Grinding of small ferrite bars (1954)
In the early 1950s, the Physical Laboratory tried to perform measurements on the Faraday rotation of microwaves with a wavelength of 1.25 cm (24 GHz). With Faraday rotation, the polarisation plane of a microwave is rotated with a magnetic field that affects the intermediate medium through which the wave propagates. A special property of Faraday rotation is that the rotation is independent of the direction of the electromagnetic wave (e.g. light) that passes through the medium in case the magnetic field retains the same direction. In that case, the resulting rotation angle only depends on the length of the intermediate medium and the strength of the magnetic field.
The linear polarization of a microwave occurs through the dimensioning of the waveguide. To be able to polarise the electromagnetic waves circularly, a waveguide with a circular cross-section must be used. Small ferrite bars are used as the intermediate medium. For Faraday rotation measurements, small ferrite bars were needed with very accurate ground dimensions. How such small ferrite bars were manufactured is described below.
The measurements between two waveguides
The ferrite bar is located inside the waveguide. A coil is wrapped around the waveguide to magnetise the ferrite bar in the longitudinal direction. Between two waveguides with rectangular and circular cross-sections, a gradual change of shape takes place. The ferrite bar is placed in that transition section of the waveguide. The bar is sandwiched with a ring of foam plastic and is pointed at the ends to reduce reflections. The diameter of the ferrite bars varies from 5 to 2 mm with a total length of 25 to 10 mm. The ferrite bars are chemical compounds of Fe, Ni, Zn, and O2. They are marketed and sold under the name “Ferroxcube” as sintered bars.
The grinding of small ferrite bars
The grinding of the small ferrite bars is mainly a manual process. After a piece of material has been sawn with a diamond saw, it is pre-grounded on a grinding dish with water and carborundum powder in an octagonal manner.
The small ferrite bar is then cemented with one end into a brass tube, which in turn is mounted in a horizontal grinding spindle. The grinding is now performed by pressing two aluminium strips against the ferrite bar while adding water and carborundum powder. During grinding, the two strips are simultaneously moved back and forth horizontally.
However, the following inaccuracies occur easily with this method of grinding:
- diameter variations, which can be reduced to a minimum by taking many intermediate measurements, and
- the ferrite bars are always slightly curved.
The latter could be prevented by taking grinding strips that are as wide as the length of the bar. However, with that method, it is hardly possible to keep the bar intact. A slight curvature of the bar, however, gives no difficulty in mounting it in the small foam plastic ring.
The final operation is grinding the two conical points on a lathe using a support grinder. Clamping in a collet of a curved bar is impossible as it will break. Even when the ferrite bar is straight, so that it can be clamped, difficulties still arise during grinding, since the bar breaks off due to the pressure of the grindstone. To solve these difficulties, the bar is flexibly clamped by sliding a piece of rubber hose around it and clamping the whole fixture in a collet of the correct size.
Grinding was performed with a carborundum stone. With ferrite bars that were sintered harder, grinding with a carborundum stone was no longer possible due to wear on the stone. This difficulty was overcome by using a small diamond grinding disc.
However, the Faraday rotation measurements using the produced ferrite bars were found not to be sufficiently accurate.
For that reason, Faraday rotation had to be determined by using another, more precise method. The ferrite bars are therefore placed in a so-called vibratory cavity or cavity. A piston is used to change the length of the cavity. Certain resonances occur. The ferrite bar must be placed accurately in the centre of the cylindrical cavity in such a way that the piston can slide around it, while the bar remains fixed. The small piston can be adjusted with an accuracy set of 0.001 mm, which is achieved with a 0.002 mm/division micrometre.
This brings up the problem of having to manufacture small ferrite bars that meet the following requirements:
- The diameter must be the same for each bar over its entire length (deviations of 0.01 mm are permitted).
- The bars must be straight.
- The end faces must be perpendicular to the axis.
These requirements must be set because the accuracy of the measurement is partly dependent on the small ferrite bar being applied as precisely as possible centrically: a bent bar can never be placed centrically, while a thin bent bar runs the risk of being broken by a moving piston. No right angles to the end faces may cause the bar to be pressed away to one side causing asymmetry.
