Chapter 44, Interface Engineering 

1. Introduction.

As systems evolved, there was always a need for electronics and/or special equipment to interface with computers.

     The greatest need was for communications between devices within systems and between multiple systems.  Our engineers helped set the standards for the defense industry in many key areas.  Today's most common interface use is the internet, the history is documented in many places thus we do not try to summarize that history hereunder.  The UNIVAC/UNISYS perspective of networking is related in a John Nemanich article.

Thanks to Marc Shoquist, Dick Erdrich, and others for their contributions to this page.

2. Point to point by Lowell Benson, U of MN - BEE 1966

     The first Naval Tactical Data Systems interface was a 30-bit wide parallel communications path using data and control level signals of 0 volts and -15 volts. The computer program would open up either an output or an input buffer. When a peripheral device was ready to accept a word, it would turn on an output data request line. The computer would accept the output request, enter data into an output holding register, four microseconds later an 8 microsecond output acknowledge signal line was activated. The holding register information was held for another 3 microseconds - meaning that it took a total of 15 microseconds to transfer a word. The peripheral would drop its request line, then turn the line back on when it was ready for another word. If it was input to the computer, the peripheral would load a register, then activate the input request line. The computer would sample the data lines then send an 8 microsecond acknowledge pulse. The peripheral would then drop the request line, reload its register with the next word (or byte), and repeat the process. A similar method was/is used for control signals to the peripheral and status information back from the peripheral. This request/acknowledge process allowed reliable point to point communications up to 300 feet - a communication rate of 45,000 words per second. Obviously printers weren't this fast but inter-computer communications were. ERA documented this communications method into Design Specification DS-4772.  This Design Standard was the basis for MIL-STD-1397 that evolved to encompass both parallel and serial computer and peripheral interfaces.

{Editor's note: I recall Ernie Lantto relating that a JPL installation had had one communication link operating over several miles using this '-15 volt' method - from a top of a local mountain to their facility at the base of the mountain.}
As Germanium transistor use became prevalent instead of vacuum tubes, research found that using a -3 volt signal swing could reliably communicate over 200 feet at a faster rate. The register loading had a 2 microsecond advance time followed by a four microsecond acknowledge pulse. This went into DS-4772.
     Silicon transistors brought a new era to this interface method. The signal swings could be 0 volts to +4 volts [or so.] DS-4772 was updated, now labeling these three interfaces as Type A, Type B, and Type C. The Type C became the interface of choice for the airborne 1830 using a 2.5 microsecond acknowledge. This provided a data transfer rate of up to 167,000 words per second, i.e. a word every 6 microseconds. This was nicknamed the ANEW interface (yes that means 'a new' interface - ANEW is not an acronym!) I believe that Ernie Lantto was the engineer responsible for the accuracy of these revisions to DS-4772.
     The Navy Laboratories doing research had UNIVAC experiment with point to point serial interfaces using co-axial cable. This led to the Type D interface in DS-4772. A Type D interface was designed for the AN/UYK-7 computer, Ken Graber was one of the engineers. The Navy then transitioned the UNIVAC specification into a Navy standard, Mil-Std-1397, so that it would 'have control' over their interfaces. These three parallel interfaces and beginning of point to point serial led to development of a Low Level Serial and then Fiber Optic communications between computers and their peripherals.
     The first 30-bit Input Output Processor (IOP) developed for FAA applications took the ANEW interface a bit faster, using a two microsecond acknowledge pulse and less register setup time to effect point to point communications of almost 400,000 words per second. [lab] 

