Fig 1.1-1 Pre-Silo Minuteman I Fig 1.1-2 Post Silo Minuteman III
The defense department realized that the long countdown delays, associated with rocket engines that required very cold liquid oxygen, were simply unsatisfactory. There had to be a faster response in the event of an attack. Contracts had been let to develop solid propellant motors – they looked promising and the Minuteman came into being. The concept was to use a stack of three solid propellant booster motors, put the missile in an underground silo, and require countdown to launch in one minute – thus the name Minuteman. The Minuteman was to carry a single Uranium Bomb and accurately place it on the other side of the Earth.
To play it safe, the AF decided to use an Associate Contractor arrangement. The contractors were: Thiokol stage I, Aerojet stage II, Hercules stage III, Autonetics Guidance and Control, GE Warhead system, and Boeing Silo design & Missile assembly. The AF hired TRW as their Technical Advisor. Each of the above operated under a separate AF cost plus incentive contract.
Minuteman was to prove to be a great success, very reliable and reducing the AF cost of ownership to 1/6 that of prior liquid Intercontinental Ballistic Missiles (ICBM). For the planners it was a bit risky, they didn’t even know if you could fire a missile from a hole in the ground.
The Minuteman I program scored several firsts, “breakthroughs”: 1) System wide commitment to apply semiconductors (solid state electronics). 2) Solid propellant motors that would fire on command, provide reasonably constant thrust for 60 seconds, after being stored vertically in silos for years (not sag, crack, out gas, etc) then fire reliably. 3) Silo launch technology.
The subsequent descriptions will get into the details of how things work. There is a reason for this. Many of the items did not exist during WW II but are now fundamental to our societies needs, they are used on aircraft, farm machinery, and industrial equipment – things which are electronically controlled.
The B-70 Bomber and associated personal learning was going on at the same time as the Minuteman. These are included as they interweave with the development process.
1.1 Introduction - Index
1.2 Autonetics Organization
1.3 Proposal for a thing called Minuteman
1.4 What is a Minuteman
1.5 Guidance and Control Proposal
1.6 How do you do Attitude Control
1.7 Fly Wheel Pre-proposal Tests
1.8 Actuators-Pumps-Electronics for Booster Tests
2.0 Autonetics Wins
2.1 Autonetics Wins the G&C Contract
2.2 Eisenhower’s Request
2.3 Inertial Navigation
2.4 The Silo
2.5 Tethered Launch
2.6 Reliability becomes a Religion
2.7 The Missile System
3.0 Guidance & Control System
3.1 Guidance System
3.1.1 Guidance System photos
3.2 Flight Control System
3.2.1 Flight Control System
3.2.2 Flight Control Computer
3.2.3 Nozzle Control Units
4.1 Servo Loop Elements
4.2a Actuator Body
4.2b Servo Valve
4.2c Position Transducer
4.3 Feedback Demodulation
4.4 Summing Junction and Amplifier
4.5 Servo-actuator packaging
4.6 Make or Buy
5.1 Hydraulic Power Supply
Learning and Planning Ahead
6.1 Mechanic Learns Electronics
6.2 Preparing for Supersonic Transport
6.2a ”Hot vs Cold Isolation
6.2b B-70 and Supersonic Transport Cancellation
Servo Loop Testing
7.1 “Missile Electronics” for Servo Tests
7.2 Hi Quality Square Wave
7.3 Torque Motor is an Integrator
7.4 Frequency Response Tests
7.4a Failed Frequency Response Test
7.5 Shaping Networks & Filters
Move to Fullerton
8.1 Downey to Fullerton
8.2 Analog Computer Controlled Load Cylinder
8.3 Operational Amplifiers – Chopper Stabilization
8.4 Load Simulator Completed – Not Used
8. 5 No Shaping Networks Required
Lab to Lead
9.1 Greer-Purpura to Project Landau to Lead Engr
9.2 Servo Valve Contenders
9.3 Rechecking Design Requirements
9.3a Two Sentence Note to Curci
9.4 90 Day Redesign
9.4a New Servo-Actuator
9.4b New APS
9.4c New Battery – Blown Cork
10.1 TD Meetings with LTV
10.2 TD Meetings with Vickers
11.1 Failure Modes Log
11.2 Failure Reports Fail
11.3 Minuteman Special Parts, IIDs
Move to Anaheim
12.2 Insulation of the NCUs
MM I Electronics
13.1 MM I Electronics state of the art
Heat Protection for MM
14.1 Heat Protection for MM Flt Ctrl Equip.
15.1 People Epilogue
Fig 1.1-3Downey Plant Site
Bldg 1 location of Navaho & Flight Control for MM proposal, bldg 9 cafeteria, bldg 123 bombshelter, location between bldg 3 & 4 was temporary location of Autonetics after MM start , later sold to city of Downey, after the move to Fullerton and Anaheim.
Bldg 6 and 250 were location of Inertial Navigation prior to move to Downey.
This site became home of Space Shuttle
1.2 Autonetics Organization
The Autonetics Minuteman organization had it’s origin in the MACE, (the medieval hand weapon) organization located in Downey, CA where the Navaho missile series was developed. Rocketdyne and Atomics International had already spun off when the Navaho program was cancelled – 4000 people were laid off in one day. Space division and Autonetics division were formed out of what remained at the Downey site. Space division began work on the Hound Dog missile and would follow on to do the Space Shuttle – the man on the moon program. Autonetics division was split off to do Guidance and Control work, first for the Hound Dog missile, Navy submarines, and then Minuteman. To make room, work was started on a new plant in Anaheim CA once the Minuteman contract was awarded.
Fig 1.2-1,2,3 John Moore heads of Autonetics Fred Eistone head of Inertial Navigation Don Williams head of Flight Control
John Moore was made head of Autonetics, Fred Eistone head of Inertial Navigation and Don Williams head of Armament & Data Systems which included Flight Control. Our Flight Control organization began Minuteman operations in Downey then temporarily located in Fullerton while the Anaheim plant was being built. Tom Shuler was head of Flight Control in Fullerton and Elliott Buxton was his assistant. Under them was Project Engineering under Ray Curci, Engineering Section Chief R. Bond and under him Systems Engineering under Frank Henderson and Components Engineering under Fred Morgenthaler. Under Morganthaller was Odel Taylor supervisor of Electronics and Paris Stafford supervisor of Hydraulics. I worked under Stafford in charge of a test lab in Downey, the previous Navaho extreme temperature lab and still being used for the B-70 Supersonic Bomber program. Eventually all of us, including the Lab were moved to the new site in Anaheim. Operations began in Downey, but Minuteman didn’t really get under way until Fullerton. We were at Fullerton for over a year, still there at the time of the first flight.
1.3 Proposal for a thing called Minuteman
My first awareness of there being a Minuteman was a conversation with Art Greer when we happened to be sitting together at the cafeteria in Downey. Art worked in the main building and I was operating the “Bombshelter” a separate extreme temperature test lab. Art had been assigned to work on the proposal for a thing he called Minuteman.
1.4 What is Minuteman
I asked what is this thing you call Minuteman. He said the AF decided they needed a missile they could launch fast, and not wait for extended count downs that can last for hours and be held up for days. They’ve come up with a way to use three solid propellant motors , put the missile in an underground silo, and countdown to launch in one minute – thus the name Minuteman. [The name Minuteman was taken from the Minuteman of the American Revolution who were ready to defend the nation on one minutes notice.]
1.5 Guidance & Control Proposal
Art said we are working on a proposal for the Guidance and Control. The AF likes the Inertial Guidance system developed for Navaho, so the Inertial Navigation department is proposing the Navaho Inertial Measuring Unit (IMU) for the Guidance. The AF has been working with solid propellant motor contractors to come up with a way of putting four tilting nozzles on the back of the motors for attitude control, so we in Flight Control are proposing our Navaho developed hydraulic servo actuators for the Nozzle Control Units (NCU).
1.6 How do you do Attitude Control
I asked how do you do pitch, yaw & roll? Art said the four nozzles at the back of the booster motor tilt. You tilt one pair for pitch the other pair for yaw and use a differential pair of pitch +- yaw to do roll. And how do you power the hydraulic pumps for a solid propellant motor, there are no rotating shafts. That’s the part Lou Purpura and I are working on. Lou’s looking into pumps we might use and I’m looking into what we could use to power them. One way is to use a battery to run a motor but those could be quite heavy – do you have any ideas on what we might do?
I said, there were number of methods used to start aircraft engines. One was to use “shot gun shells” and spin the starter mechanism with gas. But most used an electric motor to spin up a fly wheel, which when up to speed, was engaged to rotate the engine. These were heavy duty on the B-29 where the flywheel was the armature of the motor and even after engaged the motor would continue to help turn the fly wheel. The shot gun shell method was soon dropped and the B-29 used what had proven to be the best.
1.7 Flywheel Pre-proposal Tests
Art was fascinated by the idea of using a flywheel and we talked about it’s possibilities. I said, you could rotate stage II before launch and let it spin, relative to stage I, to power the pump after launch. We agreed that was not practicable. I was not surprised when Art came out to the Bombshelter with an experimental flywheel to test the concept. We made many test runs and plotted the data. Ron Frazinni had just come to work, right out of school, and took on the job of doing data reduction. Ron establishing a math model and used the test data to show what such a mechanism would do. Ron’s calculations revealed that the task to be done was more demanding than the flywheel could handle. Art & Lou were reluctant to give up on the idea but the numbers made it more and more obvious that it wouldn’t work. The more they learned of the demand, the more they proved the flywheel could not do the job.
1.8 Actuators and Portable Pumps for Motor Contractor Test Firings
Project engineer Ray Curci found the solid propellant motor contractors needed servo actuators for their test firings. Ray knew that if we could get there first with the equipment needed, we could get our foot in the door for supplying the control servos – before someone else did. Working with Paris Stafford our Engineering Supervisor, they had Lenni Nauman, lead design engineer, come up with three servo-actuator designs. These were essentially the same design but sized to fit stage 1, 2, & 3 test motors. Lenni had designed the actuators Rocketdyne used to gimbal large engines they were developing for the Atlas and Titan missiles. They also started putting together a portable hydraulic pump packages. This was a suitcase with an electric motor driven hydraulic pump. I don’t recall who came up with the motor-pump arrangement but it was quite clever, adapting a carpenters portable saw motor to run the pump. This was done in a remarkably short time, fellows came in week ends on their own to complete the packages. The motor contractor could hook them to the tilting nozzles and command movement as required to prove their designs. This is where the Navaho experience really paid off. [Following the heavy layoffs when they folded the Navaho program, none of us turned in overtime so as to stretch what funding there was to the most people.]
2.0 Autonetics Wins
2.1 Autonetics Wins the Guidance and Control Contract
The AF awarded Autonetics the Guidance and Control (G&C) contract, primarily to make use of their Guidance System, developed for the Navaho missile and then being placed on submarines. Thanks to the Curci initiative to provide controls support to the motor contractors, making use of equipment developed for the Navaho, the AF combined Guidance and Controls into one G&C contract.
2.2 Eisenhower’s Request -- Place Orders -- Now
There had been a huge reduction in the aerospace industry when the Navaho and similar programs were cancelled. The abrupt layoff of 4000 people in one day from the Navaho program had a ripple down effect on subcontractors as well. Many very capable people were out of work seeking jobs. The Eisenhower administration wanted to use the Minuteman program to stimulate employment. I was in the test labs and told to place orders for some $100,000 dollars worth of equipment -- now. The idea being that if orders were placed, business would retain people. I was given about two days to place the orders. I grabbed Bob Kelley & Bob Parkinson to help go through equipment catalogs. We didn’t know what we would need but figured we couldn’t go wrong buying good general purpose test equipment. We placed orders for some 20 HP Oscilloscopes. We were hard pressed to spend and not waste. As it turned out, almost all the equipment was put to use and the money was not wasted.
