Interview: Erv Gangl

Mil-Std-1553, the 30-year-old bus standard that gave rise to "plug and play" digital avionics, is taken for granted today. But its advent was a major milestone. Before the standard, avionics was largely analog, and signals were transmitted over dedicated wires. Aircraft harnesses featured thousands of point-to-point connections. It’s hard to imagine how complex and tricky avionics data communication was at that time.

Erwin (Erv) Gangl appreciated the complexity and made a major contribution to remedy it. In an interview with Avionics Magazine, Gangl, whom many consider to be the father of Mil-Std-1553, explains how revolutionary the idea seemed at the time and how hard it was to convince early customers of its benefits. It took enormous persistence to achieve acceptance by the three services, leading to the first tri-service version of the standard, Mil-Std-1553A.

Erv worked for the U.S. Air Force from 1964 to 1988, retiring as chief avionics engineer, strategic systems, Wright-Patterson AFB, Ohio. He now serves as the director and manager of the Dayton, Ohio, office of CACI Technologies Inc., a unit of the information technology and network solutions firm, CACI International.

Avionics Magazine: How did you come up with the concept that eventually led to Mil-Std-1553?

Gangl: I was working on an F-111 avionics upgrade in the late ’60s. The F-111 had digital computers, but the avionics subsystems were analog. There was a huge A/D [analog-to-digital converter] box that brought in analog signals from the sensors all over the aircraft, sampled them, converted them to digital and passed them on to the computers. The upgrade modifications added hardware, which required a significant A/D converter redesign to handle the extra analog signals. I felt there had to be a better way of integrating avionics and making the design more modular for ease of updating.

I recalled programming IBM mainframes in college. Peripheral devices were linked to the mainframe computer on a digital bus, and devices were selected by using addressing techniques. I asked myself, why wasn’t this possible in an avionics system? It would require putting the A/D converters at the signal source, in the radar and navigation black boxes, and interconnecting them via a digital parallel data bus, using addressing to discriminate between them. At that time my ideas were still too immature to be credible and therefore were not taken seriously.

Avionics Magazine: How did the idea finally catch on?

Gangl: After leaving the F-111 office in the late 1960s, I was assigned to the F-15 program office. I was the engineer responsible for the central computer and the A/D converters. As a young engineer, I approached the lead avionic engineer and told him I would like to try using a digital data bus for integrating the avionics. But he brushed me off because the concept was viewed as high-risk and unproven. They wanted to stay with tried-and-true analog integration, what contractors had always bid before.

When the initial F-15 design was found to exceed the required gross takeoff weight, hardware and functionality were trimmed as much as possible. But there was still a need to reduce the weight a couple hundred pounds more. So the lead engineer grabbed me and asked whether my idea of time-sharing wires would reduce the weight in cabling. I said, "definitely!" There would be weight savings due to fewer wires, fewer connectors, etc.

So the government gave the three program bidders the opportunity to come up with proposals and demonstrations based on the concept of a 1 MHz, multiplexed bus that I described to them. All three companies built some proof-of-concept demo hardware. One couldn’t get it to work, one got it to work but it was cumbersome, and one [the former McDonnell Douglas] took a simple approach and made it work. Luckily for me, they won the F-15 contract and we were on our way. The designation of McDonnell Douglas’ multiplex data bus spec was H-009. The F-15 design thereafter became known as the "HOO-9" bus. The F-15 was the first application of my digital bus idea and the forerunner of 1553. [It was also the first implementation of a multiplexed digital avionics data bus.]

So the F-15 SPO [system program office] accepted the bus concept because it saved weight, which was like taking a pill for its side effects [laughing]. It actually added architectural flexibility, ease of reconfiguration and future updates, provided digital built-in test capability and increased reliability. At the time, one of the biggest reliability problems was the avionic connectors. With analog avionics, these could be huge, sometimes with hundreds of pins. Every analog signal had to be connected to the central computer through the A/D converter with a separate wire, which then naturally required a separate connector pin.

Avionics Magazine: So the F-15 bus transitioned into Mil-Std-1553?