3. LOW LEVEL SERIAL

During the 1970’s a group under Marc Shoquist's direction developed engineering and demonstration models of the Low Level Serial Input/Output (LLSI/O) interface, the SHINPADS Network for the Canadian Patrol Frigate, and a Parallel to Serial I/O converters for the Combat Data Switching System at Mare Island, all of which became production systems.  For the Aegis system, building the ship with serial interfaces instead of parallel interfaces would save 10 tons of weight above the water line. [lab].   For the AN/UYK-7, the AN/UYK-20, and the RD-358 magnetic tape unit we engineered this interface on 3.5" x 3.5" normal sized cards for those units.   Ken Graber had been an instrumental design engineer for the earlier AN/UYK-7 point-to-point 10 Mhz MIL-STD-1397 Type D serial interface so was quite familiar with interface design disciplines.  As LLSI/O was being designed for the AN/UYK-44, special packaging was needed so Ken Graber transferred from program management back into engineering to lead the custom component design.  Ken, Larry Bolton, and Jeff Parker have written about the development, SNERT. (Serial Nato Electronic Receiver Transmitter)

4. FIBER OPTICS

by Marc Shoquist - U of M graduate, Electrical Engineering 1951

I joined ERA in 1953, (34 years with ERA/Sperry) then was Manger of Fiber Optic Systems (1973-1988)

