Tony, a reader of my Blog posts, has informed me he has an Autonetics G5 gyroscope. (Tony, I hope the gyroscope ends up in a museum and not in the dust bin.) This model of gyroscope was used on the “Hound dog” cruise missile INS during the “cold” war. This missile was carried on an underwing mount by the B-52 bomber and it was nuclear capable. The G5 was the last of a series of NAA gyroscopes which used ball bearings for the rotor bearings. However, the G5 owes the success to the employment of what is known as the “NAVAN” cycle to operate the gyroscope. The “NAVAN” cycle was clever in that it enabled the navigation system to meet the performance requirements AND created a market for twice as many gyroscopes as is normally required on that type INS. Before I explain this, I want to tell you what my role was in the development of the G5 gyroscope.
I became an NAA employee in 1955 and my work was that of a research machinist in the shop that was devoted entirely to the support of the NAA Aerospace Laboratory. Part of this group later became Autonetics. I was fortunate to have worked on many of the early projects. Several years later, I became a test technician in the Instrument Test Laboratory. In 1961, I obtained a Bsc. degree in physics and became a member of the engineering staff of NAA. My work continued with gyroscopes and INS until my retirement in 1990. As a consequence of this employment history, my memories of the G5 gyroscope are those of a machinist and laboratory technician. For example, One of my projects was to machine all the parts of a G5 for a cutaway display of the gyroscope. This project afforded me the opportunity to become familiar with the arrangement of the various parts of the gyroscope.
The classical gyroscope is used in the INS because of its inherent ability to remain in a fixed angular position in “inertial space” unless it is acted upon by an externally applied torque. Torque comes in two varieties – those you can control and those you can’t. (the latter are known as error torques). It is the ultimate goal of the gyroscope designer to reduce the sum of the error torques to zero. (This may sound like an impossible goal, but it was effectively achieved in the SINS gyroscope for the Navy. The bearing drag is so low, it takes over a year to coast to a stop.) In an INS, it takes three, single degree of freedom, gyroscopes to define the orthogonal XYZ space within which navigation takes place. The G5 is a single degree of freedom gyroscope. This begs the question: why did the N5 system require six G5 gyroscopes?
The classical gyroscope is basically a wheel spinning on an axle. In the development of inertial grade gyroscopes, it became clear that “plain bearings” (sometimes referred to as “friction bearings”) would never meet the requirements of the inertial grade gyroscope due to high friction levels. High friction requires the motor used to spin the rotor to apply correspondingly high torques. High motor torque results in high error torque – not good. The only practical alternative in that era was the “ball bearing”, but it was well known by then that ball bearings would never support the high level of Ins performance the visionaries of the time wanted and the technology required for radically different types of low friction bearings proved to be in the distant future. This meant “ball” type bearings would be the only choice in the near term. So, gyroscope designers sought novel ways to get the highest level of performance they could out of ball bearing gyroscope designs. One of the ways was the use of the “NAVAN” cycle.
The NAVAN cycle was developed from this concept: A+(-A)=A(1-1)=0. Simple! Yes, but difficult to achieve in practice. (The G5 gyroscope rotor used ball bearings for rotation and ball bearings allowed the rotor to be spun up in either direction. The ability of the G5 to be spun up bi-directionally is what makes the use of the NAVAN cycle possible.) What does “A”, above, represent? Assume the torque necessary to maintain a constant rotor speed is T=A+A(t).
A= constant torque
A(t)= small variations in torque over time
Due to imperfections in the the gyroscope assembly process, a small part of the rotor spin torque (k[A+A(t)]) will be directed along a line in the plane normal to the rotor spin axis. This small (k<<1) torque, being normal to the spin axis, has no effect on the rotor speed. This torque, however, will cause the gyroscope to precess (rotate) with respect to the inertial reference. This small torque is a source of system error and must be somehow reduced to a level that will still result in system navigation performance that is within the INS specification requirement. It must be remembered that during the era that the N5 system was developed, computers were very slow and were capable of only the most rudimentary computations. The computations to perform the calibration of a compensation scheme for gyroscope error torques (bias) were beyond the state of computer technology of that era. The designers of the N5 system were able to devise a method of gyroscope operation that automatically compensated for what we will call “gyro bias” by exploiting the ability of the G5 gyro rotor to spin in either direction and that the gyroscope precession, or bias drift rates, differed only in that one was the negative of the other. The system was designed such that it had two sets of the customary three gyroscopes, ie, six gyros. The INS stable platform was controlled alternatively in time by each set of gyroscopes. While one set of gyroscopes was in control of the stable platform, the other set reversed its rotor spin directions. The time allowed for of the periods when the gyros were controlling the stable platform was identical so that the error angles accumulated were equal, assuming there were no changes in the bias rate. Thus, over a four part cycle of the rotor spin reversals, the stable platform error angle should return to its starting value. During the cycle, the maximum +/- error angle would be equal to the product of the maximum bias rate and one forth the total cycle time. This process can be considered one in which the stable platform error angle at the end of each cycle can be equated to the product of the time averaged drift rate (averaged over one NAVAN cycle) and the time required for one NAVAN cycle. Since I first became aware of the NAVAN cycle, I have considered it to be one of the more clever things accomplished by the navigational wizards at Autonetics.