the magnet at nominal field current, i.e. the value found in the factory or the value used in the last energization in case of a magnet moving or quench. With the magnet “Parked” the helium boiloff is minimal, and you have plenty of time.

d) Search for the NMR signal changing the observe frequency instead than the field; the frequency steps should be compatible with the spectral width used, say 50 kHz step for 100 KHz spectral width, to be sure you don’t miss the signal. Soon or later a FID will appear; it will be very probably with very bad shape as the superconducting (SC) shim coils will not have the optimum current (or no current at all!) but at least a spike and a very broad line) must show up. Now you can play some trick

e) Adjust the frequency so that the signal is on-resonance (or close to it) and note the value. A quick linear proportion will give the current value for resonance at nominal 2H frequency

f) Shim the SC coils to improve FID decay and lineshape; do not spend time looking for optimum, the main current has to be readjusted and the SC shim currents will change too.

g) When a satisfactory situation is achieved, change the 2H observe frequency back to the nominal value, get in control of the magnet current and slowly ramp up (or down, as required) until the NMR signal at the exact frequency is reached. Give the magnet some time to stabilize and close the main switch; as well known, the magnet will be much more stable when the signal is approached decreasing the current rather than increasing it.

This is of course a very basic procedure, and every magnet type has his own variant; almost all require the “overcycling”, which is to “park” it at a current slightly higher than the nominal one, leave it there for some time, and then go down again and find resonance; and, of course, how much overcycle and for how long time is variable from a magnet to another.

5) Optimizing Magnetic Field Homogeneity

This is the last step and the most exciting one. The target is to obtain a magnetic field with the very same value, i.e. homogeneous, at least in the volume of the sample or, better, in the biggest possible volume all around the sample. The question is. how much homogeneous? Well, if you want to obtain a 0.3 Hz line width on a 300 MHz spectrometer, you need an homogeneity better than 0,1 part per million, not a simple task. Such a degree of homogeneity can be obtained only using the NMR signal itself as detection and measuring device, and once you have located the H2O (or D2O) resonance several steps are needed to reach it.

1) Load on the S.C shim coils the current values found in the factory when the magnet was first energized; these values were obtained plotting the magnetic field with a moving probe head and applying the proper corrections, a true specialist’s job.

Once and for all: S.C. = Super Conducting; R.T. = Room Temperature

2) Optimize the current in the coils looking at FID decay and/or water line shape; on almost all spectrometers it’s possible now to display the FID and/or spectrum with a numerical value showing the height of the peak or the level of the FID, and this makes things much easier.

This was not always true, the early models (Varian XL series, to say one) were not so clever, and the time consuming field plotting method had to be adopted, using a very primitive shimming probe with a bubble water sample. The only system capable to do real-time acquisition was our MiniFID (link all’articolo) and proud of it.

Since only 2 coils can be controlled together, and there is always some interaction between different coils since they are wound very close together each other, the adjustments must be cycled several times. In top of it, Z2 and Z4 gradients coils, the most critical ones, changes the main field quite a lot; therefore, if you have to extensively adjust one (or both!) of them, the signal shifts, and the main current must be readjusted. I hope that now you understand why the energizing service man is somewhat nervous when this process is on the way.

It should be possible now to obtain a lock (of course use a 50% H2O/D2O sample!), which allows to proceed with the final step.

3) Obtain the lock and shim the S.C. coils to maximize the lock level,as they would be the normal R.T. coils in the routine homogeneity adjustment; of course they are much more touchy, and with a somewhat longer time constant to react, so that one has to change the values slowly and wait some time to see the effect on the lock level. The Z1, X and Y gradients are quite critical, a very small change produces dramatic results, and some practice is needed, but the method is always the same: change the shim current values (following the proper method and always noting down what you are doing!) and evaluate the results. Do not hurry, the process needs time, a couple of hours is the absolute minimum, but remember the energizer’s golden rule: one hour spent in S.C. shimming is one day saved to obtain good line shape with R.T. coils shimming! This rule cames from Mr. Denis Thompson, the designer of Varian R2D2 S.C. Magnets, and it’s 100% true.

No need to talk here about optimizing homogenety methods in details, there are countless texts about it; if anybody is interested in theory, mathematics and practice of shimmimg, this is a good source:

G. N. Chmurny and D. I. Hoult, The Ancient and Honourable Art of Shimming, Concepts in Magnetic Resonance, Vol. 2 1990, 131-149

The result can be very easily checked, just do one pulse on H1 Observe and have a look at H2O line shape: the line width ad mid height should be definitely better than 1 ppm; maybe even more important, the line shape must be “beautiful” i.e. very symmetrical and without feet and humps in the bottom portion, and this implies an optimal adjustment of Z2 and Z4. The method I adopted time ago was to use the Doped D2O sample, (D2O + 1% H20 + a little GdCl3 dopant agent) with calibrated line width of 4/40/80 Hz (4 Hz @ 50%, 40 Hz @ 0.55%, 80 Hz @ 0.11%); a 4 Hz linewidht means 0.25 s relaxation time, so that you can pulse at 0.5 s interval and adjust S.C. shims in short time. I am satisfied when I reach 50 – 100 Hz line width @ 50% and “beautiful” line shape, as I know that with some easy work on R.T. shim coils the line shape specifications will be met. But real experts do it better: Mr. Ivano Rastelli, master magnet energizer and shimmer from Agilent, never stops the process before reaching  0.1 ppm at 50%, or 30 Hz for a 300 MHz.

Well, now you got it! Job completed? Not jet, the final step is missing:

4) Remove the power connector, or the “stick” in energizer’s parlance. This piece of stainless steel, fiberglass and copper wires is at LHe temperature at the bottom and at room temperature at the other end from hours; the upper part is completely covered with ice, and it looks like being completely soldered to the magnet. Well, you definitely have to take it out, and without disturbing the magnet too much, and definitely not shaking it. Once removed the ice and dried the upper side with a heath gun, the (in my case, nervous) service man catches the connector and applies force; if the stick comes out sweet and easy, the man is happy and the job is really done. If it resists, even with increased traction, reactions are different: someone cries, some other displaces his best repertory of bad words, some other extracts a big hammer from the tool kit. There are  tricks to convince the stick to emerge, no need to discover them; as mentioned before, in the worst case I had it took me some 5 hours of work to see the stick coming out from the depths of the magnet.

The final thing to mention is the ever present possibility of a “quench”. The sudden loss of superconductivity in the coil is called “quench”; all the stored energy is released in a very short time, which produces the evaporation of at least a good portion of liquid Helium in the magnet chamber. A quite noisy and dramatic event, with the room almost full of very cold Helium vapor and very little oxygen left, so that the poor service man has only one option: run out of the room, and fast. A day or more of work gone literally in smoke. A couple of shots give a pale idea how it looks like.












Therefore, do not pronounce the word “quench” in proximity of someone running up a S.C. magnet: his reaction could be harmful.

For those interested, have a look at “Quench!” article (in italian!).

That’s all. folks. See you on the next page!







Leave a Reply

Your email address will not be published. Required fields are marked *