"Production Line for Permanent Magnets - A case study at Fermilab"

Hannah Volk


An "antimatter bottle made of refrigerator magnets"1 could be in the future of Fermilab if enough precision permanent magnets can be built. Permanent magnets are desirable as a means of improving reliability of antiproton storage while reducing the cost of their storage [Jackson, 9]. For this to happen though, a production line for them must be developed. To keep construction costs down use can be made of what is already available.


Refrigerator magnets are readily available. They are a ferrite based on FeO·Fe2O3 [Henry,49]. Their magnetic properties can be altered by swapping the Fe2+ ion with another metal in Group II and by adjusting the ratio of M2+ ions to the Fe3+ ions [Henry, 50-56]. A strontium ferrite1, SrO·6Fe2O3 , in a standard 4"x6"x1" brick and which has wide use in industry was chosen for this project. When magnetized these bricks will be assembled to form the 'precision permanent magnets'.


The bottle will occupy the space just above the main injector [Foster '95, Intro]. Its size and shape were chosen because it was available real estate. The main injector is in the form of a ring about 10,000 ft. in circumference2. Plans call for over 500 magnets using 20,000 bricks2.


Antiprotons will be the antimatter used in this machine. Fermilab happens to make them in convenient batches. They will arrive with 8GeV of kinetic energy. An electron volt (eV) is the energy an electron acquires when it 'falls' through a potential difference of volt. It is a small unit, only 3.82 X 10-20, calories. So 8 GeV i.e.. 8 billion eV is still only 3 X 10-13 calories, not even one banana. But, it is enough energy to get the antiprotons going at 99% of the speed of light and this means that their mass is increased by 9.5 times. Yes, relativity isn't just a theory. All this is needed to determine how strong the magnetic field will have to be in order to keep the antiprotons within the 'bottle'. The greater the mass and speed of charged particles the greater the magnetic field must be in order to maintain them in a particular sized orbit.


Because the antiprotons would be retracing their path until used, this field must be just right. A little error repeated many times in the same way will result in loss of the antiprotons as they collide with the side of the container. This need set certain criteria for field quality.


By the summer of '95 feasibility studies were underway. Several possible problems had been ruled out [Foster '95, Jackson (Ed), '95]. Corrosion isn't a problem and they don't catch on fire. Irradiating magnetized bricks does not affect the field. But the temperature is another matter. As the temperature rises the field strength goes down. Since the magnets would be located in an underground tunnel where temperatures could be expected to fluctuate from -4šF to 120šF an unacceptable change in field strength would occur.


The electric company has a similar problem with electric meters. Permanent magnets are used to construct a simple motor. As the current flows through the meter the field it generates causes the motor to run operating the dials. The electric meter is located outside where temperatures change. So, they use a high nickel stainless steel to compensate for change in temperature [Jackson (ed.), 4.11; Telcon]. Suppose the temperature goes from 80šF down to 50šF. At the higher temperature, the bricks have a higher magnetic field. But the compensator absorbs some the field so the charged particles 'see' less field. As the temperature heads downward, the magnetic field of the bricks declines also. Only now, the compensator absorbs a smaller percentage of the field. If the right amount of compesator is used, the charged particles will 'see' no change in the field. How much and exactly what kind to use would have to be worked out by testing. This was carried out at the Magnet Test Facility with samples from various manufacturers [Wufei et al].


Meanwhile a set of the same three bricks had been measured day after day. This was to determine just how stable the bricks were once magnetized. If the bricks changed in their magnetic field intensity in unpredictable ways, then the magnet they made would not work for recirculating the antiprotons. The first brick, #592, averaged 623.85 Gauss (G) with a standard deviation of 2.32 G. The second brick, #593, averaged 606.17 G with a standard deviation of 1.37 G. The third brick, #600, averaged 624.68 G with a standard deviation of 3.57 G. These measurements3were of the maximum field to which the bricks could be magnetized.


These bricks, as would all the bricks, arrived at Fermilab unmagnetized. Since Fermilab uses magnets extensively, finding one suitable for magnetizing the bricks was not a problem. A small electromagnet dipole was brought in for the job. By small I mean something about the size of my minivan minus the engine compartment.

