HD1973DBDC

Digital Electronics Concept:

One digital signal cable for all stages

Pitch Yaw digital servo loop closer on each stage.

Prior Electronics Education, Navaho, MMI, MMII, MMIII

New dual inline TTL chip sets

Serial Data Bus Demonstrator – For Flat Cable:

TTL & CMOS Logic Devices:

Home Projects:

1970 Patent Disclosure -- Electronic Commutation of a Motors Magnetic Poles:

Learning Logic Symbols and Truth Tables:

TTL Takes the Digital World by Storm:

Use of TTL Chips to Control Servo Actuators:

Digital Up/Down Counter Summing Junction:

Parallel Load, Unload Shift Register:

Signal Latch Holding Register:

Voltage Controlled Multi Vibrator

Pitch Yaw Command Generator:

Valve Command from Pulse Width Shift Register:

Converting Binary to Pulse Width:

A Simple Way to Make a Clock -- no radio oscillators or crystal clocks:

Counters and Decoders:

Analog to Digital Feedback Converter

Patent Disclosure for Variable Frequency Position Feedback:

Successive Approximation Register:

Electronics Rack Missile:

Up-Stage Simulator:

Down-Stage Simulator:

Demonstrations

Concept Review Meeting:

Digital Electronics Concept: I had in mind what we needed, but didn’t know how to go about doing it. I knew one digital signal cable for all stages would do the job, I knew we would need Pitch & Yaw digital servo loop closer on each stage. I decided I should first set about demonstrating we could control down stage flight control servos by use of a single digital using two way serial data. I was confident this would work in a manner similar to that used for rural telephone systems where each party responded to their ring. When I began I only had a vague idea on how to digitally close a servo loop. But I believed I could convert analog signals to frequency counts and could use an up-down counter as a summing junction to compare command with feed back, one up or down counting the other to leaving the error – which could be converted to a valve command. I proceeded on blind faith that it could be done.

I initially called this a “part line” control system, similar to rural single wire “party line” phones.

Prior Electronics Experimentation: I had been striving to learn more about electronics ever since I went to work for NAA / Autonetics.

Navaho period: Ever since I became part of NAA/Autonetics I had rubbed elbows with the electronics world. Initially it had been of necessity while operating a Navaho Extreme temperature test lab where it was necessary ot instrument and operate systems remotely. From military service I was accustomed to many kinds of electrical systems, but the test lab world entered a new domain. Thanks to Bob Kelley, a ham operator and experienced electronics test engineer – I learned much not available in books. At that time I bought electronic kits from Knight and Heathkit, building my own signal generators, oscilloscopes, vacuum tube voltmeters etc plus a Heathkit sterio hi-fi that used vacuum tubes. However I learned most when I built my own sterio hi fi amplifier using all transistors and no transformers – all were early version transistors – before these kinds of kits were on the market. That sterio amplifier is now in the Oberlin KS Museum.

MM1 & MM2 period: During this phase I dabbled with electronics but focused mostly on making cabinets and setting up a music system including a “real” type tape recorder. This was before cassette tapes. It was also set up with high quality speakers and phono deck for playing then new LP (long play) “flex” records while retaining backward compatibility with the old 78 rpm “hard” records. I also experimented with making my own radio’s, from simple sets to full superhetrodyne systems using transistors but modeled after vacuum tube carrier frequency transformer signal coupling. I bought a set of “Miller” coils, (oscillator plus three IF coils) only I couldn‘t get it to work. I set it aside – not wanting to give up on it.

MM3 period: For the first two years there was no time for dabbling with electronics, then as an escape I got out the radio that didn’t work and decided to give it another try. This time I tediously took the coils out of their cans and did a continuity check on each coil against the wiring diagram on the can – that was it!!! The wiring and diagram did not match! Within a few evenings I had the thing rebuilt and it worked. I felt relieved as I didn’t want to let something like that defeat me. This again proved that tenacity is an essential ingredient when experimenting with something new.

New dual inline TTL chip sets: While reading trade magazines I discovered that semiconductor manufacturers were advertising new logic devices based on then new TTL (Transistor to Transistor Logic). I sent for suppliers catalogs to find out what this was about. Our MM III program had been based on new advanced integrated circuits and the government funded extensive reliability study contracts. This was a huge stimulus to the semiconductor industry – making it possible to make better and more reliable integrated circuits. What was learned on military contracts was put to use in commercial products. In time the commercial products were just as reliable as our special Minuteman Proven Parts, however we could only use those qualified for MM. The AF provided special funding to semiconductor suppliers to maintain production lines and personnel to make the “qualified” MM parts – in the event more would be needed.

