SMALLER IS BETTER
I went to work for NORTH AMERICAN AVIATION in 1955 as a Research Machinist in the AEROSPACE LABS; I knew from the beginning this was my dream job. The work was very challenging to my newly acquired machinist skills and was of great interest to me as the projects within the lab were very “cutting edge” in nature. It seemed like little was known; everything had yet to be invented. I soon began to do work for the potential “inventors”. I did not know at the time that I would work with many of these people for the next 35 years or that I was destined to become one of them. One of the Engineers I met was Joe Boltinghouse. I did not know then that Joe’s work would play a large role in my future work life. For me, the tyro machinist, my work life was looking good and destined to get better.
The AEROSPACE LABS were established in the early 1950’s to use the knowledge gained from testing of the captured German WW II ballistic missiles.This knowledge became the basis for the beginning of the U.S. missile programs. My work was concentrated in the field of “guidance and control” although I did not realize this until later. I found myself making parts for things called “gyroscopes” and “accelerometers”. My lifelong involvement with gyroscopes had begun.
The gyroscopes of the day were designed to achieve performance goals by using rotors with large angular momentum to offset the large errors inherent in ball bearings. High angular momentum was achieved through the use of heavy rotors that were spinning as fast as ball bearings would allow, that is, slow. At the time, ball bearings were the only practical bearings to be had and to say they were not good enough to meet performance goals is an understatement. I did not know it at the time, but better bearings were being developed in a nearby laboratory by a group of engineers which included Joe Boltinghouse. However, our ball bearing gyros were good enough to guide two U.S. Navy atomic powered submarines under the polar ice to the North pole and back. These voyages demonstrated the feasibility of using Inertial Guidance Systems to navigate ships. SINS was born!
A short time after the polar success, I left my job as Machinist to become a Test Technician in the Instrument Test Lab. It was there I was introduced to the fruits of the work done by the bearing development group. Their new bearings were the were the basis for two new gyro designs. The first, a single degree of freedom gyro using a journal gas bearing, was to become the gyro used for guidance of all the Navy’s ballistic missile submarines. The second, a two degree of freedom gyro using a spherical gas bearing, became the guidance gyro for the entire Minuteman ballistic missile fleet. My work testing the prototype two degree of freedom gyro made me aware of the significant role Joe Boltinghouse had in the development of this gyro. The design of this gyro was based entirely on Joe’s ideas. In fact, Joe’s name is on all the patents today. The testing of these gyros began what for me became a career-long involvement with Joe’s inventions. Unknown to me at the time, Joe was already at work developing his ideas for the the next generation of gyroscope. The gyro design which resulted from Joe’s new work employed a spherical Beryillum rotor supported by an electro-static bearing. The rotor is one centimeter in diameter (about one half inch) and made from beryillium metal. The new gyro is unconventional in design and operation and it has become the gyro with the lowest performance errors ever produced in quantity.
A simplified way of explaining why gyro rotor bearing errors become a source of gyro performance errors is as follows. Gyro performance errors are proportional to the product of the gyro error torques and the reciprocal of the rotor angular momentum. Given this relationship and the design goal of producing a gyro with the lowest possible performance errors, the designer has but one option . It is to minimize gyro error torques and maximize rotor angular momentum. A complicating factor is the desire to produce a gyro of minimum size. Gyro error torques are reduced by selecting a bearing that is inherently low in drag producing torques. The electro-static bearing fits the bill. The desire for a small, very low error, gyro led to the choice of a small spherical rotor supported by a low error electro-static bearing. Electro-static bearings depend on modulation of the force of attraction that exists between the plates of a capacitor. Because of the limitations placed on the maximum amount of force producing charge that can be maintained on the capacitor plates, the forces available to support the rotor are small. This constraint requires that the rotor be light weight in order to survive in the real world. So the designer ends up choosing a one centimeter diameter ball made of beryllium and supported by an electro-static bearing.