Gangl: The F-15 implementation was the first trial of the concept. I wasn’t happy with the initial hardware implementation, which used two twisted pairs [clock and data], where the signals on the data line were summed with the clock line to recover the digital signal. Problems occurred, for example, if one twisted pair was shorter than the other; the transmitted signals wouldn’t arrive at the same time at the receiver, causing waveform skewing and difficulty in recovering the signal. So you got a lot of discrepancies, which in turn required really tight control and heavy shielding in harnesses. All this went away when we went to the Manchester waveform because of the single transmission line and the embedded clock.

Avionics Magazine: How did you decide to use a serial time-division Manchester encoded data bus?

Gangl: My first inclination was a parallel bus. But that would not have been an optimal solution. The question was, if you go to a serial bus, how to do that? I was stuck.

One day, as I was driving to work, the newscaster reported on the Apollo space program, describing how [NASA personnel] were monitoring the astronauts’ vital signs while they orbited in space. The data was being beamed down to the control center on a time-shared single frequency. The telemetry system was using Manchester bi-phased encoding, as described in Mil-Std-442. So instead of inventing something new, I thought, why not adopt that for serial transmission within a twisted pair wire bus?

Avionics Magazine: The technology was ahead of its time?

Gangl: At the time we came up with the concept, we had trouble getting manufacturers to build the chip sets that could encode and decode the algorithms fast enough. When I said I wanted to transmit at 1 MHz, they said they didn’t know if they could build the encoder/decoder logic chips that would run that fast. The state of the art at that time was around 500 KHz. And the editors of technical journals rejected my early papers because the concept was so new and not widely understood. Later on, we had to have government-sponsored Multiplex Conferences to spread the word.

Avionics Magazine: Was it easy from that point on?

Gangl: Not really. We went and sold the B-1 program in the early 1970s. I had convinced the program office to let me provide a digital bus specification to be included with the aircraft requirements while the competition was still running. I knew more by then because of my experience with the F-15, so I could tell them more about what we wanted. With the F-15, I had only talked about the bus architecture and protocol; I hadn’t spec’d anything out in detail.

With the B-1, I wanted to have them use Mil-Std-442, which defines and standardizes the Manchester encoding technique. [Manchester encoding permits the use of a single transmission line because it has the clock embedded in the data. In H-OO9 there were two twisted pairs, one for clock and one for data, using sine/cosine waves.]

It all ended up with Rockwell International [the B-1 prime contractor] putting together a bus transmission spec using Manchester encoding. But the protocol was not that well defined. There were three buses on the aircraft; the Rockwell-defined electrical bus [EMUX], the Boeing avionics bus [AMUX], and the IBM centralized built-in-test bus [CMUX]. The meanings of the bit assignments being used, however, were slightly different between each of these electronic contractors, leading to some incompatibility in communications.

The B-1 program resolved these problems through a workaround. This, however, pointed to the need for additional refinements in the bus spec. I started to work on standardizing the bus specification. I approached the Society of Automotive Engineers [SAE], whose subcommittee, SAE/A2K, was working on a coax multiplex communications bus for Navy submarines. This group eventually became a partner in the development of the standard and also was our Mil-Std-1553 users group.

Avionics Magazine: How did these strands combine into a standard?

Gangl: I went to the Avionics Laboratory at Wright-Patterson and asked for help with studies and analyses to prove out architectural issues and put together a prototype system design. The lab then introduced the Digital Avionics Information System [DAIS] program to evaluate and refine the concept. A co-worker, Chuck Gifford, joined me in defining the waveforms, impedances, response times, etc., while I focused on defining the multiplex data bus concept, protocol, message formats, etc. We used all we had learned to define a digital, time-division, shielded, twisted pair multiplexed data bus.

This eventually became the Air Force version of the standard, published as Mil-Std-1553 (USAF). All this happened before we got to the point of interacting with the other services.

Avionics Magazine: And 1553 (USAF) became the tri-service Mil-Std-1553A?

Gangl: With 1553 (USAF) published, I marched into the F-16 program office to have them implement it on their system. But the Navy, not to be outdone, was developing its own concept for a bus. So a guy [Mike Keller] from the DDR&E [the Office of the Director of Defense Research and Engineering] called in the Navy guy [Dick DeSipio] and myself and told us to combine our resources into one spec. We were forced to negotiate, and the Army joined in as well.