I was told to buy some test consoles. Ted Lenney had arranged for Space Division to make test consoles for Flight Control patterned after their factory check out consoles. Ted was upset when I said no, we’d design and build our own using standard electronic racks. The control rooms in our Bombshelter test facility were to small to accommodate the large consoles. We bought standard electronics racks plus parts for a mobile base and the technicians assembled them. They worked out very well and were used for years – being moved to new facilities in Fullerton and later in Anaheim.
2.3 Inertial Navigation
Though this chapter is primarily my recollections about Flight Control, it is appropriate to tell about Inertial Navigation and the Guidance System – the primary reason Autonetics won the contract.
The AF had been highly impressed with the work done by North American Aviation on the Inertial Measurement Unit (IMU) for the Navaho missile. This included a set of three accelerometers, designed by MIT Draper Labs, mounted on a gyro stabilized platform oriented to measure x, y, z motion. Think of an accelerometer as if it is a plumb bob hanging from the rear view mirror of your car. If you accelerate it’s inertia will cause it to swing back, if you decelerated it’s inertia will cause it to swing forward, the amount depending on the rate of change – which is acceleration. Three rate (accelerometer) instruments are placed on a platform. The platform is gyro stabilized, to hold a constant position and is placed inside a set of gimbals. The inner gimbal ring holds the platform and the outer ring gimbal is attached to the missile structure. Thus the stable platform remains in a constant position in inertial space while the missile moves about it. By keeping track of acceleration in x, y, z coordinates, from the time of launch, the location of the missile could be computed. The Guidance computer compares a planned mission trajectory with where the missile actually is. The stabilized platform of accelerometers is known as the Inertial Measurement Unit, the associate computer as the Guidance Computer and the combination as the Guidance System.
Fig 2.3-1 IMU used for submarines Fig 2.3-2 Route of the Nautalis submarine under polar ice using Navaho IMU
Fig 2.3-3 When Zane Sandusky made a quick trip to install a Navaho IMU in a submarine for the first test, stenciled on the IMU was:
“Do not exceed Mach 3, Do not exceed 14 G’s” – the sailors got a kick out of that.
The Germans developed a primitive inertial navigation systems for their V2 rockets, and Sperry developed such for Navy submarines; but it was the Navaho IMU that achieved the precision necessary for our submarines to go under the Arctic ice to the North Pole. [Zane Sandusky had an office next to mine and we worked closely together on the B-1B program. During this time Zane told of being rushed from west to east coast to install a Navaho IMU in a submarine. The Sperry system needed repair and the Navy had designated the Navaho IMU as the backup. They were told they had two weeks if they were interested in installing their IMU. Zane told of taking the unit by plane, renting a station wagon and showing up before being expected. He told of having to take the unit out of it’s case and grind off rivets to “push” it down the conning tower. That he developed means of calibration after it was installed. The test run was very successful, accurately defining where they were, there after the Navy never went back to the Sperry system. Within a few months they sent the Nautilus Submarine under the North Pole, Zane went on their second trip – surfaced and had a “walk about the pole”.]
Sled test proved this Navaho IMU would hold up under severe G (gravitational) loads. The Minuteman can experience up to 14 G’s and thus requires a special IMU. Moon shot G forces were kept low because of the people on board. [Don O’Neil, who Sandusky and I worked with on the B-1B, had been an engineer on these tests and told us of the AF being impressed with how well the IMU performed under severe G forces.]
2.4 Silo Based
Fig 2.4-1 Silo casing Fig 2.4-2 Looking into a missile Silo
The silo was to be a “hole in the ground” which was to protect the missile from all but a direct hit and be available in a minute to fire out of the hole – if ordered by the President.
2.5 Tethered Launch Tests
Very early on Tethered Launch tests were performed at Edwards Air Base. These were a partial stage I with it’s controls and a dummy forward part. Some 8 of us went to see the first launch. Dalton Davis was running our equipment from a launch control and we could hear the “countdown process” by the pitch of the high speed, 20,000 rpm, motor driving the hydraulic pump. There was a billowing puff of black smoke – our eyes searching to see the missile – then there it was already out being pulled to the ground by it’s long tether. The stage I motor was loaded with enough propellant to provide 2.8 seconds of burn time. [In 1996 I was to find that M/Gen John Carpenter II, was in charge of the tethered launch at Edwards. John was a member of our 19th Bomb Group association, and had been at Clark Field Philippines when WWII started, his plane lost two days later, he was taken out by submarine to Java to fly again.]
2.6 Reliability – Becomes a Religion – Levied by Contract
From the very beginning reliability received tremendous emphasis. I was consulted about tests we had run in the lab, failures and their causes. We would discuss what seals and joints were the most vulnerable to failure. These discussions had a considerable influence on design.
A special team of people were set up under Bill Yetter, called design assurance to handle this aspect. As the program progressed this effort became more sophisticated.
2.7 The Missile System
The following figure is of Minuteman I in flight, with close up view identifies section.
Figure 2.7-1 Minuteman I
3.0 Guidance and Control System
3.1 Guidance System
3.1.1 Guidance System Photos
Fig 3.1-1 Inertial Measurement Unit, in G&C section (MM3 version)
Fig 3.1-2 Inertia Measurement Analog Electronics in G&C section (MM 3 version)
3.2 Flight Control
3.2.1 Flight Control System
The Flight Control System which consisted of: a Flight Control computer (P-92), (3) Nozzle Control Units (P-89, P-90, P-91), and (5) Cable sets for connecting: G&C to stage III, stage II, stage I plus G&C and stage I to silo power, checkout and launch control. The following schematic shows the set up for a single downstage NCU, where this repeats for each of the NCU’s.
Fig 3.2.1-1 Location of major elements
Fig 3.2.1-2 Flight Control Block Diagram
3.2.2 Flight Control – Missile Digital Computer
The Missile Computer (D-37) was packaged as part of the G&C section of the missile, forward of stage III and aft of the Warhead. It was built by Data Systems Division, of which Flight Control was a part. According to Elliott Buxton this was the first Digital Computer used for Avionics controls. It was an all transistor design.
Fig 3.2.2-1 The Minuteman D-37 Computer, in G&C section (MM 3 version)
Fig 3.2.2-2 P-92 Flight Control Electronics, in G&C section (MM3 version)
Fig 3.2.2-3 Angular Accelerometer unit
Fig 3.2.2-4 Angular Accelerometer
Fig 3.2.2-5 Angular Accelerometer system schematic
Fig 3.2.2-6 Angular Accelerometer Electronics MM I vintage
Fig 3.2.2-7 Minuteman I Flight Control Schematic
3.3 NCU – (3) Nozzle Control Units
The Nozzle Control Units fitted between the tilting nozzles at the back end of each of the solid propellant booster motors. The NCU structure attached to the motors via four identical Servo Actuators, one for each of the four tilting nozzles. Fig 4.2.1-2 left, shows two bolt hole attachment locations for actuator to nozzles and fig 4.2.1-2 right shows the flat plate attachment to the NCU structure. The structure, shown in fig 3.3-1, was called “the dog bone”. The structure supported the Auxiliary Power System (APS), a motor driven hydraulic pump system, that powered the servo-actuators. The structure also held the Analog Electronics that command the four servo-loops. The NCUs were connected by shielded cable  , fig 3.4-1, to the D-37 Computer at the top of the missile.
Fig 3.3-1 Stage I NCU
Fig 3.3-2 Stage I APS and Battery in NCU
Fig 3.3-3 Stage III NCU
3.4 Missile Cables -- Battery Power -- Raceway
Cables: Missile cables carried control signals from and to the G&C section as well as electrical power to operate the hydraulic pumps during check out. The cables were heavy and shielded with stainless steel braid. Most of the weight was in the battery power cables.
Batteries: There was a battery at the top in the G&C section and also used to power stage III, a battery in the middle stage II and at the bottom for stage I. There was an umbilical cable at top and bottom, from G&C and stage I, to provide silo “battery” power for checkout.
The batteries were sealed and dry until activated just before launch. A squib (an electrically activated propellant charge) was activated at the last moment in a count down and this pushed the liquid electrolyte into the battery cells and thus activated the batteries. There was no “starter switch” between batteries and the electric motors, when the battery power came up, it ran the motors – until that stage burned our, it’s job done.
Fig 3.4-1 Missile Cable Segment
Fig 3.4-2 Raceway cover
The Raceway: The interstage cables were enclosed by a six inch high raceway cover, which also covered the “missile destruct” explosive charges. The motor cases are pressure vessels, lined with an inhibitor to insulate and stop buring through a case. Wiring had to go on the outside.
The raceway protrusion was no problem so long as the stage I used four tilting nozzles, as roll control torque came almost for free. Later when making studies for an MX missile with a single nozzle it was found that side winds on these protruding raceways could cause high roll torque demands. Some of us had participated in meetings with Lockheed engineers who told of how they placed wiring flat against the outer surface, as they “squirted” the missiles from tubes in the submarines using steam pressure – with a flat cable they had no roll torque problem. At a meeting at Norton Air Base, while waiting for a meeting to start I discussed this with one of the AF officers. It soon followed that the MX missile would have “flat” cables and not need for a protruding raceway.
4.0 The Servo Loop
4.1 Servo Loop Elements
The combination of electronics and the servo-actuator with it’s; actuator body, servo valve, & position transducer is called a Servo Loop. The servo loop parts are part of the NCU.
The missiles Guidance system tells the Flight Control computer the amount of thrust vector correction required.
The D37 Computer sends a new position command to the down stage electronics.
The down stage electronics commands a new position. This causes the servo valve to port hydraulic fluid to make the actuator move. The actuator position signal is called negative feedback as it subtracts from the position command. When the new position is achieved, the position feedback cancels the command and the “error” signal to the servo valve is reduced to zero -- the servo-actuator is “held” in it’s new position until it receives a new command. The difference between the command and feedback is called the “error” signal and when applied via the servo valve becomes a rate command. If the error is large flow to the actuator causes it to move fast, as the error decreased the actuator slows to a stop and does not overshoot. This can become a sophisticated process, discussed later under Shaping Networks.
A hydraulic servo-actuator has three parts: actuator body, servo-valve, & position transducer; and is electronically controlled. It’s called a “servo” because it goes to a commanded position.
Fig 4.2-1 Servo Actuator Parts
4.2.1 Actuator Body
Most people have watched a hydraulic actuator at work on construction equipment, where hydraulic pressure moves a shaft in or out of a cylinder to move a heavy load. Missile actuators are kept small to save space and weight. They achieve the needed force by use of high pressures. The standard pressure during WWII had been 1000 to 1500 psi (pounds per square inch), this was increased to 3000 psi for Navaho missiles and to 4000 psi for the B-70 experimental supersonic bomber. We used 3000 psi for the Minuteman.
Fig 4.2.1-1 & 4.2.1-2 Minuteman Stage 2 Actuator Body
The fig 4.2.1-1&-2 actuator body is far different than what you’d see on construction equipment. The large hole on the right is where the piston goes. The round hole on the left where the servo valve plugs in. The upper left and lower right is where the actuator mounts to the solid propellant motor with two bolts. The flat face on the right and lower left is how each of the four actuators mount to the Nozzle Control Unit structure. In addition to the four bolt holes there are two others. The upper of these on the right is for the electrical wiring and the one below it where hydraulic fluid from the pump goes in and out. Designer Jim Jewel, came up with a way to make a “plug in dual port”, which is more readily seen in the subsequent stage 3 servo photo. The position transducer body screws in at the back end of the piston housing. These are obviously costly to machine – almost works of art. But what is not apparent is that with 3000 psi hydraulic pressure, they have more force than the hoist that lifts a car in a repair shop.
The Minuteman actuators were also a very innovative packaging design. They used an unbalanced piston, where the shaft came out of one end only – which eliminated one shaft seal. The position transducer body, fig 4.1 , screwed into the back end of the cylinder, extending inside the piston. The movable flux coupling probe screwed into piston shaft.