The dimensions of the ferrite bars to be manufactured are 35 mm long with diameters of 1, 1.5, 2, 2.5 and 3 mm.
To manufacture these small ferrite bars, they must first be sawn off and pre-ground, as previously described. However, the length is now relatively large and the diameters of 1, 1.5 and 2 mm have become so small that it is practically no longer possible to do the rough round pre-grinding. This is because the small ferrite bar breaks with the slightest movement at the place where the bar is cemented. To prevent this, a coil spring is soldered in the brass tube. The bar to be ground is cemented in the protruding part of that spring so that it is flexibly attached. In this way, it is possible to pre-grind ferrite bars to a thickness of around 1.3 mm with a length of around 40 mm.
As has already been stated, it is extremely difficult to manufacture ferrite bars in this way which have a constant diameter along the full length and are straight. That is why a different method of grinding has been developed. The ferrite bars are first pre-ground according to the first-mentioned methods up to 0.3 mm above the desired size, followed by a different method of grinding.
The basic idea for this method has been to rotate the ferrite bar while it is trapped between three grinding faces, one of which is adjustable. When warped, the ferrite bar will automatically be ground in a straight line. When the three grinding surfaces form a three-sided prism, the ferrite bar will obtain the same diameter throughout.
The grinding surfaces
The practical implementation of this idea can be seen in the schematic figure below.
Two of the grinding surfaces are formed by the two surfaces of a V-shaped groove, which is milled in a brass block (a). The third surface, which is also adjustable relative to the other two, is a narrow strip of a second little brass block (b). Each block is mounted in a larger aluminium component.
The three aforementioned surfaces with which the ferrite bar comes into contact must be prepared in such a way that they grind the bar without wearing off itself. To achieve this, an abrasive must be affixed. Diamond swarf 120 mesh is used as the abrasive. As a means to secure the abrasive, Araldite is used, a thermo-curable synthetic resin. In the application of the abrasive, it has to be ensured that each surface has as many as possible granules in one plane, which is achieved as follows: the block is heated to +/- 130 C. A little Araldite resin is melted on each of the heated surfaces. The resin will be spread very irregularly on the surface. Using a prepared scraper, the surplus of resin is taken away so that a +/- 0.2 mm thick layer remains. Thereafter, the surfaces of the block are then pressed in a quantity of around 2-carat diamond powder. A part of the diamond sticks.
After the block has cooled down, the residual powder is swept away with a soft brush. The block is reheated, after which the grains are pressed into one surface as much as possible with a hardened, flat-cut piece of steel. In the V-shaped groove, a counter-profile piece of steel is pressed.
Finally, the blocks are placed in an oven to cure the resin, after which the blocks are ready for use.
Adjustment of the third plane
The adjustment of the third face is carried out using two screws, which are part of the upper block holder. A steel ball is soldered at the end of each screw. One of the faces rests in a V-shaped groove and the other one in a cone-shaped hole in the holder of the lower block. The grinding surface of the upper block can be adjusted by turning one of the screws so that the three surfaces together form a prism. This can be checked by placing a steel bar in the V-groove of the lower block.
The upper holder can pivot around the two balls and thus follow the ferrite bar when it is being ground. An advantage of this construction is also that the upper holder can easily be taken away for inspection, while it can be ensured that, after being placed back, the block is again in the correct position relative to the bottom holder.
Depending on the thickness of the ferrite bar to grind and the brittleness of the material, weights may be placed on the top container to have sufficient grinding pressure.
A third screw (E), which is set in advance, takes the pressure of the weight of the container plus the extra weights as soon as the ferrite bare is ground to the preset thickness.
Driving the ferrite bar
As a result of the grinding, the ferrite bar will become thinner and will lie deeper in the V-groove. The drive shaft driving the ferrite bar must be able to move downwards without continuing to exert additional forces on the bar because this would increase the risk of breaking.