          Fiber Optics at Univac Defense Systems in St. Paul was an out growth of the NAVSEA development of Low Level Serial I/O for Aegis Combat Systems as well as NAVSEA’s participation in NATO Naval Study Groups on Combat Data Systems. The Division was a charter member of the NATO NIAG Sub Group 6, which was organized in 1973 with a goal of development and standardization of a high speed serial I/O interface for NATO Navies Combat Systems. The NAVSEA’s effort was led by Captain Eric Swenson and Harvey Kloehn, NAVSEA Chief Engineer. The initial Univac Defense Systems member was Bill Geiger who served on the committee from 1973-75. NAVSEA sponsored the Development of Low Level Serial I/O, which was implemented in Aegis Ships and became a NATO standard in 1979. Development of a Fiber Optic medium interface standard for Low Level Serial I/O started in NIAG Sub Group 6 in 1976 and became a NATO standard in the early 80’s.
  Univac Defense Systems and Sperry Flight Systems (Phoenix) led the early development of fiber optic interconnect systems for Sperry Corporation. It started with a fiber optic technology interchange meeting set up initially between the two divisions, but soon grew to shift the focus to the review of 1974 project requests to the Sperry Rand Research Center (Sudbury) in the optics area. The meeting was held on February 8, 1973 at the Sperry Flight System facilities in Phoenix and in addition to the above divisions, representatives from Sperry Gyroscope (Long Island) and Univac-Roseville attended. It was a milestone meeting as it initiated semi-annual optics technology interchange/seminar meetings which grew to include all Sperry Corporation Divisions (military and commercial including New Holland) in which the meetings were hosted by the Divisions throughout the corporation.
        UNIVAC Defense Systems began IR&D sponsorship of fiber optic interconnect links in 1974 as it was seen as a medium option to the NAVSEA Low Level Serial I/O as well as for airborne interconnect systems.  The marketing focus was on the U.S. Navy development of Fiber Optics which was centered at the Naval Electronics Laboratory (NELC)  in San Diego led by  Don Williams, Head of the Fiber Optic Systems Branch.  Their initial priority was on the development of a first generation of low-cost, modular electro-optical transmitter and receiver components for bundled fiber point-to-point applications.  In 1975 Defense Systems was awarded NELC contracts for panel mounted modules, Figure 1 on the left,  followed by flat pack modules for circuit board cards, Figure 2 on the right.  NELC viewed the panel-mounted modules as a retrofit replacement for their electrical equivalent RG-8 or RG-58 coaxial connectors on the back panel of current equipment.   Manufacturing technology contracts on these components were received from the Naval Air Development Center (Warminster, PA) in 1976.  These modules were developed by John Kolling, then manager of the Hybrid Laboratory at Defense Systems and his expertise in both hybrids and circuit design was instrumental in establishing the Division’s reputation as a leading supplier of fiber optic interconnect systems for military applications.  
        The R & D marketing of fiber optics was done by myself and George Tkach (A P-3 Naval Reserve Flight Officer) from Defense marketing.  George established a close relationship with the NELC Fiber Optic staff and particularly LCDR Jim Ellis who provided liaison to NELC in an in-flight demonstration of fiber optics interconnect for the A-7  Navigation and Weapon Delivery System.  This project was sponsored by the Naval Air Systems Command to demonstrate the feasibility and effectiveness of fiber optics on an airborne military platform.   The fiber optic multiplexed system replaced 115 point-to- point channels of parallel wire cable with 13 simplex  channels and accumulated a total of 150 flight hours of failure free operation.   
        Our fiber optic modules were a key to winning contracts, so a fiber optic data link demonstrator, Figure 3, was developed to market the modules and the capability of fiber optics.  It consisted of a pair of transmitter and receiver modules and 100 feet of fiber optic cable in which a 10 Mb/s data link was turned on and off by a hand held push button switch.  The switch activated a digital counter which measured the time interval between the start and stop operation of the switch.   By successive operations of the switch, individuals could measure their response time in milliseconds. Our customers would like to test their reaction time on the system.  