Coils of wire inside the magnet are arranged to leave a gap right through the middle from front to back. When current flows through the coils, a compass needle inside will point straight up and down. The direction of the magnetic field depends on the direction of the current. The strength of the magnetic field depends on the amount of the current up to a point. Eventually a point is reached where cranking up the current no longer increases the strength of the magnetic field in the gap. This is called saturation. The magnetizer saturates at about 23kG. This is well beyond saturation for the bricks.


To make the bricks into permanent magnets, they are placed in the gap of the electromagnetic dipole (refered to as the magnetizer). A field is applied. Because of the material of the bricks, they become magnetized and stay that way even when they are removed from the magnetizer. Their saturation point, in the neighborhood of 600G, is ensured by running the magnetizer to 10kG. This figure2, half the original estimate, was determined experimentally.


Of particular interest was what is known as the knee of the hysteresis curve. This is where the rate of increase brick's magnetic field as the field of the magnetizer increases suddenly changes. The next change in rate occurs when the brick becomes saturated. Away from the knee the curve is mainly linear. Since the bricks are magnetized well past the knee, those with too great a field can be fine tuned downward. This is done by reversing the direction of the field in the magnetizer and applying just the right amount of field. As long as this region is away from the knee the reverse current needed in the magnetizer can be calculated.


Unfortunately the curves were a little different for each brick. One batch of bricks magnetized nearly all the same. Another batch of bricks had large differences in how the bricks were magnetized. It is too early to tell if this will make the project impractical


It was time to build some prototypes to see if the bricks would make suitable magnets. The first ones were only about 3 feet in overall length. Still, it was not easy. Left alone the magnetized bricks would simply stack themselves into one big brick.



To do the job of guiding the antiprotons, the bricks had to be arranged against their 'will'. One arrangement of bricks is called a dipole. It bends charged particles passing through the field they create.



The frame that holds the bricks is called the flux return yoke. It has three jobs: hold the bricks, provide a path for the flux (magnetic field) to circulate, and smooth out the field of the individual bricks. The other important part of the magnet is the pole tip assembly. By shaping this, the magnetic field in the gap can be adjusted.



With various arrangements of bricks and gaps the magnetic field can be molded to guide the particles on the desired path. As they all have the same charge they tend to repel each other and spread apart. So, some magnets are designed to focus the particles.

An early prototype was a the double double dipole (PDD002-0)2. It was constructed of half sized bricks so magnetizations are about half full size fields. Bricks used for this magnet came from a batch that had a median of 325 G, average of 323.7 G, and standard deviation of 5 G. These numbers3 do not seem bad. When the median and the average are close together the histogram is fairly symmetrical. The standard deviation looks small. The magnet was constructed by taking random bricks from this batch. When the magnet was measured, however, the field in the gap was peculiar. It was higher at the ends and one end had a bit of a hook in it. Definitely not suitable for batches of antiprotons to repeatedly circulate through it.


Another prototype was of a gradient magnet. This magnet has beveled pole tips so that the gap is wider on one side that the other. This time the bricks, full size, were selected so that bricks which were a little low were next to bricks that were a little high2. In this batch3 the bricks on the top had a median field of 618.1 G, an average of 614.3 G, a standard deviation of 17 G. The bricks on the bottom had a median field of 617.9 G, and average of 614.5 G, a standard deviation of 15.1 G. When measured the field was smooth as hoped by alternating weak and strong magnets. But, there were higher orders of magnetic moments. I am not sure what that means exactly except that it's bad.


It was beginning to look like all the bricks in any magnet would have to be all the same field within a small standard deviation. Some optimism was generated when one batch3 of bricks arrived that magnetized to a median field of 636.1 G, an average of 636.9 G, with a standard deviation of 4.1 G. Alas the next batch3 was once again spread out. The production line would have to include a means of measuring and tracking each brick.