These choices were made knowing there were some practical difficulties to be confronted. One such was the choice of the size of the gap between the rotor surface and the plates of the electro-magnetic bearing. The choice of gap size was driven by the physics of capacitor plates and the requirement to produce sufficient bearing forces for the rotor to survive the real world environment. The resulting gap size was approximately two hundred millionths of an inch. This very small gap in turn requires the ball to be spherical within approximately two millionths of an inch when the rotor is spinning at operational speed and at operating temperature.
The design process is about to become complicated. The rotor fabrication process that was developed produces a rotor which has the required roundness and diameter. Note that the rotor is not spinning and is at the process temperature when made to be round. When this “round” rotor is taken to operating speed and temperature, it takes a shape similar to a USA football and gets “bigger”, that is, it is not the desired shape or size. This “shape and size issue” could have been a show stopper. That it was not a show stopper and how that came to pass is where the phrase “Elegant Design” come to mind.
Beryllium metal, due to its properties, cannot be worked like ordinary metals. It also is very toxic to humans when it enters the body. It, however, is safe to handle under ordinary circumstances. For these reasons, Beryllium stock is made by compacting and fusing powdered Beryllium in a cylindrical die using high pressure and temperature. The result is a billet suitable for the fabrication of instrument parts. Because it is toxic, Beryllium parts must be fabricated within the confines of a very special machine shop.
One of the important parameters of materials used in the manufacturing of precision instruments is that of the coefficient of thermal expansion. The method used to manufacture Beryllium stock is hot pressing to form billets. The thermal expansion coefficients of the pressed billets are found be different when measured along the billet pressing axis and in the radial plane normal to the pressing axis. This difference is not a good thing for most applications. However this attribute played a key role in the design of the electro-magnetic bearing gyro. Joe’s genius allowed him to realize that the difference in coefficients between axes could be used to design a rotor that would be round when at operational speed and at operating temperature.
This is how he accomplished this tour-de-force. Joe’s design required the rotor to made round at a defined temperature above room temperature. The defined temperature was chosen such that the rotor, when at operating temperature, expanded deferentially to a shape that exactly cancelled the the shape produced within the rotor by spinning at operational speed. The rotor is now round! Genius! however, there is another problem to overcome. In order for the shape cancellation to occur as planned, the axis about which the rotor is spinning (the spin axis) must lie along the axis defined by the pressing of the billet used to fabricate the rotor.This requires that knowledge of the location of the pressing axis be maintained throughout the entire rotor fabrication process. The wizards in the shop accomplished this. How they did it is a story for telling another time.
The design of a practical rotor is not sufficient to produce a practical gyroscope; it is required that provisions to accurately spin-up the rotor be provided (a motor) as well as a pair of hemispherical cavities which contain the conductive plates to provide the capacitor surfaces required to complete the electo-magnetic bearing. These cavities, when joined and aligned, must form an internal spherical surface such that the designed bearing gap is achieved and the rotor can be spun-up. A successful cavity machining process was developed by the rotor fabrication wizards.
So far we have provisions for a rotor bearing which permits levitation of the rotor such that the bearing gap is uniform. Successful levitation assumes servo electronics are provided. (This is another wizard story to tell some day). The electro-static bearing is required to operate in a very hard vacuum to eliminate the threat of gas ionization due to the high voltage gradient that that results from a very small gap. Our practical gyro also requires a means by which the angle between the spin axis and reference axes fixed in the cavity set is determined on a continuous basis, i.e., a pickoff.
The requirement for a pickoff must be met by means other than mechanical, magnetic or optical devices. Given the size and precision of the rotor/bearing design, the use such devices is not practical. It is noted that the cavities must and do have electrical conductors passing through the cavity walls which provide the means by which the levitation servoes modulate the levitation charges. The question becomes: How to utilize the servo conductors for the the pick-off requirement? The answer was provided by Joe Boltinghouse through his invention of the MUM pickoff scheme.