We would end the demonstration by disconnecting the fiber optic cable, moving the terminals closer together and aligning the transmitter and receiver ports to show the system could operate with no cable be at all.  Our demonstrations would frequently wind up with our customers rounding up their managerial staff including VP’s to come and see the demonstrator.   The demonstrator was an effective demonstration of the small size of 100 feet of fiber optic cable as compared to its electrical coaxial cable equivalent.  In addition to its low size and weight its other attributes included high data rates, longer transmission distances, immunity to interference and a non-radiating transmission system, all attractive to military applications, particularly airborne systems and  ground transportable links.  Our initial production contracts were won on those type systems. 
        The first multi-year production program for the modules was a 100 MB/s  transmitter/receiver card, Figure 4, for the AN/UPQ – 3A Airborne Data Multiplex system manufactured by the Defense Systems Division of Sperry Univac in Salt Lake City.  This was a fully militarized system in which the modules were required to meet an operating environment of – 55° C to 125 degrees C.   Sixty transmitter/receiver cards were required per system with distances up to 80 feet for the airborne system and 1000 feet for the ground system.   The initial system used fiber optic bundled cable but was later upgraded to small core single fiber cable.  Acceptance testing was completed for the system in 1977 with production beginning in 1978.  At that time it was the only production militarized airborne fiber optic system in the U.S. Air Force.   
        A benchmark development/production award was received in 1980 for the fiber-optic based Common Weapon Control System for the Ground Launch Cruise Missile (GLCM) sponsored by the Joint (Navy/Air Force) Cruise Missile Program Office.  It was one of the first internationally deployed weapon systems to be interconnected by fiber optics, which was specified because of its ability to survive in a hostile environment (including the effects of a nuclear event), its ease of deployment, its immunity to electromagnetic interference and absence of radiated emissions.  The GLCM system, Figures 5, consisted of two Launch Control Centers (LCC) and four Transporter Erector Launchers (TEL) interconnected by eight 300 meter cables each containing 3 full duplex digital communication channels.   Each cable carries multiplexed computer data, digitized voice, and discrete signals in digitized form between elements of the system.  The Signal Transfer System hardware is shown in the photograph of Figure 6 below left.  The development contract for the GLCM system was the largest fiber optic award of that time of about $3 million followed by multi-year production awards starting in 1981 exceeding a total of $ 40 million dollars.  Of interest is that one of the technical evaluators of the proposal was the NELC Fiber Optics staff from which we had received the initial $ 25,000 contract for development of the transmitter module. 
        Following this award, the group went on to develop transmitter/receiver modules for the SAE standard High Speed Data Bus (HSDB) which were low profile and suitable for mounting on standard computer I/0 cards.   In addition, an active fiber optic coupler was developed for the HSDB under contract to McDonald Douglas for use on the AV-8B , the proposed ATF-23 and other McDonald fighter  aircraft.  Jim Herrmann designed the coupler and was project engineer on the program which was awarded in 1988.  The active coupler extended the number of users beyond that achieved with passive couplers. 
        The Fiber Optics engineering staff at the time of the GLCM award besides myself was John Kolling, Jim Hermann, Bill Davis, Ernie Griffith, Ross Starkson,  Terry Thorvelson, and Al Osberg.   The Defense Marketer was Bill Sanda who received a “Salesmen of the Year” award for winning the contract against strong competition from the ITT-Roanoke Fiber Optics Division, Lockheed Electronics Co. of Plainfield, NJ, Lockheed Missile and Space Co. of Sunnyvale, CA and  International Telephone.  This was an advanced development group responsible for engineering and demonstration models which were passed on to design engineering for production systems.  Fiber Optics was just one of the activities of this group whose focus had been on communication components and interconnect systems. 
        Members of the group were also active participants on military and SAE standardization and working groups on Computer interfaces and Local Area Networks. [ms]