4.2.2 Servo Valve
Fig 4.2.2-1 Diagram of Moog Servo-Valve
The servo valve was a very important post WW II invention – that serves all of us today.
The servo valve can be divided into three parts: an output stage, pilot stage (hyraulic amplifier), and torque motor (actuating motor).
The output stage ,at the bottom, has four openings called ports. The pressure and drain (return) ports from and to the Pump, and the piston extend and piston retract ports to the actuator. Flow through these ports is controlled by the positioning of the “spool”, centered between two springs. The spool is shown in the cut off position where there is no flow to or from the piston.
The pilot stage (hydraulic amplifier), at the center, has two tubes that flow oil, as if free flowing garden hoses, emptying into the return. Oil from the pump flows up and branches to left and right passing through two small orifice-filters, dark spots. Each side continues on to the end of the spool and branches up to the pilot stage left and right tubes (garden hoses) dumping into the return drain.
The torque motor (actuating motor), at the top, is commanded by up to 8 ma (miliamperes) of control current, which moves a central flapper by electromagnet pull. The flapper extends down between the two free flowing tubes. When the flapper moves to one side or the other, it tends to block the flow from one tube and allow more flow from the other tube. When this flow is impeded or aided, it creates a pressure difference against the ends of the spool. By moving the flapper, the spool is moved against the centering springs.
When the spool is moved it ports fluid to one piston port and from the other port. The flow is proportional to the electrical signal applied. In this form it is a flow control valve. (there are also pressure control valves.)
Electrical current to the flow control valve controls the rate (speed), and direction, of the actuators piston. When there is no applied signal, the piston is held (locked) in place.
The 8ma valve current became a standard maximum when vacuum tubes were used, that was all a vacuum tube could produce. The middle pilot stage is called a hydraulic amplifier because it boosts the feeble torque motor flapper motion to a powerful differential force on the spool, it uses the abundant hydraulic power to move the spool. In fact it was required that the force on the spool be powerful enough to sheer piano wire – as proof it could overcome contaminates.
The cost of hydraulic amplification is the continuous hydraulic leakage from pump to drain, failure to account for this was to cause redesign of the system just before first flight.
Dry Torque Motors The Navaho servo valve torque motors had been immersed in oil, contaminates collected in the torque motor inhibiting motion and the designs were changed to be dry torque motors. This attribute which would lead to a single torque motor bi-propellant valve for small rocket engines. We had worked very close with Moog during the Navaho program helping to refine servo valve designs.
Fig 4.2.2-3 Similar Servo Valves
Fig 4.2.2-4 Raymond Atchley “Askani jet pipe servo valve, used on experimental models
4.2.3 Position Transducers
The position transducer is used to tell the servo-loop electronics where the actuator piston is.
[A transducer is a device which converts something like position, pressure, temperature, etc to an electrical output.]
Fig 4.2.3-1 Collins Position Transducer. Fixed body bottom screws into the back of the cylinder body.
Movable flux coupler at top slides inside the transducer body and screws into the piston shaft – it is immersed in oil.
Fig 4.2.3-2 Unbalanced (single shaft) actuator body and piston, with transducer body and movable probe installed
The transducer body, lower part of fig 4.2.3-1 screws into the back of the actuator body fig 4.2.3-2, is immersed in hydraulic fluid and extends inside the piston. The movable probe (slug) fits inside the transducer body and extends to attach to the inside end of the piston shaft. Thus it is mechanically protected and there is no binding because the actuator and transducer are concentric. [For other designs it was often necessary to place the transducer on the outside and the transducer shaft would often bind if the actuator piston rotated.]
Fig 4.2.3-3 Position transducer schematic
The transducer body houses a transformer wound about an inner tube, the movable slug fits inside.
The transformer has a primary (input) and two secondary (output) windings connected at the middle. One output winding produces an extend (plus) and the other a retract (minus) signal. The input is flux coupled to the output by a movable slug. [When a primary (input) winding of a transformer is excited with an alternating current (AC) signal it creates a flux field. This flux field will magnetize iron material within it’s influence, and induce a current in a winding within it’s influence. A “slug” (core) is used to concentrate the flux influence between windings.] When the slug is centered, the extend and retract cancel and the output goes to null, see centered slug position above. When the slug moves toward one end or the other the output increases plus or minus, to represent extend or retract position. The output is a linear definition of position. [When centered the output does not go to a full zero. A vacuum tube voltmeter (VTVM) was used to find center by searching for when the output reached it’s minimum level – called null. A VTVM was used to completely isolate the instrument from the signal being measured.]
For Navaho application the primary coils had been excited using sine wave alternating current (AC). Since only battery power was available on a solid propellant motor, it was necessary to chop DC battery current which produced “square wave” AC current. Gary Colins was the only one who could produce a linear output transducer when square wave excitation was used. [Gary Collins worked at the company and played cards at lunch with Flight Control people working on the Navaho missile. After listening to their problems with position transducers he came up with one, using methods used to tune isolation transformers in a radio. These were very successful and he left to set up his own company. Soon all US missiles used his position transducers.]
4.3 Feedback Demodulation
It is necessary to convert the square wave output of the transducer to an equivalent plus or minus varying magnitude DC (direct current), this is done with a Demodulator.
During the Navaho program I had studied and studied the design of a vacuum tube demodulator trying to figure out what it did and how it did it. Being told it was a fancy, full wave rectifier didn’t help much, however that’s what it is. Fig 4.3-1 is a diagram of a full wave rectifier which converts alternating current to direct current. The diodes are like liquid check valves, fluid can go one way and not the other – a rectifier accepts an alternating signal in and sends a DC signal out. A standard rectifier could not be used because the diodes will not pass a low level signal, they are like a small dam, you have to lift voltage high enough to flow “over” them.
Fig 4.3-1 Transformer followed by Fullwave Diode Rectifier
The demodulator solves the signal threshold problem by using a transistor (or vacuum tube) switch in place of a diode, this will pass low level signals. Using opposite pairs of transistor switches, one pair commanded on and the other pair commanded off, using the excitation signal as the command to the switches, does an excellent job of solving the problem. This mechanization is called the demod.
The DC output from the demodulator is filtered -- smoothed – by resistor-capacitor filters before it is sent to the summing junction.
The Square Wave problem: Square wave, corners are the equivalent of very high frequencies and have a bad habit of radiating “noise” which in this case would show up on the servo valve signal – something not encountered prior to Minuteman. How this was solved is discussed later.
4.4 Servo electronics -- Analog Arithmetic – Loop Gain
Fig 4.4-1 Summing Junction & Gain Control
The position command and “negative” position feedback come together, via resistors, to a “summing junction” the output of which is the “error”. This is analog electronics way of doing “arithmetic”. The amplifier boosts the error signal up to 8 ma of servo valve torque motor current.
Loop Gain, is adjusted by controlling amount of electronic amplification. If too little the servo is sluggish, if too much it can go unstable. This is taken up under the topic frequency response.
4.5 Servo-Actuator Packaging
Mounting the servo valve was more controversial. After much debate it was decided the face seal method, fig 4.5-1, used on Navaho servo valves was not reliable enough – they had been a frequent cause of leaks, usually due to the use of the wrong size “O” rings when replaced. Thus the Minuteman servos were designed to use a “plug in valve” with shaft type “O” ring seals. They never leaked but were more costly to make and test. We were very sensitive to the fact that the missiles would be out of sight for years in a silo, and even small leaks would be unacceptable.
Fig 4.5-1 This Navaho high temp experimental servo valve by Cadelac Gage shows four face seals top left.
Fig 4.5-2 Stage 3 Servo actuator aft view Fig 4.5-3 Stage 3 Servo actuator shaft removed.
The left view shows where the servo valve plugs in and where the position transducer screws into the back of the cylinder cavity. The right view shows “O” rings on the dual (pressure and return) hydraulic port, there are no face seals. The right view also shows where all electrical wires come out as one wires bundle. There are no exposed wires on the servo actuators and all wiring and hydraulics pass through drilled passages in the structure that holds the four actuators and shared pump and control electronics. A very compact and reliable design.
Fig 4.5-4 Stage 3 actuator housing, shaft & fluid port Fig 4.5-5 Stage 3 actuator housing servo-valve & transducer port
“Z-Links” attached in slot lower left and to the motor nozzle “stove pipe” that held Tilting Nozzle
4.6 Make or Buy Servo-actuators
Our organization was one of the most advanced in servo actuator design manufacture and test. We were doing them for the B-70 mach 3 bomber, the Hound Dog missile and were proposing them for a new F108 fighter. Those operations were under Jim Anderson. The Minuteman actuators would operate at room temperature and were not considered a technical challenge. It was decided to buy the Minuteman servo actuators from Ling Tempco Vought (LTV) in Dallas, TX and the Hydraulic Pump systems from Vickers in Torrance, CA. Art Greer and Lou Purpura had been responsible for writing the specifications and getting the first buy’s under way. At the time it was anticipated that Autonetics would be building the more complicated high temperature control servos for the B-70, Supersonic Transport and new F-108 fighter – none of which materialized into follow on work.
Fig 4.6-1 Experimental stage 3 servo front view Fig 4.6-2 Experimental stage 3 servo aft view
These used the Raymond Achley kind of valve mechanism
This early experimental actuator shows LTV’s solution to the valve seal problem, the servo valve was built into the actuator body. The round object on the right is a pressure transducer used on the first experimental flights only. The glistening object at the top is the valve torque motor cover.
5.0 Hydraulic System
5.1 Hydraulic Power Supply
Each stage had it’s own hydraulic pump driven by a battery powered electric motor and it’s own supply of hydraulic fluid. It was made up as one composite package, similar but of a different size for each of the stages. The pressures for stage I and II were 3000 psi and for stage III 1500 psi.
5.2 APS – Auxiliary Power System
The APS was an electric motor driven hydraulic pump system which included a reservoir, contamination filter and pressure transducer with plug in ports to the structure. Lou and Art did an excellent job of figuring how to package the APS parts to increase reliability, save space and weight. They did this in part by removing non essential parts. For example the pump housing was removed and the pump mechanisms placed inside the reservoir. The reservoir body was a bellows which expanded to contain more fluid when the actuators retracted and contracted to provide more fluid when the actuators extended. A filter to remove contaminates, and a pressure transducer for experimental test flights, were built into the assembly.
Art and Lou wrote and delivered a paper on the APS unit telling how they achieved innate reliability while at the same time reducing size and weight – their packaging was admired by those skilled in the field.
There was no conventional hydraulic tubing or fittings. The pump input and output ports “plugged” into the NCU structure which was drilled out to carry fluid to and from the actuators. The “plugins” were sealed with shaft type O rings and did not leak. (The actuator “dual port plugins”, fig 4.5-3 lower right with pressure inside and return outside, were designed by Jim Jewel to save space and weight – and be reliable.)
The hydraulic passageways, from pump to actuators, were drilled through the NCU structure. The “inline” hole, left in the structure as a result of the drilling operation, were sealed using “Lee” plugs. The “Lee” plugs are driven into the structure. The aluminum structure surrounding the plug is driven above the elastic limit (permanent deformation). This makes the aluminum structure sensitive to stress corrosion cracks – which did and still do occur, according to AN people in 1996.
Fig 5.2-1 Stage I APS
5.2.1 The Pump
Fig 5.2.1-1 Typical pump mechanism
The above figure, though not of a Minuteman pump, shows the primary parts. The cylinder block is at an angle, and as the piston driver plate rotates, the pistons move up and down in the cylinder block. The valve plate exposes downward moving pistons to the intake and the upward moving pistons to the outlet. By rotating the valve plate the pump output can be adjusted from none to maximum. By changing the angle of the cylinder block relative to the shaft, the maximun output of a given pump size can be adjusted.
For the Minuteman design, the valve plate was operated by a pressure control servo, the plate was rotated automatically to cause the pump to deliver the amount of fluid necessary to maintain system pressure. Thus it was called a Compensating Pump, it only delivered what was demanded. This reduced the energy drain on the battery power supply and heating of the motor armature.