The best solution for this was a drive with a Cardan joint. The coupling at the ferrite bar side is suspended utilizing a flexible coil spring to prevent the weight from being absorbed by the bar; this is especially the case for the thinnest bars. The weak point in the drive is the connection with the ferrite bar, especially for the thin bars. A cross-sectional drawing for this connection can be seen in the next figure.
The rod is honed 0.2 mm thinner than the desired measure on one side of the bar over a length of +/- 2 mm. This grinding is done manually using a diamond grinding disc. After this step, a brass canister with shellac is put around the thinner ground part of the bar. Thereafter, the other end of the canister is fixed to the Cardan joint (PTO shaft) using a small pointed screw.
Originally, the intent was to only rotate the bar. The quality of the ground surface, however, was less than desired. Also, grinding took a lot of time, which is to be explained that each grain cuts a groove in the surface, whereby the resulting surface becomes somewhat irregular. Therefore, the resistance for each grain to penetrate the material increased. For that reason, a back-and-forth motion was added to both blocks during grinding. A stroke of 15 mm was applied using an eccentric with a drive rod that was connected to the bottom holder.
The surface quality became much better, while the grinding time was significantly shortened. The speed of the ferrite bar, originally 60 revolutions per minute, was later increased to 370 revolutions per minute. The grinding speed increased by a factor of six. Reducing the ferrite bars with diameters of 2, 2.5 and 3 mm with 0.3 mm lasted approximately 20 minutes. Grinding ferrite bar with diameters of 1 and 1.5 mm, the time was +/- 35 minutes because the grinding pressure had to be reduced otherwise the ferrite bar would break.
Before grinding a ferrite bar, about 300 grams of weight was placed on the top holder. For grinding a ferrite bar of 1 mm, the weight was 150 gr. The hardness of the material also affects the grinding time. Six different types of ferrite materials have been ground, one of which was very hard and which required a grinding time that was two to three times longer than for the relatively softer ferrite types.
To wash away the grit during grinding, a small pipe with several holes was located parallel to the blocks of the machine. Water was jetted against the ferrite bar during grinding.
An even bigger challenge
The Faraday rotation measurements showed that it was desired to use even thinner ferrite bars. In the same manner, we produced perfect circular ferrite bars of 0.7 mm in diameter. The ferrite bar was at first normally ground to 1 mm. Then the weight on the upper block was reduced to 30 gr. The horizontal stroke for both blocks was reduced to 5 mm. The weight was made lighter to make the grinding pressure smaller because the torque needed to drive the ferrite bar had to be delivered via the cemented connection. A rivet of 0.5 mm had to be ground to the ferrite bar after the diameter was reduced to 1 mm. The resistance against twisting was reduced by a factor of four. The stroke length had to be reduced. Otherwise, the ferrite bar extended too far beyond the blocks and could result in a bending load that could break the ferrite bar.
The time needed to grind such a thin ferrite bar was significantly increased due to all these limitations. Depending on the hardness of the material, the time required for grinding a bar of 0.7 mm diameter could take between 2.5 and 5 hours.
The end faces are polished at right angles to the axis in the following way: a hole of the same diameter as the ferrite bar was drilled in a brass block. This block was placed on a grinding scale with the hole in the vertical direction. By inserting the ferrite bar into the hole, it will touch the scale with one end. This top surface could be ground now at a right angle. In this way, it could be prevented that the ferrite bar could wiggle during the grinding and that edges could chip as the ferrite bar was entirely locked up in the hole.
Using relatively simple means, it has been proved possible, despite the difficult to-process sintered “Ferroxcube” ferrite bars, to manufacture 35 mm long ferrite bars with diameters of respectively 0.7, 1, 1.5, 2, 2.5 and 3 mm and a tolerance of 0.01 mm. As a pleasant side-effect, the ferrite bars, which were initially bent, were delivered automatically straight by using this method.
Text and figures are derived from an article by P. Leemans “Het slijpen van ferrietstaafjes” published in Dutch in the Polytechnisch tijdschrift Nr. 19-20 – 12 mei 1954.