5. More about Fiber optics by anonymous

     The Marc Shoquist GLCM FO technology that was fielded in Germany and along with Reagan convinced the Russians that they we could outspend them leading to their downfall. This evolved into the CVN77 RFOF IRAD and ultimately evolved to Rick Stevens F16/JSF FO based Mission processor.  LMCO is now building fiber optic back panels for the F16 and several other Lockheed aircraft.

6. 1773: by Dick Erdrich

     Marc, your fiber optics article above is a very good read. I have fond memories of John Kolling - wish he could have counted his Mondays down to zero.
     I was aware of most of the projects that were mentioned, but was only connected with John for the 1773 effort because of my position as support engineer for the AN/AYK-14. I don't believe we ever did any production release of a fiber Discrete and Serial module - MIL-STD-1553B equivalent - but I thought that the design was completed and could have been released if we had gotten a turn on from the Navy.
     We were very fortunate to have worked for a company like UNIVAC during those years. Now that the division is controlled by an aircraft company the opportunity to do wide ranging development and research is gone unless it applies to their aircraft. There are limited chances to work on something interesting but I did get lucky. [rae]

7. SHINPADS contributed by Gene McCarthy

     The SHipboard INtegrated Processing And Display System (SHINPADS) Serial Data Bus is a high-speed digital interconnect system designed for real-time Naval Electronic Systems. The SHipboard INtegrated Processing And Display System is a concept that was originally conceived by the Canadian Navy. ®SHINPADS is a registered trademark of the Canadian Department of National Defence.
     The Serial Data Bus (SDB) is used to interconnect devices currently using interfaces such as Military Standard MIL-STD-1397 Input/Output Interfaces or NATO standard interfaces such as STANAG 4146 (parallel) and STANAG 4153 (serial).
     The SHINPADS Serial Data Bus was specified, designed and implemented by Sperry Univac Defense Systems. Sufficient growth margins are incorporated by properly distributing functions between software, firmware and hardware. The Serial Data Bus also provides for future evolution to Fiber Optic Trans mission without changing operational or system soft ware.
     The concept is based upon the need to achieve a high degree of survivability and operability in battle damage situations; system reliability through use of reliable components, redundancy, and system reconfiguration; low life-cycle cost by standardizing major system elements and by reducing the complexity of interfaces; and improved performance through the extensive use of digital computers, digital support equipment, and a distributed system architecture.
     The architecture of a system is largely influenced by the techniques used to interconnect the equipment used in the system. Interconnection techniques using MIL STD-1397 interfaces with various switching combinations, duplexers, multiplexers, etc., have enabled the development of a variety of system structures. The SHINPADS system has a distributed architecture while other examples of military system architectures may be federated, centralized or any combination of these architectures. 
     The simplicity of current standardized hardware interfaces has provided system designers with a flexibility enabling the structuring of required systems. However, the expansion of digital technology and the new system requirements, such as those defined for SHINPADS, have created the need for an efficient data bus system interconnect which allows complete intercommunications between computers without great numbers of cables and switches. 
     Right, Richard L. Seaberg, Defense Systems vice president and general manager, emphasizes Sperry Univac commitment to Canadian military programs in an address to a group of 36 Canadian customer representatives at Univac Park.
     Below, Dick Olson explains operation of SHINPADS® serial data bus access modules to visiting Canadian Department of Defence personnel in an equipment demonstration following Mr. Seaberg’s remarks. {Editor's Note: Gene McCarthy is circled on this snapshot.} More than 200 customers have visited Defense Systems during the past few months to view SHINPADS (Shipboard Integrated Processing and Display System) demonstrations conducted in a specially prepared demo room. The demo room contained four militarized computers with a fifth system computer, an AN/UYK-502, operated in a remote site. More Than 40 Groups from Nine Countries have viewed the SHINPADS Demonstration.
     Many functions formerly centralized in computers have been distributed throughout the subsystems. The interconnection of subsystem elements by a data bus network provides for easier reconfiguration, increases capacity for functional and physical expansion, and provides ability for more direct intercommunication between any combinations of subsystem elements.
With these combined features, the potential for fallback and survivability in the event of loss of one or more subsystems is very high.
     Conventional interconnection of military systems via point-to-point wiring has many problems. Because of the sheer mass of cabling involved one cannot afford to provide redundant hardware paths between all platform subsystems. Therefore, the platform is vulnerable to outages caused by failure of these dedicated cable paths.
The design and customer use of SHINPADS Serial Data Bus incorporates role simulation, modeling and system protocol levels, the relationship of software architecture to previous software architectures and reconfiguration policy alternatives.
     The SHINPADS concept is based on the extensive use of standard digital computers widely distributed throughout a combat ship and connected by a data bus. Theoretically, the data bus allows the digital computers to serve as global resources capable of taking on any function because of complete interconnection. The use of the data bus in SHINPADS also provides a practical means for redundant physically separated interconnections to improve system survivability.
Sharing a communication channel, such as a data bus, requires a blending of competing demands for service to satisfy operational requirements. The designers of a bus interconnect system must consider and ask questions such as: “What is the average data transmission rate over a given line?” ‘‘What is the peak data transmission expected during any given time interval?” “How long will any user have to wait to use the line?’’ “How will time critical functions be granted access to the line? The systems designer must also consider the question: “How can information be protected and controlled in the distributed network?” The answers determine important parameters in the bus interconnect system design. The SHINPADS Serial Data Bus design team considered these questions and many more to identify the important parameters for a naval combat system bus interconnect system. The SHINPADS itself defined certain performance requirements which by their nature required examination of naval combat systems in general.
     Key SHINPADS Serial Data Bus features which distinguish the Sperry Univac Defense Systems data bus from other buses include:

• Data Throughput
• Access Time 
• Transmission Distance 
• Number of Users Fault Tolerance 
• Broadcast and Point-to-Point Addressing 
• Bus Control
• Compatibility 