The Minuteman pumps had a very special bearing, a ring of fixed angle Kingsberry pads to support the trust load on the cylinder block. Hydraulic pressure pushes the cylinder block in the direction of the rotating plate, and the Kingsberry pads served as skies, permitting the cylinder block to ride on a film of oil. The drive shaft, spinning at 20,000 rpm, was supported by a conventional ball bearing.
Fig 5.2.1-2 Location of flow plate determines output (prepared for depot crews)
Kingsbury had invented the bearings used to support the power generators at Hoover (Bolder) Dam. The generators spin in a vertical axis and rest on the ends of their shafts. Think of the Kingsbury bearing for that application as a plate cut in slices like for a pie, with the center cut out to make room for pivot pins for each slice of the pie. The generator shaft rested on the pivot-able slices. As the shaft turned the slices would pivot acting a ski segments – the shaft would lift up on a film of oil between the shaft and tilted slices. These worked so effectively that once when flooding replaced the oil pool with water, the generators continued delivering power while skiing on water. These small fixed angle pads on the Minuteman pumps were one of the main reasons the pump packages were so small.
The shaft seal became the most controversial part of the pump. What constituted a leak, a drip, was a wet shaft indication of a leak or a normal condition. It was finally decided that wetting was not a leak. Seals were tested until there was assurance that they did not drip.
5. 4 The Motor
The electric motors were of the DC kind as they operated from battery power. This kind of motor has a fixed electromagnetic windings in the case which pull on “rotating” electromagnets on the armature (the rotating part). Brushes carry current through cooper segments, called the commutator, at the back end of the armature, these segments connect to windings on the armature which create the electromagnets. The segments become the critical part of the pump motor and thus the entire APS. The armature cannot be cooled, and builds up heat – it’s designed to last long enough to do the job, before it burns out. The motor spins at 20,000 rpm (revolutions per minute). The armature coil wires solder to the cooper commutater segments. Holding the commutater segments in place as the solder gets hotter and hotter is very important – and difficult. The segments were held in place by insulated “piano” wire wrappings. Thankfully the motor manufacturer knew his business and we never had a motor fail prior to it’s design life.
It was important that the pump duty cycle be well thought out and defined in advance. A mistake was found before first flight which required a complete redesign.
5.5 Reservoir -- Oil Supply
The Reservoir was required to have a variable capacity. The actuators hold more oil when extended than when retracted. Thus the reservoir had to handle the worst case maximum and a worst case minimum. A metal bellows was used which was spring loaded to close. When in a condition of worst case minimum it was still required to maintain an inlet pressure on the pump. The reservoir also served as the housing for the naked pump inside.
There was some extra fluid allocated for seal leakage – though great measures were take to assure there would be no leakage.
6.0 Learning and Planning Ahead
6.1 Mechanic Learns about Electronics
My training had been as an airplane mechanic and mechanical engineer; with prior experience as a Civil Engr (materials testing and surveying for bridge and road construction), Process Engr (tooling for aircraft-& automotive); thus I found myself in a completely new element as Lead Engineer at the Bombshelter test lab. It was essential that I learn something about electronics. It was embarrassing to not understand the vernacular or names of parts. Some of the engineers were taking classes at night on topics like “Controls Systems Synthesis” where they learned the new mathematical methods applicable to servo controls. The company seemed to have an abundance of those fellows. I was daily confronted with how to deal with the hardware devices they brought to be tested, and build systems, that could be operated remotely, to test them – some tests were hazardous. They were using names and terms I didn’t understand. I began the learning process by reading parts catalogs and studying the wiring diagrams for hobby kits – they didn’t teach what I needed to know in colleges or night classes. I sent for catalogs, pamphlets, parts & kits – I learned by reading and doing – there was a wealth of material available for the person who looked for it.
Fig 6.1-1 Diode radio Fig 6.1-2 Transistor superhetrodyne radio right.
During the Navaho program, I built a “crystal” radio from a schematic in a Sylvania booklet advertising their new semi-conductor diodes. It had five parts: tuning capacitor, fixed capacitor, diode, antenna wire and ear phones. I fit the three “radio” parts inside a Skippy Peanut Jar. It worked great for the LA basin where there were powerful radios like KFI beaming down from Mt Wilson. The circuit diagram for this is shown in the next figure..
I next built an all transistor radio from a Miller Coil diagram. It would not work after my first attempt, I was very discouraged -- and very busy – I put it aside. A couple years later, determined not to let it defeat me, I took the coils apart – the windings did not match the diagram! I corrected my wiring to match the way the coils were wound and it worked! The quality was lousy, by then I knew how it should have been built to suppress noise and I moved on. It had served it’s purpose, tenacity had refreshed my confidence. I had emulated how IBM had made their computer cards, by drilling holes in formica sheets, bending the wires and soldering them. Printed circuits evolved from these and similar methods. The diagram for this is shown in the next figure
Upper left of Figure to the right shows the parts and diagram of the circuits inside the Skippy peanut jar. This is now in the museum at Oberlin KS.
I modified a later version that included one transistor by adding a second by direct coupling NPN and PNP transistors. I connected that to a small speaker with enhanced home made sound box. I used that as a Garage Radio for many years.
The bottom right diagram is the Intermediate Frequency transformer coupled full up radio of the kind later built and sold in large quantities, usually Japanese made.
Fig 6.1-3 All Transistor Hi-Fi with base, trebble and volume control for both channels
Fig 6.1-4 Included selector for Tuner, Tape, & Photo inputs – using very early transistors
Stages aligned using square wave generator & oscilloscope
When Bob Kelley joined us, at the very beginning of Minuteman, several of the fellows were assembling vacuum tube Hi-Fi Systems from kits, usually Heath or Knight kits. Bob said, why don’t you build a Hi-Fi using all transistors? I asked, what’s the advantage. Bob said you don’t need transformers especially those in the output amplifiers – transformers are the greatest source of Hi-Fi noise. He said his friends, at the radio transmitters on top of Mt Wilson, who beam broadcasts down over LA, pay close attention to their output amplifiers. Their amplifiers use very large transformers to achieve quality. From this beginning I proceeded to build my own all transistor hi-fi amplifier, now in the Museum at Oberlin KS.
From these experiments and Bob’s teaching I was to learn much about square wave. Bob said it requires a high quality amplifier to create the sharp corners of a square wave – the corners are the equivalent of very high frequencies. Bob recommended that I use square ware to establish the bias resistors for the amplifier stages of my hi-fi – replicating the square wave is much more demanding. Selecting the bias resistors is how you align signals to the linear range of the transistor. I adjusted transistor biasing this way for each stage, often having to go back and adjust a prior stage. When finished it indeed produced high fidelity – and I’d learned how to shield to block out noise. These home experiments proved very helpful at work.
This was ahead of it’s time, HiFi enthusiasts were still assembling Heath Kit Vacuum Tube sets and had not entered the world of transistors. Motorola & Delco of GM had just come up with the “door knob” power transistors as output stages for automotive radios. These Germanium Transistors were some of the first on the commercial market. The small signal transistors used were early devices made available to hobbyist, those used had double digit numbers – within two years many companies were making transistors and identifying numbers went up. The biggest drawback was finding good “large capacity Capacitors in small envelopes. I was using some I bought from the salvage yard and salvage from experimental board – in fact many of the small parts such as variable resistors came from such boards. It was a very interesting and rapidly changing time. I would over and over use things & principles I learned from these experiments on other applications.
While the Minuteman program was consuming more and more of Flight Controls attention, we were still working on aircraft systems, primarily in support of the LA Aircraft division. Bob McCoy was the aircraft projects engineer, with Jim Passwater an assistant project engineer, and Jim Anderson the lead design engineer. Jim had done the experimental designs for some of the B-70 servos, which were used on the flight vehicles, and Jim was also working on the F-108 proposal. Most people believed that there would be follow on work on a Supersonic Transport.
There was also research efforts going on headed by Fritz Gardner, with Harry Horowitz, RE Smith and Jim Jewel doing design work. Attention at that time had been on developing a “digital” servo valve, or “digital servo actuator”. Conventional wisdom was that now that transistors permitted building digital electronics, that there would be a need for digital components. Several years later, after Minuteman III, I shifted to doing studies on how to replace analog servo controls electronics with digital. At that time I came up with a way to adapt digital electronics to the analog servo valves and analog positions transducers. At the time of early Minuteman this did not seem possible and much work was done by Autonetics and others to develop digital hydraulic components.
Transistors 1956 style: Upper left options on how to connect a transistor, lower left how to bias a transistor, Upper right package types, lower right heat sink methods for power transistor. These kinds were used to make HiFi amplifier in prior photo and radio below.
Bob Kelly, a remarkably capable engineer.
I owed much to Bob Kelly. I’d gone to Odel Taylor, supervisor of MM I Flight Control Electronics, to obtain an Electronic Engineer to build electronics for us to use in testing the Minuteman parts. He assigned, in fact transferred Bob Kelley to Paris Stafford my boss who assigned Bob to work with me in the Bomb shelter test lab. Bob did not have an engineering degree and Odel wanted to retain his “best” to do the design work using the then new transistors. Bob Kelley was an early day Ham Operator who learned radio by building his own transmitter. Bob had used glass plates from a green house to make a high voltage capacitor so they could communicate with Europe from his home in Massachusetts. In time I noticed that the engineers with degrees, working for Odel, were consulting with to Kelley to determine how he’d managed to build his Minuteman servo control loop electronics. I watched and listened as Kelley demonstrate what he’d done showing outputs on an oscilloscope. They were all having a problem with noise generated by the 6 khz square wave used to excite the position transducers. Batteries were Minuteman’s only a source of electrical power, thus it was necessary to chop DC to make “AC” to excite the linear Inductive Transformers used as position transducers. This “noise” penetrated the otherwise clean analog signals from servo valve amplifiers -- the retained engineers were baffled on how to solve this. I listened as Bob demonstrated and told them that high frequency noise only looked bad on a scope, that servo valves were unable to respond to such frequencies, that the noise source was known and balanced, it was simply averaged by the valve torque motors – the valves operated on current noise only detectable to with instruments as high frequency voltage spikes. I would later make use of this principle when using digital chips as servo valve drivers. Bob was very capable and much appreciated by persons like me. Bob died about 1970.
Electronic Kits: This experimentation was essential to my learning how to perform remote control and instrumentation. I used Allied Radio parts catalogs to learn the names of parts and sent for kits to build at home to understand how things worked. Electronics engineers spoke to me as if I understood, it was embarrassing not knowing the names of common electronic parts. This hands on learning was very helpful. Transistors were new and I applied them to simple devices.
For example I used a phono crystal and amplifier from home built phonograph setup I’d made for my kids to remotely measure vibration of a high speed drive out in a test cell. It worked surprisingly well.
To align and test my all transistor hi-fi amplifier system it was necessary to build kits as Vacuum Tube Voltmeters, Signal Generators and Oscilloscopes to generate square wave signals and fine tune amplification stage biasing. Bob Kelley sketched how to make a transistor HiFi speaker driver without a need for conventional very large transformers used for vacuum tube HiFi systems. That all transistor Sterio Hi Fi Amplifier, shown above, was used as the family HiFi system in our living room while out kids grew up. It’s now on display at the Oberlin KS museum.
6.2 Preparing for B-70 and Supersonic Transport Testing
(Emulating a Transformer to Isolate Hot and Cold)
Fig 6.2-1 Capacitor coupled Amplifiers
Fig 6.2-2 Transformer coupled Amplifiers
Fig 6.2-3 B-70 Mach 3 Bomber by NAA
The B-70 equipment operated very hot, as had the Navaho equipment. To do the Navaho testing we had built expensive heat exchangers, of my own design, so we could use room temperature pumps for operating high temperature servo actuators in an oven – oil at 600 deg F and oven environments up to 1100 deg F. Johns Hopkins University had us build heat exchangers for them after having used ours. The B-70 program was still on going and LA division needed the equivalent heat exchanger system. My thoughts still included how to test systems operating at very high temperatures.