So far the bricks for the magnets were being magnetized and measured by hand. I tried a few and that was enough to convince me that some automation was needed. I had to:

1. check each 4.5 lb. brick for cracks, chips, dings and paint bottom blue

2. complete the start up procedure for various power supplies --get the key, go to the building next door, unlock the power supply for the big magnet, turn it on then check interlocks for the magnetizer

3. load the brick into an aluminum holder -the holder can take up to two bricks, however only one brick at a time can be measured


4. shove the holder into the magnetizer --many problems here - there is a dusty film from the bricks, a kind of grit that gets everywhere - the mechanism slides into the magnetizer, but the slide consists of metal on metal so there is a lot of friction - it has to reach a switch in the back of the magnetizer so that it can be turned on - there is no indicator light or any other thing to let the operator know if the brick is in position other than the failure of the magnetizer to ramp up

5. complete the interlock routine --then the operator goes to a workstation (Sun in this case) and enters the serial number of the brick - the computer asks a lot of questions (but it can't tell you if the brick is in position)

6. wait while the magnetizer is ramped up --finally (provided the brick was pushed all the way back) - the brick should now be magnetized to around 600 Gauss

7. pull the holder out --this is harder than it sounds because now that the brick is magnetized it does not want to come out of the magnet - it wants to stay with all that lovely steel.

8. measure the brick's field --remove the brick from the holder and put into the measurement device - the measurement device consists of loops of wire in a capacitive resistive network - as the brick slides beneath the coil, current is induced in the coil due to the moving magnetic field of the brick - the current then charges up the capacitor, the greater the field, the greater the current and the greater the charge on the capacitor - this capacitance is proportional to the voltage across it and this has been calibrated with known magnetic sources

9. record the serial number of the brick and its field -- along with this comes a decision as to whether it is good enough

10. Whew! and they need 20,000 of these?


A system was designed by Eric Haggard to relieve the drudgery of the process. In the design, a timing belt pulls a brick in a container into the magnetizer then through the measuring coils and back to the operator. The serial number of the brick can be scanned into a data base along with its measured field.


The timing belt is turned with a motor. Recalling that some amount of tugging was needed to pull the magnetized brick out of the magnetizer, a dc shunt motor was selected. With this type of motor, the speed depends on the voltage across the armature or the voltage across the shunt field. Its speed remains constant despite an increase or decrease in the load [Rockis & Mazur,458]. And the motor's speed and direction are controlled with a motor controller.


A programmable logic controller (PLC) directs the motor controller. The PLC is the device that gives the system its flexibility. It is made to function in factory conditions and to interface with the devices found there like limit switches, temperature sensors, or motor controllers [Rockis & Mazur, 383]. Best of all it is designed to be easy to program. So if the magnetizer needs to be ramped up to a different current only a number in the code needs to be changed.

A GE Fanuc was chosen for the project. It uses the traditional ladder logic. Inputs are on the left and outputs are on the right. The big vertical bars represent control voltage just as would be used to wire up buttons and contactors. Many companies use this type of programming with a few modifications unique to a particular brand.


This gives an idea of the programming4. It is convenient to change. And, it can be used to monitor the PLC while it is running. As the brick chugs on its way, the various devices being used will light up in the program. This is useful when troubleshooting.


The carriage to carry the brick is under construction. To make it work smoothly will take a few adjustments. This is in the capable hands of Lee Benson5. Once the automated magnetizing system is completed it will be taken to MP-9 where the prototype magnets are currently being assembled. It will run alongside the manual magnetizer where final adjustments in its programming will be made.


MP-9 is a converted experimental hall so no new construction was needed to house the permanent magnet project. It has an overhead crane - handy for maneuvering large pieces of the magnets around. Full scale magnets are up to 12 feet long, and weigh over 2100 pounds. There are two assembly stations for them. Each consists of three long non-magnetic tables. Along the sides of the tables are wooden slots to hold the magnetized bricks. They are spaced far enough apart to keep the bricks from jumping out. The center table holds the top or bottom of the magnet while the two outer tables have the side pieces assembled on them. Fitted with lifting fixtures the crane is used to lift the side pieces to the center table where they are bolted to the bottom. Then the assembly can be turned over to get the top put on. The order of assembly is not quite worked out. Some fixtures may have to be devised to ensure a gentle assembly.