     Individual communication requirements for a system such as SHINPADS were examined and a real throughput requirement of 2.2 million bits per second (MBS) or less was determined. Although 2.2 MBS is a SHINPADS peak condition, the overall real throughput of the bus was set at over 8 M BS to provide fast access time and room for growth. Serial data busses, such as MIL-STD-1553B or Shipboard Data Multiplex System (SDMS), have single channel real throughput capabilities that are significantly less than 8 MBS, and higher rates can be achieved only by duplicating channels.
     Time critical access requirements were defined in terms of critical message transfer rates. The extensive use of front end computers for dedicated time critical functions such as signal processing tends to reduce overall access time requirements, although access times less than 500 microseconds are still needed for some systems. The SHINPADS Serial Data Bus guarantees specified access times.
     The maximum length of the bus was specified to be 300 meters. Three hundred meters was selected because it is sufficient for shipboard applications. Point-to-point interface connections to the SDB are made by stub cables of up to 30 meters. The SHINPADS system required in excess of 90 bus connections. However, the maximum number of connections specified for the SHINPADS bus is 256 to provide room for growth. Bus systems, such as MIL-STD-1553B, SDMS, and others, provide a user addressing capability ranging from less than, to significantly more than 256. The number selected is a “trade-off” between the amount of performance required and our adequate number of user addresses. Excessive addressing capability increases the overhead factor which limits throughput.
     The shipboard environment requires survivability of system functions, not only for single failures but for multiple failures. The bus system contains up to six redundant transmission cables which can be rapidly utilized in case of problems.
     The SDB is capable of interconnecting existing sub systems with minimum modifications to the sub systems. Existing subsystems covered by this requirement are characterized by the extensive use of MIL-STD-1397 interfaces and the characteristic of exchanging information through the use of variable length messages. A significant feature of MIL-STD-1397 interfaces is that the interface protocol is simple and only controls the transmission and receipt of information and does not operate on the content of the data word.
     Fundamental to the SHINPADS concept is the use of standardized interfaces. The use of standardized interfaces in digital systems has been a goal of Sperry Univac Defense Systems since the development of the U.S. Navy Tactical Data System in the 1950’s. Standard interfaces must have special qualities to have a useful long life since many of the peripherals interconnected in the future will not have been designed when the standard interface is specified.
The SHINPADS Serial Data Bus design was based upon an understanding of the interconnect system philosophy which forms the foundation for MIL-STD-1397 interface options. This is important because MIL-STD-1397 is currently being used on most, if not all, Navy equipment developments. However, the bus was also specified with an eye to the future. A large growth margin was designed into the system to allow for future considerations. The implication is that distributed system architectures are not only possible, but are now becoming practical for applications requiring a network of small interconnected computers.
     As technology advances, higher data transmission rates will be possible. Therefore, the SHINPADS protocol was designed so high-speed transmission systems can be used without changing the protocol of user system software. For example, when fiber optic technologies become cost effective, the SHINPADS SDB can be upgraded.
[efm] 

8. Communications Systems by Larry Debelak

     Ole: Here are some links to briefs that show the history since the NTDS to 1997 Chart you slipped under my door--What is missing are the Non-Navy systems and the classified systems to include the antenna couplers [e.g. Air Force SMA], Project Spring [Salt Lake], Israel Project 6977 [Eagan], the Salt Lake DOD [Micro wave TDL] and the commercial Univac Switch product story [Salt Lake]---Bruce Olson or Marc Shoquist should be able to fill these in----and Harry Wise should have the Eagan Remote Control Communications switch he put in unmanned Navy communication sites in the Southeast that were the fore runner of the MATCALS/ABCCC switch that Joe Pobiel & Marc Shoquist have the book on.

     Also Lauren Cady worked on the Home grown Automated Voice Telephone Switch system for the Eagan facility.
This along with the ABCCC JTIDS, TMRC/ASOC Data links and the SOS Po Sheng JTIDS/FO based network fill out your chart and could provide quite an impressive legacy for the Rick Udicious newly created Communications and Networks Business Unit. Now we just have to fill the vision with a win on AMF JTRS, evolve the LCS CRR success to other Surface ships and leverage the Owego US101 COSITE/UHF SATCOM subsystem to airborne platforms--Then the Univac to LM Communications and Networks Legacy story will be complete and I can retire. thanks.
     We also used Harry's/Joe's/Jack Schaubert's switch, Joe's/Dr. Don Kleven's Digital Voice SW with silence detection and Steve Andersen's SHINPADS/FDDI/SAFENET/ATM/InterPhase experience that along with ABCCC Automated Communications Selector GUI's was the foundation for the NSSN CRAD effort with NUSC Newport/Groton [Now NUWC Newport] & PEO SUB. From this success [Bob Siegfried was the Network CRAD lead that simultaneously hosted the Q-70 Combat Control/Sonar & Voice over IP with 5 MS latency], I got the first PEO SUB $3M Contract for COTS Radio room prototype that became the EB Virginia CSRR, LCS and the Current PMW 770 CSRR contracts for all Sub Classes with an unprecedented 3 years of Congressional Plus Ups. [LF Debelak]