MUM is an acronym which stands for Mass Unbalance Modulation. THE MUM pickoff is based on this characteristic of a charged capacitor: If the gap between the plates of a charged capacitor is changed, keeping the total charge on the plate constant, it will be observed that the voltage measured across the plates changes in proportion to the change of the gap. Joe figured out that if a way could be found to cause the surface of the spinning rotor to modulate the bearing gap, the voltage across the bearing gap would be modulated also. This is the basis of the MUM pickoff.
A perfectly round, homogeneous, spinning rotor (the geometric center and the mass center are coincident) will not produce a modulation of the bearing servo voltage because the bearing gap will not change as the rotor spins. No pickoff information is at hand! However, if the spinning rotor is not homogeneous (the geometric center and the mass center are not coincident), a modulation of the servo voltage will occur because the rotor spins around the mass center. The rotor surface does not run true and thus modulates the gap at rotor speed. A pickoff is thus possible.
To complete the pickoff design several issues had to be addressed. The first is how to fabricate a non-homogeneous rotor from homogeneous material. It was learned that it was possible to make long, thin rods of Beryllium by extruding the rods from hot Beryllium billets. This capability was exploited to achieve the desired non-homogeneity by making an extrusion of a multiple part billet. The major parts are an inner cylinder and a close fitting outer sleeve, when mated, would comprise the extrusion billet. Before mating the billet parts, several tiny grooves were cut into the surface of the inner cylinder and small diameter wires, of density greater than Beryllium, were laid in the grooves. The outer sleeve was then slid over the inner cylinder to complete the billet assembly. It must be noted the determination of the parameters of the wires, the diameter of the inner cylinder, and the extrusion ratio were calculated starting from the MUM pickoff requirements. Because of the toxicity constraints of Beryllium metal, the billet assembly was sealed into a steel can prior to the hot extrusion process. What emerged from the extrusion die was a beryllium rod encased within a thin steel coating. Several more important benefits were gained using the extrusion process. The axis defining the thermal coefficient of expansion is very well defined. The wires are known to be parallel to the extrusion axis and are located within the rod as the rotor design requires.This requires careful design of the billet sealing can.
There is more! The wires that produce the MUM pickoff signal also result in the rotor having unequal moments of inertia along three orthogonal rotor axes. The placement of the wires is such that the greatest moment of inertia is along the extrusion axis. The other two moments of inertia, one greater than the other, are orthogonal one to another and to the first. The rotor, when spun-up, most often ends up with the spin axis not along the axis of the greatest moment of inertia (principal moment). When this happens, an effect called Polhode Motion appears in the MUM signal. Polhode Motion corrupts the Mum pickoff signal to the extent it masks the useful pickoff information within it. The polhode motion must be reduced to zero by moving the spin axis into coincidence with the principle moment of inertia axis. This is done by modulating the voltage applied to the spin motor as determined from the polhode motion harmonic signature in the pickoff signal. When the polhode signal is reduced to zero, the motor is used to inductively heat the rotor to its operating temperature. The motor is then turned off and remains off until the rotor is spun down months later. The rotor speed is maintained at 3600 revolutions per second by a “speed trap” designed into the levitation servo electronics. The levitation servo produces the forces necessary to maintain levitation and counter the effects of loads applied to the bearing. The modulation of levitation forces by the bearing gap modulation, necessary to produce the MUM pickoff signal, produces a torque directed along the spin axis which speeds the rotor up. This torque is reduced to near zero by the “speed trap” designed into the servo electronics and the rotor speed is maintained at the required 3600 rps. Amazing!
I do not have a definition for “Design Elegance”. As the saying goes: “I know it when I see it” and I see it loom large in the designs that flowed from the inventions of Joe Boltinghouse. I was very fortunate to have the company of Joe at our weekly retiree luncheon for over twenty years. We had many discussions about the historic details of the invention and development of inertial instruments that our group of Engineers, technicians, and Machinists participated in. Joe Boltinghouse passed away over a year ago, as so many of our cohorts have, but his friends still miss him at lunch. The USA Navy and Air Force will not soon forget Joe either.