My experiments with electronics providing new insights on how to solve lab testing problems. I was quite interested in how amplifier stages were isolated yet passed an amplified signal from one stage to the other. Most hydraulics people thought of electrical resistors as being similar to hydraulic throttling devices and as hydraulic accumulators being similar to electrical capacitors. Quite suddenly I saw transformer isolation as being similar to hot and cold isolation! Where a mechanical link could serve the same isolation function as flux coupling. The idea came to me when looking at our high temperature seal test in operation.
The test rig was a pair of actuators mounted inside a pair of I beams with their shafts connected. The actuator with the test seals, was inside the oven, and the load actuator, connected to a relief valve system, poked outside the oven. One actuator was hot and the other cold an they were isolated by a shaft --but no fluid was exchanged.
I thought why not drive the cool load cylinder with room temperature hydraulics and convert the hot test cylinder into an oscillating pump. The shaft connecting the two was like a flux coupling between transformer coils. The external driving piston was like the primary winding, with AC strokes, and the one in the oven like the secondary winding, with AC fluid force. By connecting the oven piston to a full wave rectifier, using check valves in lieu of diodes, we could produce hot DC hydraulic pressure. We could “filter” the pulses by connecting “rectifier” inside to an accumulator outside the oven – like a capacitor – we could filter the ripples. And we could use cold accumulators to filter hot fluid.
Fig 6.2-4 Electrical Diodes to Hydraulic check valves
I proposed the idea to several people who said, “it wouldn’t work”. But I knew it would, I’d figured out how. In the main plant hydraulics lab there was a load fixture connecting two actuators. I had one of the technicians help me. We connected the “cold” actuator to the labs room temperature pump and commanded it using a pressure control servo valve. I rigged up a command signal using a knob operated potentiometer, and for negative feed back used a pressure sensor. Thus the pressure control valve would be commanded to maintain a defined pressure. I used a mechanical switch to reverse current to the pressure control valve when the actuator got to the end of the stroke.
We next rigged up a check valve, full wave rectifier, to the output of the “hot” actuator and filtered the output with an accumulator. Then to prove how well it worked, we applied the “hot” output pressure to a Minuteman servo which had been set up for test. With the Hot-Cold reverse motion pump going we ran a frequency response test on the Minuteman servo actuator. It was worked just great and I called in each of the fellows who said it couldn’t be done. They conceded it really did work. It had taken the technician and I less than a day.
I went to Paris Stafford and Fred Morgenthaler to see if someone would follow up on this with LA division. However nothing came of it, the B-70 was soon canceled and the US made the decision to not go into the Supersonic Transport business. Thus the great idea came to an end. That method would have saved us much time and money had I come up it three years earlier.
Fred called me in a day or so later and said we are bringing you in from the test lab and having someone else sign out for all that equipment, so report to me as soon as you can. Bob Parkinson took over that lab and later arranged for it to be moved to Anaheim, all the equipment was moved from Downey and installed in a new separate hazardous test building at Anaheim.
7.0 Servo Loop Testing
7.1 Emulating Missile Electronics -- for Servo Testing
We used the Bombshelter test facility in Downey for testing the first experimental servo actuators. This required that we have electronics that emulated what was used for the missile. I talked with Paris Stafford, in charge of Minuteman Hydraulics, and Odel Taylor, in charge of Minuteman Electronics about our needs. I needed electronics from Odel, or an engineer to help build our own. Odel assigned Bob Kelley to our lab – this choice could not have been better. It had been decided that Minuteman electronics would be all semi conductors, no vacuum tubes. Odel felt uneasy about using Bob for design of vehicle electronics because Bob did not have an engineering degree. We were to find that what Bob knew was not taught in school.
Bob had been a Ham operator in the early days of Radio and had built his own transmitter using such things as glass plates from a green house to make high voltage capacitors. He kept up with the state of the art, and amazed all of us with what he knew. He would often spend part of his weekends visiting with the fellows his age operating the radio stations on top of Mt Wilson, beaming signals down on the LA basin. I also became aware that the missile electronics designers were coming to Bob to find out how to make their designs work. Bob hovered over and experimented with the electronics he was putting together for us, often telling me of problems he was having. He’d often go see his buddy George Dyer, one of the best electronics designers in the company, to talk over some detail or learn about new transistors. Bob constantly kept himself up to date. Transistors were so new they were having difficulty finding power transistors to do thing like chop square wave for transducer excitation.
7.2 High Quality Square Wave Excitation for our Lab
Since they were having trouble finding transistors that could chop battery voltage without burning out, Bob recommended that we not wait, he had a solution -- buy commercial high quality hi-fi amplifiers -- the same kind his friends used on Mt Wilson. We did and they worked great. RCA finally came up with the transistor they used for the first flight.
7.3 The Servo Valve Torque Motor is an Integrator – it Averages Noise
The missile electronics designers found their valve signals were very “ratty”, as if something was very wrong with their design. Bob previously encountered this and told me, we’re picking up noise from that 5 kc transducer excitation, but we don’t need to worry about it. The valve coil cannot respond to such high frequency signals, the torque motor coil integrates high frequencies to an averaged DC level. Bob was not worried about noise on the valve command that appeared on an oscilloscope, he was only interested in the resultant applied torque. If valve flow behaved as it was supposed to, that’s what counted. It was not practical to fight inherent noise when it did no harm and for which there was no practical solution. Paris Stafford was to appreciate how valuable Bob was to us, but I’m not sure Odel Taylor ever realized Bob was teaching his graduate engineers.
I was to apply these lessons from Bob several years after Bob died – Bob made it possible for me to operate an analog servo valve using digital electronics when others said it couldn’t be done. Bob would have gotten a kick out of that.
7.4 Frequency Response Tests
The quality of a servo-loop is measured by performing a Frequency Response test. If something is wrong with the design or assembly it will not pass this test. To perform the test, an actuator is commanded to cycle back and forth at faster and faster rates. Two parameters are measured, the Amplitude Ratio and the Phase Shift. The actuator is commanded to a certain stroke amplitude at low frequency and the amount is defined as 1.0. As the frequency of cycling is increased the amplitude will sustain, and if the amplifier gain is high, the amplitude will tend to peak before dropping off. This ratio was sometimes expressed in db (decibels) by the people who initiated the measuring method. The command amplitude and frequency is controlled by the test operator using a test panel.
Fig 7.4-1 Plotting and measuring Amplitude Ratio and Phase shift
The Frequency Response control panel had a command amplitude adjustment, a frequency control dial and a phase shift dial. The phase shifted output was applied to X axis of the scope and the servo feed back to the Y axis. The setting on the phase dial when the scope pattern showed an “hour glass” , as shown to the right above, was the amount of phase lag between the command signal and the feedback signal.
The frequency is increased in steps and the amplitude and phase lag plotted. A manual plot of Amplitude Ratio and Phase shift, above left, uses the same X frequency scale and a different Y scale.
If the gain is set too high it can cause overshoot as frequency increases, overshoot is not to exceed 1 db, more could cause control instability. The servo loop is considered OK if it was not down more than 3db by a certain frequency and phase lag.
Amplitude ratio was often measured using a Sanborn recorder, the kind developed for hospitals for recording cardiac performance. Amplitude ratio could also be measure from the Y amplitude change on a scope, however it was preferred to use the plotting recorder as the plot could go into a test report.
Fig 7.4-2 Allowable limits for Amplitude Ratio
Fig 7.4-3 Allowable limits for Phase shift
7.5 Failed Frequency Response – A Measure of it’s Value
During the Navaho missile program I had been asked to run a frequency response test on the G-26 booster. It had been failing the test and was becoming an embarrassment when why it failed remained unknown, The prior chapter tells of the test and the determination of what was wrong. That serves as an excellent example of the value of a frequency response test – it can detect a “sick” servo loop, that works but is not right.
7.6 Shaping Networks – Filters
The Hound Dog missile was found to have a body bending problem, an accelerometer would sense body bending and send unwanted signals to the control servos. If the system was activated in the factory, and someone slapped the fuselage, it would trigger accelerometer commands and cause the missile to have “the shakes”. To solve this problem it was necessary to design a “notch” filter which would blank out the unwanted body bending frequency yet let normal control signals pass. [Clarence Asche evolved the notch filter and some two years later was sent out into the field to test all Hound Dog missiles to see if the notch filters had shifted with time. I would later use this experience as and argument for using digital electronics which did not shift with age.]
Servo loop electronics would often include “shaping networks” or “filters” placed in the command or feedback leg of the control loop. All control electronics at that time were analog, and signal shaping was done using combinations of resistors and capacitors to modify the dynamics of a signal. There was concern that the Minuteman might need such signal conditioning.
8.0 Move to Fullerton
8.1 Downey to Fullerton
Autonetics had decided to build a new plant in Anaheim, Space Division remained in Downey. While the Anaheim facilities were being built we in Flight Control moved to Warehouses, owned by Howard Hughes, in Fullerton. Bldg 60 for Engineering and bldg 61 for Manufacturing.
Table I Flight Control Engineering (Fullerton) dept 3446
T K Shuler
F H Gardner
W F Harrington
Missile Projects, Proj Engr
E W Velander
Missile Proj, Asst Proj Engr
R J Nadalet
Minuteman Proj, Proj Engr
R L Curci
MM Proj, Asst Proj Engr
Airplane Projects, Proj Engr
R N McCoy
Design Assurance, Proj Engr
W F Yetter
Develop Engr Sect, Sect Chief
R W Bond
MM Sys. Grp, Grp Leader
R F Henderson
E D Simkins
J E Stewart
Components Grp, Grp Lder
F W Morgenthaler
J E Hawkins
W O Taylor
P E Stafford
Systems Grp, Grp leader
E E Ward
Airplane sys unit
S W Bresenoff
Missile sys unit
A F Cooper
Prel Engr Sec, Sec chief
E A O’Hern
System Analysis Unit
R K Smyth
H L Ehlers
J M Johnson
Note: Don Williams and his Asst Elliott Buxton were in charge of Data Systems which included Flight Control. The Data Systems D-37 Computer was under G. B. Way and missile Cables were under Frank Henderson. Engineering troops at Fullerton included: Jim Anderson, Bill Strobel, Art Greer, Lou Purpura, Dave Byles, Kurry Woo, Clarence Asche, Ron Frazinni, George Leonard, Jim Erb, Parker Gasper, D Landau, and others
8.2 Analog Computer Controlled Load Cylinder
Art Greer came to me and said they had no idea what the loads would be for tilting nozzles. He was aware of the unexpected problems with the Hound Dog missile control instability and wanted to be prepared for the unforeseen. Paris Stafford and I had often discussed our method of creating an actuator load, for high temperature seal tests; as it did not represent what the actuator would actually see.
Most of the controls analysis fellows had been taking after work classes, using text book Control Systems Synthesis by Truxel. These analytical methods had not been taught when many of us were in school. Though I was not doing that kind of work, I bought a book about servo controls by Savant and had studied it on my own -- so I could at least understand what they were talking about. Years later, Joe Cherney who worked on the MM I Digital Computer, told me that servo control theory began with fire control systems in WW II; he had worked on those and the method was kept secret – thus not generally known. Walt Evans one of the Inertial Navigation engineers during the Navaho program had come up with the “Root Locus” (1948) method of determining if a servo loop was stable. Behavior at specific frequencies was plotted to predict stability or instability.
Calculations were made, using analog computers, using methods called controls synthesis and a mathematics called Laplace Transforms. When I was in school we were taught the use of differential equations which was performed using Differential Calculus, we did not learn of Laplace unless you were a math major. Laplace would be superceded by Z-Transforms and all of that by Sample Data digital methods when digital computers took over from analog computers.