Serial Data Bus Demonstrator – For Flat Cable: I decided to build a serial data path which could be use for each stage by giving each stage a code. I worked out a message format in which the first part was the code that identified the stage address. Following it was the Pitch & Yaw servo commands , and discrete roll control & ordnance commands. I allowed 8 bits for address, 16 bits each for Pitch, Yaw, Roll and 8 bits for discrete signals. I set up a clock which ran upstage and down stage functions. For the first cut I wanted to keep it simple. I built the system into an electronic rack with the upper part “up stage” and the bottom part “down stage”. I used four 4 bit parallel load, parallel unload shift registers in series to move data in 16 bit packets. With my clock-counter I could move data down stage and return data back. It require five small wire data bus wires: two data, one clock, one +5 volt and one signal ground. This was the easy part, but to make a credible data bus, and how to you command and control something?

TTL & CMOS Logic Devices: I was constantly reading magazines and sending off for information. What follows cuts the chase and leaps from TTL to CMOS. The design began using TTL parts and migrated to CMOS. The first Serial Data Bus demonstrator used all TTL parts.

Upper left compares TTL and CMOS devices. Upper right shows how to make a NOR gate logic gate. Zoom view to display family of logic symbols on left. These were placed in chip sets of four as shown in center. Mid right shows D Flip Flop and JK Flip Flop with Truth Tables. Lower left shows a 4 bit shift register made from D Flip Flops. Lower right shows how to make a square wave clock generator.

Home Projects: I used my electronic work bench at home to making digital devices. One of the first devices was a transistor tester and a flip flop. Previous work with analog electronics was very helpful. I used formica circuit boards and applied signals with my signal generator and look at outputs on the Oscilliscope or VTVM (vacuum tube voltmeter). Almost every night at home I’d read and read and built parts, slowly becoming acquainted and comfortable with what I was doing.

1970 Patent Disclosure -- Electronic Commutation of a Motors Magnetic Poles: While writing this I came across a 1970 Idea submittal that I worked on in my home hobby shop. It was well known that a squirrel cage motor speed locks on to the frequency of the power signal. If a user had the means of changing that frequency they could cause the motor to generate maximum power at a frequency the user selects. Heavy Duty Transistors permitted commutating the motor pole pieces at any frequency determined by the operator. I had visited about this idea with fellows at lunch. Everyone knew such a thing would work, but there had never been the means to switch heavy duty power before the availability of power transistors. I decided to make a model in order to prove the switching electronics. I had submitted the idea and was building the power switch modules for my home set up when I happened to look at a recent Controls Magazine. There it was, the very thing I was trying to do – the magazine article presented the idea – indicating work was already underway in the industry. I dropped my effort and moved on. Years later I read where Union Pacific has been deriving greatly improved efficiencies due to their new Electric Engines which permit adjusting the motor pole frequency to the rate of motion – thus getting maximum power, at any speed, without loosing wheel to rail traction.

Learning Logic Symbols and Truth Tables: I spent hours studying and experimenting with transistors, diodes, flip flops and logic devices. IBM and other computer makers manufactures made logic devices with discrete parts on plug in circuit boards. Semi conductor manufactures were making rapid progress combining more and more devices on a single chip. I just happened to come off the propulsion program in need of a new job at the time these new goodies were becoming available. The technology was rapidly changing, I was memorizing terms and truth tables and sending for sample parts. The main devices were AND, OR and EXCLUSIVE-OR and the inverse of these the NAND, NOR and EXCLUSIVENOR; where N means Not AND. A Line above an identifying letter means NOT. Outputs were usually designated a Q rather than O to save confusion. A line above a Q Ō indicates the inverse of Q, or Not Q. IBM cards were using resistor or diode couple discrete transistors to make logic elements and chip makers were integrating these in new chip families.

TTL Takes the Digital World by Storm: The Transistor to Transistor Logic took off driving all other devices out of the market. These were packaged in dual inline 16, then 18 then 24 pin standard sizes. A designer could look these up in manuals and put systems together as if arranging dominos. Without fully being aware, I had entered on the ground floor with an application eagerly ready and waiting that could make use of the new devices. I just couldn’t learn fast enough – and didn’t have the funds to move out fast – so I steadily plugged away at it.