As mentioned earlier, magnetized bricks want to stack themselves into a big brick. This makes them difficult to work with and around. The tools have to be non-magnetic. People working here have to pay attention at all times. For example: one day Jeff Wittenkeller was about to place a brick, but he got a little too close to a container of fractured pieces which leaped up to join the brick he was holding. Unfortunately Jeff's hand was in the way. I tried to place some bricks on the flux return yoke to see how it felt. The forces due to already placed bricks changed quickly depending on exactly where my brick was. As I approach its final location, it tugged forward to jump on the next brick, but then, lowering it slightly, it suddenly was pushing against my hands. Then, moving just a touch lower it tugged strongly and, warned by Jeff, I got my fingers out in time. One of the things that made this particular part of the job hard was getting the brick between two rails. These rails are bolted to the flux return yoke and keep the bricks in place once the magnet is fully assembled. This is to ensure low maintenance down the road. But, it turned out that the rails could be put in after the bricks were laid. Something new is learned everyday.


Another reason for putting in the rails after the bricks was to lower breakage. Since these bricks are ceramic, a sudden shock can crack them. During one assembly, the side bricks caught on the edge of the rail as they were swung into place causing damage. So this change of assembly order helped in two ways.


The production line is not quite there yet. It is close. The amount of deviation in individual brick magnetization is being narrowed. The hardware for turning raw materials into completed and tested permanent magnets is being organized in one place. There is flexibility in the system to allow for changes as they come along. But most of all, none of it would have been possible without a lot of hard work by a vast assortment of people from physicists to engineers to technicians. Every time I have visited, and it was a fair number of times, everyone was working together. All in their blue jeans and all working to solve problems one at a time.





End notes

s(,1)Bill Foster (AKA G. Foster). Personal interview. 16 April, 1996.

s(2)Jim Volk. Personal interview. 1995-1996.

s(3)Jim Volk. Personal logs. 1995-1996.

s(4) John Eric Haggard. Engineering Notes. Fall '95 and Spring '96.

s(5) Lee Benson. Personal interview. 14 April, 1996.




Brown, B. (1996, April 2). Closed orbit effects due to longitudinal bend center displacement. Fermi National Accelerator Laboratory. MI-0162 version 1.1. Batavia, IL.

Foster, G. (1995, July 24). Recycler and 8 GeV line permanent magnets reference design & performance requirements. Draft, Fermi National Accelerator Laboratory. Batavia, IL.

GE Fanuc Automation. (1992). Series 90-30/90-20 Programmable Controllers. Charlottesville VA: GE Fanuc Automation North America.

Giancoli, D. (1989). Physics for Scientists and Engineers. Englewood Cliffs, NJ: Prentice Hall.

Henry, C. (1969). Electronic Ceramics. Garden City, NY: Doubleday.

Jackson, G. (1995, July 24-26). Overview of the recycler ring project. Workshop on Permanent Magnets and Lattice Design for the Recycler Ring. Fermi National Accelerator Laboratory. Batavia, IL.

Jackson, G. ed. (1995, July 18). Recycler ring conceptual design study. TM-1936. Fermi National Accelerator Laboratory. Batavia, IL.

Rockis, G., & Mazur, G. (1992). Electrical Motor Controls. Homewood, IL: American Technical Publishers.

Telcon Limited (1995). Data sheet no. 36. West Sussex, England.

Wufei, C., Draeger, C., Orris, D., & Brown, B. (1995, August 25). B-H curves of some magnetic material for temperature compensation. Fermi National Accelerator Laboratory. Batavia, IL.



Note: There have been some updates since this report was written. As of July 30, 1996, 19 of the Double Dipole magnets have been constructed, and are now in various stages of testing. These Double Dipole magnets will be used in the transfer line between the Main Injector Ring (source of the anti-protons), and the Recycler Ring.

The automated system for brick magnetization and testing is nearly complete. There have been some delays due to technical problems which could not be foreseen, but it should be on-line by the first week in August.

The problem of being unable to consistently magnetize the bricks has been solved. It was determined that the time that the ceramics were baked, lead to a variable amount of oxygen in the material. This means that different batches of bricks would have different maximum points of magnetization. Hitachi (the brick supplier) has worked with us, and has standardized the oxygen content in the bricks. Almost all of our bricks now meet the target range that we have set for their magnetic flux.

Return to main page.

Thomas Egan of Marist High School, Chicago, IL. This project was constructed as part of the Teachers Research Associate (TRAC) Program from the Fermi National Accelerator Laboratory in Batavia, IL. This project is also in conjunction with Aurora University, Aurora, IL.

Produced on: August 7, 1996