Art and I talked about the kinds of loads such as: Coulomb friction, Spring a function of position, those a function of velocity, and those a function of acceleration. In differential equation terminology these were expressed as: Load = Kf (friction) + Ks*x (distance spring) + Kv*dx/dt (velocity) + Ka*dx/dt(dx/dt) (acceleration). Where Kf, Ks, Kv, Ka are mathematical coefficient (values) for each term. Where x is distance, dx/dt the rate of change in distance, dx/dt(dx/dt) the rate of change in velocity [When a freshman in college I had trouble with some of the terms. To me the differential was the gears in the rear axle of a car, and I’d never heard of an increment. I finally grasped that dx/dt expressed the difference in position vs time. To save writing, x with dot above, was used as short hand for dx/dt. A single dot for the first and a double dot for the second derivative. (I’ve not found such a symbol to print here.)]
Art & I figured that if I could come up with a means of “programming” a load cylinder to simulate these various kinds of loads, and to provide a general purpose set of resistors and capacitors with which to create shaping networks; we’d be able to create any load encountered and create shaping by test. I talked it over with Paris Stafford who said go ahead, see what you can come up with.
We had I-beam test fixtures on which we could place a Minuteman actuator and operate it against a load cylinder. (You could watch the force of those small actuators bend heavy steel plate.) We were using the conventional relief valve method which created a “friction” load. With this as a beginning, I connected the load cylinder to a “pressure control servo valve” and rigged it to change pressure as a function of position. I then started to rig up a velocity and accelerometer sensors. Then one of the fellows said you can convert the rate of change in position to be velocity and the rate of change in velocity to be acceleration – they do that in the simulation lab using analog computers.
Again I talked it over with Stafford who said, fine but you’re going to be very busy, don’t try to build this yourself. Define what you want and we’ll buy it.
So I defined the setup that we wanted, and purchasing sent it out to a supplier. I spent an entire day talking with the supplier, and was very impressed -- he immediately knew what I wanted to do, how to do it. He left information on Philbrick Operational Amplifiers, the main ingredient of his solution, for me to read. He was given the go ahead and I read about Operational Amplifiers. My “load simulator” had became an analog computer operated test fixture. [Most of the Flight Control dept employees were young and eager to advance their careers, especially in becoming stronger professionally. As a result after hour education work became a way of life. Autonetics encouraged this trend by offering numerous company training/class room opportunities and by provinding financial incentives to work toward advanced degrees.]
8.3 Operational Amplifiers and Chopper Stabilization --Connections
During WW II, when being shipped by train from FL to CA, I bought a book on Radio at a lunch stop. It described the “superhetrodyne” radio, the name dropped as all were made that way. This used an oscillator to create a carrier frequency. The received signal was superimposed on the carrier by use of AM (amplitude modulation) or FM (frequency modulation) methods. This AM or FM carrier was passed from one amplifier to another using intermediate frequency (IF) isolation transformers. The small radio signal was mixed with the carrier frequency, then detected (separated from the carrier) and sent to the output amplifier to drive the speakers. Vacuum tubes could not operate when signals were in the “mud”, by superimposing them on the carrier frequency they were lifted to a linear range for the vacuum tube.
Operational Amplifiers used choppers to create a carrier frequency, it was the old stunt with a different name. The chopper was the radio’s oscillator. Inside the chopper can was a coil and points, a relay, which operated like a Model-T Ford ignition coil – DC electricity passed through the points to the coil, the coil pulled the points apart breaking the flow of current. These chopper mechanisms on an early car radio was called a vibrator.
An arrangement of electrical parts, which performed the functions of a mechanical chopper, was called a “free floating multi-vibrator”. Transistor multi-vibrators used a pair of transistors and called Flip-Flops. It was found these could be Set or Reset and hold a given state – thus the flip flop became a memory cell and could be interconnected to perform logic – digital electronics was born. There are many connections from one thing to another – by different name or method. Technology transitioned from Model-T coils, to vibrators to flip-flops to digital ignition.
The chapter on MM2 will tell of the solution to inductive kick back for transistor controlled solenoid valves – an ingredient of digital ignition that replaced distributor points on a car.
8.4 Load Simulator Completed – Not Used
Parker Gasper was put in charge of setting up the Load Simulator system when it arrived. But it had no more than become functional when progress passed it by – it was not needed.
Someone from Systems Engineering had arranged to bring in the aft end of a stage I motor case. I saw Josh Stewart and Tom Shuler looking it over outside the door of the plant. Others had gathered about and questions were asked how they were going to get a big heavy thing like that in the plant. Such a “problem” had been common in our Navaho test lab. I was standing next to Shuler so I said, give me a couple days and I’ll have it mounted on large casters so it can be moved about in the test lab. Josh heard, they nodded OK and two days later it was in the lab. A week later an experimental nozzle control unit was installed on it, the setup operated, upside down, flipping nozzles, proving it worked. There were no mystical loads to be simulated or compensated.
8.5 No Shaping Networks were required for Minuteman Nozzle Control
Kurry Woo, doing controls analysis, found all we needed was a simple “35 radian filter” – years later at retirees lunch Kurry is sometimes greets as, “there’s old 35 radian”; and smiles radiate recalling those days. Kurry grew up in Flagstaff AZ where his parents, of Chinese origin, had a restaurant.
8.6 R&D effort to develop a Digital Servo Valve
Fig 8.6-1 R&D Digital Servo Valve
At the time of Minuteman I it was considered through out the industry that with the advent of digital electronics, it would also be desirable to come up with a digital Servo Valve and Digital Actuator. Much money was spent by many companies in the business in attempts to develop same. Fritz Gardner was in charge of this effort at Autonetics. R E Smith and Jim Jewel worked on the design. Harry Horowitz was the test engineer. Before I moved to devote all my time to Minuteman Harry asked for me to buy him a Memo-Scope for testing this valve. An Oscilloscope could show a trace, but it was desired to have something that would memorize and capture for repeat display the rapid transient that vanished in the blink of a eye. Later after Minuteman III I changed to the field of digital electronics and came up with a way to operate an analog servo actuator direct from digital logic devises, voiding the need for a “digital valve” and designed the equivalent of a memo scope by use of digital memory chips, that could capture and re-sweep events on a screen. I used the case and display of a hand calculator to command the interfacing electronics – it was “home made” but it worked. This Digital Valve above did not work out and the Memo-Scope was turned in to the Instrument lab in case someone else had a use for it.
9.0 Lab to Lead
9.1 Greer and Purpura to Project – Landau to Lead Engineer
Art Greer and Lou Purpura were moved to the Project Office and I was brought in from the Lab to be the Lead engineer for the Power Elements, actuators and pump system for the Nozzle Control Units. Assigned to me were three of the new and very capable engineers; Clarence Asche, Ron Frazini and George Leonard.
9.2 Servo Valve Contenders
Dick Myers, and Bob Flippo were lead engineers at LTV and among the first we worked with. Autonetics, had pretty much dictated the design of the servo actuators by virtue of how they were mounted in the NCU’s. LTV used the Collins position transducers because they were the only one who could provide a transducer that worked with square wave. LTV looked into the choices for a servo valve. There were three contenders. Achely, Hydraulic Research and Moog. The Achley valve was different than the others, it worked on the Ascaini principle, where the torque motor aims a small tube squirting oil at two small holes; one of which would cause the main spool to command extend and the other to command retract. Initially LTV had in mind building the servo valve into the actuator using this mechanism. See photos of this actuator, now in the Museum at Oberlin KS. LTV decided to use valves completed by the manufacturer.
It was decided to use all three valve types with a different one on each of the stages. Each of these servo actuators was put through Qualification testing. The Qual tests were run by Autonetics and when a unit failed it was sent back to LTV who performed a failure analysis on the unit and presented their findings and design modification at the next Technical Direction meeting. We were fortunate to have very high quality people on the program from LTV and Vickers. Both contractors were very professional and provided in depth coverage of all aspects.
Early on there were delivery problems on the Achley valve. Then there was a critical failure of the Hydraulic Research valve which they did not pick up on and fix. The torque motor on the Hydraulic Research valve broke loose in vibration tests. An index piece, of torque motor potting compound, broke leaving the coil rattling about. Had they immediately fixed it with a metal pin they may have stayed on the program. This was just before first flight and a switch was made to a Moog’s valve.
The Moog valves performed very well so it was decided to use Moog valves on each of the stages.
9.3 Rechecking Design Requirements
When Asche, Frazini, Leonard and I took over what Art and Lou had been doing we knew nothing about it. Art and Lou had been overwhelmed with work and delegated much of the design requirement and specification details to others. One of them Jerry Colkin had moved to another job, and we were on our own. The four of us decided we had to catch up on what had gone on. We spent almost a week gathering all the correspondence and notes left behind, stacked them by date and subject and started going through the stack. In the process we found the requirements didn’t look right. We checked each others calculations and found that no one had included the servo valve leakage in the calculating demands on the pump. Also all demands had been calculated on pure pitch or pure yaw activity, not for a worst case of 45 degrees requiring in both pitch and yaw. We discovered that the actuator force levels to handle worst case were too low, we needed an actuator with 125% more force and a pump capable of about 150% more output. The impact was not so severe on the upper stages but the stage I, coming out of a silo and hit by side winds, would be very vulnerable.
9.3a Two Sentence Note to Curci
Once we were sure of our numbers I went to those in charge trying to tell them about it. I couldn’t get anyone to take their focus away from what they were doing. Curci had a room full of people trying to solve some problem and I couldn’t speak with him. I went back to my desk and composed a note:
“There is a good chance the first flight will fail.
The actuators and pump for stage I are undersized.”
I gave the note to Curci’s secretary and said be sure Ray reads this just as soon as his meeting ends.
About 15 minutes later Curci was at my desk saying; Now just what the hell is this!
I explaining the calculations we had made and how we arrived at our conclusions. I said it’s one of those things, each a very easy item to overlook when your pressed to come up with answers. Ray said nothing for a moment then said, I’ll be back. I didn’t know it at the time but Bill Strobel, one of the sharpest and most respected engineers had endorsed the calculations as being correct. Ray checked with Bill who reluctantly agreed a mistake had been made and our conclusions were correct. [I would later work for Bill on Minuteman II. He was a very good friend. He had never been married and found the girl of his dreams, got married and they had a child, while they were building a new home. Bill abruptly died of a heart attack near the start of Minuteman III after assigning me to help evaluate a proposal which turned out to be the Post Boost Propulsion System.]
I’m sure Ray also conferred with his boss Tom Shuler as well as Ed Ray and Jack Tilletson of TRW. A day later I explained our calculations to Jack Tilletson. It was decided to postpone the first flight. I was told to move out fast and fix the problem. I had told Jack Tilletson that if conditions were right it could fly just fine, but if not we could loose the flight. TRW did not want to take the gamble and put their reputation in jeopardy.
9.0 90 day Turn around
It was remarkable how fast LTV and Vickers responded to the change. Later versions of Minuteman could not have been redesigned, built, and qualification tested ready for flight in such a short time.
9.4.1 New Servo actuator
While at it I thought it was best to have LTV move the servo valve location on the stage I from at the back to closer to it’s mount. I considered where it was to be vulnerable to harmonic cantilever vibration amplification. Dick Myers of LTV and Lou Purpura of Autonetics agreed with my gut feel appraisal and the valve was moved when the actuator design as resized to have 25% more force output. Actually they were able to continue with the same servo valve and transducer – mostly making the actuator cylinder and piston larger.
9.4.2 New APS
The Vickers people were asked to go from a 5 hp to a 7hp APS. I went with the Vickers people to their electric motor supplier, those would be the long lead items. It was an interesting meeting. I found the motors were spinning the pumps at 20,000 rpm and required to keep running as they became hotter and hotter. There was no way to cool the armature. Vickers had selected well as the motor supplier knew his business. They had evolved how to use piano wire to wrap and hold the armature commutation segments in place at very high rpm – and temperature. It became important that we properly define the duty profile of pump activity, as this determined how much heat would go into the motor. The fact of servo valve leakage the moment the pump was turned on contributed to the heat load. Though the amount of fluid was not great, it was at 3000 psi and the pump efficiency was very poor at low flow rates.