Upper left shows a CMOS inverter, upper center a CMOS transmission Gate which can block or pass analog signals (will flow either direction), upper right an analog 1 of 4 signal selector. Center is a square wave clock generator. Mid left is a 4 bit parallel preload, up/down counter. Lower left is a binary to discrete 1 of 16 selector, to it’s right the inner logic of a D flip flop, and bottom right a 7 segment display selector.

Use of TTL Chips to Control Servo Actuators: It was one thing to communicate with my “party line” signal system and it was quite another to control servos. I would need a digital equivalent of an analog “summing junction” a forward loop amplifier equivalent and a feedback equivalent. My first focus was on the summing junction.

Digital Up/Down Counter Summing Junction: On the way to work the odometer of my car began acting up, a bad tachometer cable, but It gave me an idea. Digital counters were like an odometer, they could count up and count down. I could start with a command, then count down with feedback and what was left would be the error command to issue to the control valve. I could shift a command down the cable, strobe it’s command content into a latch register, from which it could be loaded in parallel into a counter. I could set up an iterative cycle during which the feedback would count down the number and the error stored in a valve command register. A digital counter was the summing junction – I could control extend or retract with a sign bit stored separately – to cause the counter to count up or down. I found there was a new parallel load up/down counter that would be great for that.

Parallel Load, Unload Shift Register: I also needed a parallel load and parallel unload shift register to send signals down the cable and response signals back. I found there was such a device made from D Flip Flops.

Voltage to frequency Set able Clock Generator

Signal Latch Holding Register: I looked for and found a register they called a latch which would serve as an unloading/loading dock for shifted data.

Voltage Controlled Multivibrator: I needed a way of converting the feedback voltage to a binary number. More searching revealed a voltage controlled multi-vibrator. I found these came in pairs, two per chip – one for pitch and one for yaw.

Pitch Yaw Command Generator: I could also use a pair of VCMs to convert pitch/yaw motions of a joy stick to binary commands. I could generate these upstage and ship them downstage. I could simulate flight control commands via the joy stick by hand, but how was I going to convert a binary error signal in the summing junction counter to a valve command. I needed to ramp small signals to a full on command. If there was a way to convert the smaller binary bits to a % on time – then I could cycle the % on time, sustain it and let the valve coil integrate the pulse width to a sustained analog output.

Valve Command from Pulse Width Shift Register: I could store % on time as a pulse width in a shift register and let it cycle at relatively high frequency until the next upgrade iteration. I could cycle the pulse width at 6 mhz, the same frequency as used for the feedback excitation. I decided to use a 16 bit shift register as providing enough command resolution.

Converting Binary to Pulse Width: It took several days before I could come up with a way to convert a binary 15 bits to a stack of 1 to 16 bits. Later I found I could us a look up table set in a memory device – but originally there were no memory devices available and I didn’t have the means to program a read only device – which were just then showing up on the market. The Micro-Electronics Division was just then coming out with a 1024 bit silicon on sapphire ROM device.

A Simple Way to Make a Clock -- no radio oscillators or crystal clocks: I was at first thrown off by looking up how radio oscillators were made. I then looked into the use of crystals. I hit pay dirt when I found it was very simple to use a pair of logic devices connected with resistor/capacitor to generate a steady clean square wave. The technician helping me was soon able to select an appropriate R/C combination to generate just what we needed.

Counters and Decoders: I found there were counter chips which could be cascaded to as many as needed. I found there were 1 of 4, 1 of 8 and 1 of 16 decoders – the building blocks were there to do whatever I need to do – once I had a high speed clock I could down count, subdivide, enable/disable/strobe actions as required.

Analog to Digital Feedback Converter: How to handle the feedback signal was a real challenge from the beginning. First I had to provide 6 khz excitation, then demodulate the AC in to DC out, then convert the DC to binary, including the sign bit, and to scale the signal. I used the 6 khz to drive a four transistor set as a demodulator. Feedback signals are so low that you cannot rectify by use of diodes because diodes have a threshold voltage drop that blocks off critically needed signal. I managed that with discrete parts – driven full on devices do not have a threshold voltage drop. Demodulators used on the Navaho program required four peanut sized vacuum tubes, as full wave rectifiers, and consumed an entire 1”x3”x5” plug in module. During the Navaho it took some time to understand the need for a demodulator and how it worked, with new transistors it was no problem.

Later during the Minuteman program I had visited with Gary Collins who invented the position transducers we were using about the idea of making a DC transducer by placing a transistor demod in the units. I found he was already working on the idea and later came out with a DC transducer as a product line.

Patent Disclosure for Variable Frequency Position Feedback: During the search for a better method I experimented with the idea of replacing the position transducer with a variable inductor – tunable with a core – that could converted to frequency as a function of position just as a capacitor could change the output of the kind of clock generator I was using. I abandon the idea because there were no useable variable inductors on the market. Later at a staff meeting George Anderson, my boss who had just returned from a Group Leaders meeting said the company had not been submitting enough patents – and wondered if anyone had something they could submit. I had been daydreaming on my immediate problem, then became aware everyone was looking at me! As they were earning the money I was consuming with my experimentation, I felt obligated to do something. I said I’ve had a number of ideas …. Immediately Frank Lettang spoke saying you don’t have to prove they work by building a model – some of those ideas you’ve told me about should qualify for a patent submittal. Thus I reluctantly submitted the inductive transducer idea for patent. The AF decided to accept it, assigned an patent attorney to work it up and some 10 yrs later a patent was issued and I was awarded a check for $1000 for the effort – all of us had been required to sign an agreement that the government not us owned any patents. I had recalled a 1930’s model Buick radio I’d taken apart which used what looked like inductor tuning slugs rather than variable capacitors to select radio signals. Thus I assumed such a device could be made – since I couldn’t buy such a thing for what I was doing I had too much to do to spend any more time on it. As it was it required many hours to put together credible information to back the credibility of the idea – especially in follow up phone calls from the patent attorney in Washington.

Successive Approximation Register: I later solved this problem by use of a Successive Approximation Register device that recently came on the market. By this process the SAR outputs a half way guess binary number, which is applied to a set of binary scaled resistors which convert the binary to analog which is fed back for comparison with the unknown analog signal. The SAR keeps homing in until there is a match and the binary number accepted as valid. The advantage is that it is very fast, requiring 16 steps to achieve 16 bit accuracy.

I don’t recall for sure but I believe I used the Voltage Controlled Multivibrator chip for this early demonstrator.

Electronics Rack Missile: I needed a way to demonstrate the concept so I set up an electronics rack with Upstage at the top and downstage at the bottom. I got the loan of two PBPS electric servo actuators from Jim Anderson to operate as Pitch and Yaw servos. Two more were added to represent another stage. In my shop at home I made two aluminum frames on which we mounted punched circuit boards.

Up-Stage Simulator: I mounted two potentiometers to the top board where an aluminum tube handle moved one for pitch and one for yaw. This worked so well I made a second one to command the second pair of servos. Voltage Controlled Multivibrators converted the potentiometer positions to a frequency output which was directed in it’s turn to a counter. The binary output from the counter was strobed into a latch and from the latch in the output shift register. A clock and timing was set up to manage the data bus protocol sequence.

Down-Stage Simulator: The feedback was connected through VCMs to up/down count the command parallel loaded into the “summing junction” counter. The valve command was shift register cycled as a pulse width applied in accordance to the sign bit to extend or retract the servo.

Demonstrations: Emil Kohler had been assigned to help and we worked together trouble shooting the circuits. We used an oscilloscope to check the storing of signals in the devices. But we could not “see” the process at work other than watching the response of the servos to wiggling the joysticks at the top.

It was a thrilling feeling to be able to wiggle the joysticks and watch the servos follow with only a few tiny wires connecting top and bottom.

Our first visitor was Dale McLoud, who sent Walt Evans to look over what we had done. Evans had invented Root Locus method of servo loop analysis and later suffered a stroke -- he car pooled with George Keller during the Navaho program.

Our second visitor was Tom Shuler. Tom looked and listen dead pan. After explaining how it worked and what we were trying to prove I said, OK Tom stand here, wiggle the control sticks and watch the servo movement below. We were delighted to see Tom’s countenance change from dead pan to a big smile.

Concept Review Meeting: We had a concepts review meeting in bldg 73 attended by many of those working on concepts applicable to new business. This included Dale McLoud, Bob Niese, Mal Johnson, Dale Leisy, and others. When asked about how the system I had put together was to accommodate servo-loop shaping problems -- I made some irresponsible statements – which Mal Johnson never let me forget it. I’d said to McLoud we don’t need that old shaping stuff any more, digital doesn’t use that. Others thought I had said we didn’t need to perform shaping loop functions. I intended to say we didn’t need to use resistor capacitor operational amplifier shaping networks. In reality I had been so intent on determining how to do the serial data bus thing and to achieve digital control of servos, that I had totally ignored the shaping network problem. Our Minuteman servo actuators didn’t require shaping networks – however we soon became aware the new kinds of MX actuators anticipated by the Motor Contractors would.