They were still able to use the same basic pump, but did have to change the “wobble” plate angle to increase flow rate. In effect it was a complete redesign, reservoir, foot print and all.
9.4.3 New Battery Requirements – Simkins Blows his Cork
Once we had come up with a new stage I pump size and duty cycle it was necessary to tell Dan Simkins system organization about it. I mailed them a memo stating the requirements – when I released the memo I didn’t know this required a new battery.
Dan Simkins called a meeting and I took Jim Erb who had just hired in with me as he’d helped with the calculations. Dan had not been in on the Curci thing and the decision to redesign. As the meeting got under way Dan became angrier and angrier directing his hostility at me. (He must have believed I was the one who sized the thing wrong in the first place.) At first I didn’t catch on, and the more he ranted the less sense it made – he was accusing me of causing problems that were not remotely related to me. I began to realize he’d flipped his lid. I turned to Jim Erb and said there is no point in staying to listen to this. Jim I agreed and we left the meeting.
I reported back to Fred Morgenthaler who was wanting to know how the meeting came out. I said, Fred Dan needs vacation – he’s exhibiting all the symptoms of overwork – he’s accusing us of being the cause of problems we have nothing to do with. Jim Erb endorsed everything I said and filled in as Fred wanted to know more. I don’t know who Fred saw but Dan took a weeks vacation starting the next morning. Josh Stewart was in charge of buying the batteries and started a new purchase the next day. [I would later work for and with Dan on the start of Minuteman III. We never spoke of this event though I’m sure he remembered it. We got along well, but all of us were wary about Dan’s sometimes being close to the edge. However we all admired and liked Dan – we all had our mood swings.]
9.5 Filling the Sealed System with Hydraulic Fluid
Jim Anderson had hired Dr Repert, chemistry professor from UCLA, to work for us during summer vacation. Someone had assigned him the task of looking into how to fill the NCU’s with oil. The request had probably come from Systems Engineering as that was their responsibility. I don’t recall how I became involved, but someone said they were running out of time and wanted me to look into it. I was to find a way to fill the four servo-actuators, structure passage ways and APS with hydraulic fluid.
What at first seemed simple task, I soon realized was a bit of a problem. The actuators being “unbalanced” held a different amount depending on where they were positioned, and the APS accumulator was to supply or absorb the variation in what was needed. It was simple enough to retract all the actuators to put the most fluid in the reservoir and then fill it to full by virtue of the reservoir bellows position. However, how did you get all the air out of the system? We were well aware that air will go into solution, be dissolved in oil and that it can cause cavitation at the pump. Air bubbles can erode pump parts or leave the pump trying to suck a bubble and pump nothing. System uses positive pressure at the pump inlet to prevent this. The reservoir bellows behaving as a spring took care of that for our system. [I was home on vacation visiting Howard Saum, the head mechanic at the Ford garage. I had worked with him when I first got out of service and we were good friends. He had just overhauled a tractor and was trying to get the front lift to work – nothing moved. I was standing in back of the seat which was above the transmission. I asked where is the oil, where is the filler. He said there in front of you, it used transmission oil – there is plenty of oil, I just checked it. There was an over head air hose, I pulled it down, cupped my hand about it and squirted air in the opening – that was it, it had an air bubble, and worked immediately. He said nothing at the moment, and later on asked now how in the hell did you know that would work?]
I came up with the idea of using a vacuum pump, in fact brought the one from the bombshelter in Downey to the lab in Fullerton. I wrote a procedure, with diagrams, on how the vacuum pump was to be connected to the Nozzle Control Unit, after all parts were installed; and to the container of fluid to be used to fill the system. The NCU hydraulic passages were to be sucked empty of air and the fluid used for filling was to be purged of entrained air – while being sonically vibrated while the vacuum was in effect. When the evacuation had been held at a low lever for a period of time, the vacuum was cut off and the fluid permitted to rush in and fill the evacuated system. It worked great for our experimental systems and so far as I know is probably still in use today. I don’t know what Repert’s problem was, he was a very nice fellow and certainly knew chemistry – but like many analytical people didn’t relate to real hardware systems. I had to find text books on how much air could go into solution in oil and the conditions required to remove it.
10.0 Meetings with LTV & Vickers
10.1 TD Meetings with LTV
The ’49 Kaiser: I was on my way to work, thinking of the meeting we were scheduled to have with LTV, when my car quit? I was about four miles from home, it was my $150 1949 Kaiser with a caved in driver side door, a casualty of a collision in the parking lot at Downey a year prior. Andy Laslofy, one of the technicians, had helped me overhaul it and it had been running just fine. I lifted the hood and saw that they engine had listed about 20 degrees to the right, an engine mount had given way. I walked to the nearest phone to call Mary to pick me up. A wire from frame to engine ignition had pulled apart. Walking back to the car I saw a six inch piece of bare wire laying in the road. I picked it up and used it to splice the broken wire. I tried the starter, the engine worked, so I drove to the phone and called Mary saying, never mind I have it working and will drive on to work.
The meeting had lasted until about 7:30 pm and while visiting with the LTV people the others had left. Suddenly I recalled my car and called out to the LTV fellows, don’t leave until I can see if my car will start. Curious, they came and looked at the engine saying, you mean it will run that way! I said it got me here, the carburetor system still works at an angle. It started and that $150 transportation car became a conversation item, it was a nice looking car except for the smashed door which gave it character.
Steve Barnaby had been the reason the meeting had lasted so late. Steve was a chemical engineer by training, a bright fellow who had never married. He was gregarious and very helpful to those seeking information -- but you could never get him to write a report. I figured he was such a perfectionist that he was afraid someone might find fault with something he wrote. This was one of our early meetings with LTV and there was much discussion about oil, elastomer seals, etc. Steve had gone to the black board to explain something and LTV people were so fascinated with his depth of knowledge they asked him questions for over an hour. When we left the meeting Dick Myers said to Lou Purpura and I, if Barnaby is ever looking for another job send him to us – he’s fabulous. Not too long after that Steve went to work for Gary Collins as specialist and salesman for Collins position transducers, a very hot item in the aero systems industry at the time.
Design Review Meeting at Boeing: Larry Hein, one of the technicians, went with Elliott Buxton to the Minuteman I Design Review held at Boeing, chaired by TRW and the AF. All the hardware was placed on display which included the stage I NCU. Larry & Buxton were kept busy, drafting other Autonetics people there in wiping drips of oil from the floor with their shoes – while AF and others looked over the NCU. They’d wipe oil with their shoes, go to the john and clean their shoes and return to wipe up the next drip. When they returned I wrote a report, sent it to LTV and the shaft seal was the primary topic at our next TD meeting. The LTV fellows had made cartoons of how the leak had been discovered, showing Larry and Buxton wiping drips – these humorous charts identified the problem and were followed by the serious discussion of the solution.
The problem was that the actuator shaft seal did not have enough squeeze on the shaft, and low pressure from the APS reservoir bellows pushed drips past the seal. For long term storage this would be a disaster. In normal use, system pressure squeezed the seal against the shaft. Army Navy Standards defined the use of front and back teflon backup rings with an O ring between. The inner teflon ring served as a pressure plate against the rubber O ring which squeezed the piston shaft. The outer teflon ring protected the squeezed O ring from being pinched chewed away. Shaft seals were a design problem for every such application. The commercial world could be allowed a shaft to seep and use materials we could not.
The solution, after LTV performed many tests, was to use a special teflon “cap seal”. This was like a U, teflon member, into which the O ring was placed. When installed the O ring had a preloaded squeeze against the thin teflon at the bottom of the U which pressed against the shaft. It worked beautifully and the NCU would remain, drip free, in the silos for years.
11.1 Failure Modes
As we put the actuators and pumps through qualification testing they received very severe treatment and the initial designs failed. They were vibrated, dropped, overheated, overcooled, endurance operated – all designed to find their weakest link. When they failed they went back to the supplier and they did a failure analysis and recommended design changes at our next Technical Interface meeting. We had an almost perfect means of keeping track as the parts passed back and forth between the supplier and us. After we had accumulated several failures I started a Log in which I defined each failure mode, to see if after the design fix, it ever happened again. There were some 8-10 failures on each of the servo-actuators and the pump assemblies. Once a failure mode was fixed, that failure mode did not repeat itself. These became very very reliable, and once in the field never failed unless abused in some way.
11.2 Failure Reports Fail
Reliability was a stipulated part of our contract, and the Reliability Organization under Bill Yetter set up a Failure Reporting system. When ever there was a failure a failure report was to be written. Bill soon had many people coping with a flood of reports but could make no sense of it. There would be a failure report against the total Nozzle Control Unit, another on the electronics assembly, another on a amplifier, another on a transistor – they soon had more failure reports than they had failures. As a reporting system it had become a mess – though failures were being fixed. Bill discovered I had excellent records at a time he had to report to management on progress. He borrowed my documentation to show progress, and later revamped their tracking method.
11.3 Item Identification Documents – IIDs
It became apparent that the new semiconductor parts were very critical to reliability. A program was set up and endorsed by the AF to write Item Identification Documents for each and every electronics part and for the suppliers to set up separate “Minuteman” production lines, with documented procedures including certified personnel for making these parts. Large sums of money went into this program and it completely changed the way electronics suppliers manufactured their parts. This program paved the way for the rapid advances toward integrated circuits and to the microprocessors that followed. By the time of Missile X (Peacekeeper) standard commercial practice had absorbed these high quality methods into how they built all their products – the IID method was no longer required.
12.0 Stafford Leaves -- Jerry Pacassi takes over
Paris Stafford had to take medical leave due to heart problems. Jerry Pacassi from Systems Engineering took his place. Wayne Gates was transferred from aircraft operations and convinced Jerry we should have a second source for the pumps. Working with subcontracts he brought in another supplier who spent the money to build a stage I pump for test – it was almost a duplicate of the Vickers design – in part because of the way our specification was written and to be interchangeable. To this day I don’t know what Wayne didn’t like about Vickers. He’d been with the company longer than I had and was certainly an experienced engineer. Jerry accepted the idea that it never hurts to have competition. Vickers came to me several times wanting to know if it was perceived that they were not doing a good job. I said there were no complaints that I knew of – it was just that Gates had been given the green light to see if someone else would like to bid on the job. When it became obvious that purchasing was going out for bids again Vickers came to me and asked what I thought they should do. I said there is nothing that I could do to help them, that it was a competition. I said you are already making the units, you have your development costs behind you, the only thing I know that you can do is drop the price low enough so another bidder cannot afford to pay the start up costs to get in the game. Purchasing is going to select based on price only – so long as the other fellows passes Qualification tests. Vickers dropped their price and the new outfit Gates brought in lost their venture capital. I felt it was a waste of the new bidders time from the beginning.
12.0 Move to Anaheim
12.1 Map of Autonetics Anaheim
Fig 12-1 Autonetics Plant Anaheim CA
13.0 Minuteman I Electronics State of the Art
Minuteman I digital electronics were made of discrete parts, where semi conductors were being used to replace vacuum tubes. The following diagrams of a Flip-Flop and a dual NAND gate are examples of how digital functions were made. By Minuteman II these functions were put together on a single chip, the beginning of Integrated circuits. Examples of Integrated circuits will be shown in the next chapter NAA-MM2.
Fig 13-1 Minuteman I Flip Flop
Fig 13-2 Minuteman I Dual NAND logic gate
14.0 Heat Protection for Minuteman Flight Control Equipment
by Frank Lettang, Autonetics Div, North American Aviation
The following, written 05-13-97, is a Special Report written by Frank Lettang to accompany the above recollections. I knew he was doing this thermal work but had no knowledge of the details. Thanks to Franks excellent memory, he has recalled those events an shared his frustrations.
Some 37 years have gone by leaving of the only an impaired memory for dates and details.
The Dynasoar program had just been canceled (~’54) and I was assigned to the Minuteman program under Mal Johnson to perform servo loop stability analysis. Since this was a new field for a mechanical engineer with a thermodynamics background, I was busy taking classes at USC and boning up on the various servo analysis tools in use at Autonetics (Root Locus, etc).
Some of the fellow engineers were Roger DuPlessis, Ron Frazinni, Guy Bayle, Dick Olshausen, Pat Sanchez & Kurry Woo.
One day (~59) Dan Simkins came by and showed me preliminary stage I engine base area heat fluxes. Considering that my management had little heat transfer background, I told Simkins that this was “hot” in the literal sense. Dan turned around to Henderson and said “Frank, Frank says this is hot.”
Now I figured that I had about two weeks to come up with how hot hot was. So when asked that question two weeks later, the answer was that not far into the stage I flight time ¼ inch thick aluminum plates would melt.
Obviously the fat was now sizzling in the pan. Again it was easy to guess that the next question was going to be as to what to do about these heat fluxes. Getting a little more serious about the subject, I collected data on various insulation materials and any associated application methods.
True enough, Frank Henderson who was responsible for the system hardware and Flight Control electronics devises wanted to know what to do. The answer was apply ¼ inch thick PR1910 (a room temperature vulcanizing silicon rubber).
Henderson so far had no official heating criteria and no action was required.
Mal Johnson sent me to a full time two month training course at the Warren HS Downey (8 hours instruction with home work for the night) – I had lost 20 pounds while at “school”. We were just completing finals when I was yanked out of class to fly to Seattle with some VP (Bowman ?). I was to obtain the latest engine base area heating criteria.
Flight Control Heat Protection becomes Full Time Job
Initially the Minuteman I Flight Control heating criteria arrived far too late to affect the base hardware design. Nozzle Control Unit assemblies were committed to fabrication when it was learned how “hot” they would get.
For example Henderson called a meeting on the Edwards AFB silo launch hardware (tethered missile) He wanted to know what to do about heat protection. When asked how much time I had to come up with an answer, the response was “5 minutes”. So the answer was given again: apply by hand PR1910 to a minimum thickness of ¼ inch. I reasoned that a welding torch couldn’t burn through a ¼ inch rubber in 3 seconds. [the time to pop out of the silo].
Well the technicians Willy Mc Grahan and Chuck Todd had a few choice words about PR1910 after they got it smeared onto their coveralls -- and couldn’t remove it any more.
After a tethered launch, for which the NCU was instrumented, I was sent to Edwards for a postmortem. None of the temperatures budged from their initial values and the external rubber surfaces were slightly charred. The first inkling that the criteria makers had (for self protection) “cooked the books” and applied too much conservatism.
Fragmentation of Heat Protection Function Breeds Problems
TRW with SAMSO’s acquiescence) had organized the Minuteman thermal protection business to assure TRW the judge and jury roll. This approach already impacted all R&D flight hardware for Minuteman I.
AVCO corporation (Nose Cone) would develop a single material to be applied by all associate contractors. This material was an epoxy polyamide called AVCOAT (of course) an ablating material.
(a) carry out material testing and propose material design parameters (temperature of ablation, heat of ablation, conductivity, density, etc.)
(b) propose heating criteria for all Minuteman hardware.
(a) approve heating criteria and material design parameters and release them to all associate contractors.
(b) would closely monitor (supervise) all thermal protection activities of all associate contractors including the use of “official” computing tools.
Associate Contractors would apply the criteria and material design parameters using approved “ablation” programs to arrive at minimum AVCOAT material thickness for all exposed missile elements.
What are some of the Difficulties of this Thermal Management Approach?
(a) Timely receipt of design input data, especially on a fast moving program such as Minuteman.
(b) Overly conservative design input data results by divorcing the criteria makers (Boeing /TRW) from the hardware designers (Associate Contractors).
(c) An insidious “leaching out” of responsibility affecting component suppliers and Associate Contractor design personnel. In Minuteman heat protection activities the decision making authority appeared to be vested with TRW. Personal responsible involvement on the part of Associate Contractors personnel weakened in proportion to the loss in design authority.
Minuteman Flight Test Hardware
Given the inputs and requests (not technical direction) Autonetics applied AVCOAT to all rigid engine base area surfaces and the G&C section. The engine base portions of the missile cables were protected with SE55 (silicon rubber) extrusions (cable runs), and boots (connectors). The actuators and Z-links were covered with silicon rubber boots. Both the Missile Division Downey thermal and lab people as well as Autonetics material labs provided a lot of help.
Fabrication of flight hardware proceeded well. The AVCOAT could be readily cast onto the outside of the NCU “dog bones” and APS covers and the material adhered well and had good handling qualities.
Then around Easter (’61) a bolt of disaster struck clear out of the blue sky (or let us say during a cool spring night).
On Easter Monday, early Bill Blamer, the Flight Control Factory Manager, called up to report big trouble (in very laud and blue English). He seemed to also have included a few choice words about incompetent engineers.
It turned out that someone had left overnight a tarp covered stage III NCU on the dock of Bldg. 61 Fullerton. The unit was to be shipped to Plant 77 Ogden Utah to be assembled into a flight test missile. The outside night temperatures had dropped low enough causing the AVCOAT insulation to shrink itself off the NCU. It lay shattered on the floor of the dock.
Seven engineering organizations (TRW, Being, Autonetics, Aerojet, Hercules, Thiokol and AVCO) had all failed to consider the elongation to failure of AVCOT. This epoxy polyamide tends to polymerize such that it’s elongation to failure drops from about 2.7% to 0.7% in four to six months. Now the thermal coefficient of expansion (or contraction) is twice that of aluminum. So it shrinks itself off a part whenever the body temperature deviates from the fabrication temperature by as little as 30 deg F.
Obviously the scramble was on to rescue the flight test delivery schedule and above all find a solution.
It was small solace to Autonetics to learn that Thiokol had a similar experience with a stage I motor that was shipped in an unheated truck and exposed to Utah spring time night temperatures. About 400 to 500 square feet of AVCOAT were found in shattered pieces on the floor of the truck. At an all Associate Contractor meeting (of course chaired by TRW), Thiokol used a photo of the aft end of the truck to illustrate the problem.
At that meeting TRW still sought to save AVCOAT by introducing AVCOAT II touted to posses adequate flexibility. However, the Autonetics team led by John Esslinger, who was loaned from the Missile Division Downey to help us out of the crisis. John told the TRW people that Autonetics would solve it’s own problem by using NAA resources, that Autonetics will select, invent, and design and characterize it’s own set of materials and if TRW didn’t agree it better be prepared to issue specific technical direction to the contrary.
About a dozen materials survived an intensive gauntlet of testing and fabrication attempts and they were “admitted” by TRW into the material pantheon along with AVCOAT II and Armstrong Cork.
During this time TRW & Boeing began to push Armstrong Cork. Cork turns out to be thermally efficient insulator, however it comes in sheet stock. It is well suited for smooth surfaces (plates or large diameter cylinders), but it is a very poor choice for applications such as NCU’s and it is unsuitable for flexible objects and moving parts. The TRW thermal Tsar, Bob Hovey attempted unsuccessful by (over the phone) instructions to impede progress at Autonetics. He hinted that he would allow only very pessimistic heat of ablation values for the Antonetics championed materials.
The later Minuteman I flight hardware and all Minuteman I production hardware were insulated as follows:
NCU’s: RTV77 (white silicon rubber)
APS covers: Fiberglass
Actuator & Z-Links boots: compression molded boots
Cables (base area): SE555 extrusions
An unsuccessful attempt was made to replace the RTV77 with sturdier compression molded rubbers. Autonetics and industry (three rubber companies) could not solve the adhesive problem in time for design release.
Stage I Heating Criteria Flap
Around the latter part of ’61 or early ’62 a flight test missile was instrumented with calorimeters in the stage I engine base area. Unfortunately the stage I failed during the first half of the stage I flight. The calorimeter recorded somewhat higher than baseline fluxes which were due to insulation burning at lower altitudes.
A delta heat flux was added to the Stage I criteria which translated to an additional 0.020 inch thick layer of RTV77 than the minimum applied. The fabrication tolerance was about +_ 0.020 inches.
So Autonetics received a CCN (Contract Change Notice) around the summer of ’62. Autonectics answer was that no change was required (technical reasons), however of the customer insisted the impact would be a mandatory recycle of all Wing I missiles, and schedule and cost impact on Wing II, etc.
TRW’s position was that the added twenty mil should be applied in the silo by technicians lowered down to stage I in a “bosons” chair. The loss of configuration control was dismissed as an unimportant “paper” objection.
So Autonetics was handed a couple more CCN’s insisting that we re-insulate. Autonetics answer was adamantly the same except the cost impact kept growing, as is usually the case when other departments (logistics etc) weigh in.
By now Flight Control management was fully aware of the issues and united in the resolve to prevent the recycle of Wing I. As a result Ray Curci, Frank Henderson, and Frank Lettang paid a visit to Lungren, General Manger of the Minuteman Division. Lungren, a former AF Officer, quickly agreed to Flight Controls objectives and authorized two operational hardware stage I NCU’s off the line for Lettang’s use to prove that the mandatory recycle can be avoided.
This hardware was to be used to prove that the NCU allowable temperatures were well above the 400 deg F baseline. Ignitron lamps provided enough radiant flux to drive many parts of the NCU from room temperature first to 400 deg F, then 450 deg F, 500m 550, 600, 650, 700 and finally to 750 deg F. All insulation had been removed and the outside surfaces were blackened to better absorb the radiant heat pulse and the NCU was operated during the exposure. At the 750 deg F level the hydraulic cylinder started squirting oil into the vacuum chamber.
The Downfall of TRW’s Thermal Office
At a fateful meeting with Capt. Motley, AF resent, Gene Flowers, MM div, Flight Control Project Engineer, very ably presented the Autonetics data. TRW’s Leroy Herald wouldn’t gracefully accept defeat so he asked Lettang what he would do if the NCU worked at 450 deg F, (50 deg above base line) Lettang’s answer “nothing”. The dialog went on as follows:
Capt Motley closed the charade, wagging his index finger at the TRW crew, told them not to repeat this kind of performance again.
From Jan ’63 until Nov ’63 there was no single call from TRW thermal types to Lettang. In November Jack Tillotson, TRW Project office, wanted to know if Lettang would review a proposed thermal criteria change prior to release to the Associate Contractors.
Winning a “victory” was clearly harmful to the Minuteman program, since it damaged existing working relationships. The new thermal types at TRW avoided Lettang. Besides the thermal management office had lost it’s political power.
Part of the flight instrumentation was allocated to measure stage I engine base area heat fluxes and on at least one flight to measure several NCU body temperatures in places where the RTV77 was thinned down to the minimum fab thickness. That flight test was successful. Pete Fogle came closest with his guess of the max temperature reading of 264 deg F, and he won the bet.
Now if the in-flight flux levels had really been what the spec called for, the NCU temperature should have exceeded 400 deg F.
15.0 The People
About a year later Gates left the company to spend full time running his book binding business.
During Minuteman II Jim Anderson took over the unit, Jerry Pacassi went to work for the LA division and I went to work for Bill Strobel under Frank Henderson. Bill Strobel died of a heart attach at the beginning of Minuteman III. Jerry Pacassi died of a heart attack about three years later.
Art Greer went to work on Autonetics Radar operations and some three years later left the company, and nobody has heard from him.
Paris Stafford, a wonderful fellow liked by all, died of a heart attack about 8 years later, he never returned to work.
Fred Frankel went to work on an Autonetics F-111 electronics program then left the company with no contact since.
Most of the others continued with the company and were part of Minuteman II.
Of those mentioned in this chapter Jim Anderson, Frank Lettang, Fritz Gardner, Curry Woo, and Darrell Landau still meet at the “Third Thursday Lunch” of retired Flight Control people.
 The first missiles did not use shielded cables.