The Navy, for example, wanted to have broadcast capability, whereas we had wanted the bus controller to tightly control all messages. So we have a broadcast option today. As a result of tri-service and industry negotiation, we developed Mil-Std-1553A, the first tri-service version of the standard.

The USAF version was only implemented on the F-16. It took about 10 years to get from the initial spark to the F-16. And, by the way, when the F/A-18 was competed, the Navy was still pushing for a bus of its own design. But their bidders all came back and said they would rather use Mil-Std-1553A.

We had three Multiplex Conferences, which were internationally supported by both industry and government. This gave us high visibility and, I believe, contributed to the success of 1553. Ultimately, with help from all participants, we came to an agreement on Mil-Std-1553B, which I froze after publication in 1978. A standard is not a standard if it is constantly changing! Future bus standards were to be a new standard, not a modification to Mil-Std-1553B.

Avionics Magazine: How many countries use Mil-Std-1553 today?

Gangl: I ended up becoming the U.S. representative to the NATO Military Agency for Standardization’s [MAS] Digital Avionics Systems Committee. I introduced 1553 as a potential NATO STANAG and it ended up being signed off by numerous countries, the key ones being the five countries that heavily participated on the committee. I believe the U.S., UK, Germany, France and Canada were the primary drivers for STANAG 3838.

I also was made U.S. representative to the Air Standardization Coordinating Committee [ASCC] Working Party 50, which included the U.S., Canada, UK, Australia and New Zealand. They also approved 1553 to become an ASCC standard.

France adopted it and published it as an option under a larger data bus spec, along with their own Digibus. There was even an unauthorized Russian translation of the standard some 20 years ago. Most other countries just use Mil-Std-1553 as is.

Avionics Magazine: Mil-Std-1553 has lasted 30 years and there are efforts to extend it further.

Gangl: I’m extremely happy about it. I was told that the lifespan wouldn’t be very long. But the standard was unique. It was the first to specify a standard interface, not a standard hardware design. It tells you what to implement but not how to implement it. The reason it has lasted is that the standard doesn’t, for example, specify technology, like you have to use CMOS [a semiconductor process]. It didn’t standardize on the hardware, circuitry, dimensions and connectors.

Now we have high-speed buses for computer-to-computer communications, but we still find [Mil-Std-1553] being used in aircraft systems. Mil-Std-1553 is like a side road, versus the freeway, for information such as control data and in other areas that don’t require extremely high bandwidth.

At the last 1553 SAE meeting I attended, they said the technology is here to encode and decode the logic much faster. I said, if you want to crank 1553 up from 1 to 5 megabits/sec [Mbits/sec], have at it. There are always organizations like the Aging Aircraft program office [at Wright-Patterson] that would like to upgrade old airplanes, but not rewire them, and need greater throughput.

Avionics Magazine: Have there been any surprises to you in the progress of Mil-Std-1553?

Gangl: Looking back now, the biggest surprise is that we were able to obtain tri-service and NATO consensus on a digital data bus standard. It was a good solution for its time, but there were many other variants proposed–some not so good, some possibly better. The second surprise is that it is still in use today. I am proud to have been involved in a concept that worked so well.

The standard evolved due to dedicated teamwork between the military, government and the aerospace industry, including international support. That was the key to 1553’s success. I may have lighted the match, but "the team" made the fire roar!

Mil-Std-1553 Chronology

1. Late 1967 to early 1968: 1 MHz, time-division, multiplexed avionics data bus concept;

2. Fall of 1968: Multiplexed data bus concept introduced to FX (F-15);

3. Fall of 1968: SAE/A2K subcommittee enlisted as 1553 users group;

4. March 12, 1969: H-009 bus spec (used on FX, renamed F-15);

5. 1970: An internal ASD Multiplex Committee established;

6. 1972: Digital Avionics Integrated System (DAIS) development started in the Air Force Avionics Laboratory;

7. Aug. 30, 1973: Mil-Std-1553 (USAF);

8. April 30, 1975: Mil-Std-1553A;

9. Nov. 1, 1976: Digibus Std (by Avions Marcel Dassault, Bruguet Aviation);

10. 1977: Start of Large-Scale Integration (LSI) chip development;

11. Sept. 21, 1978: Mil-Std-1553B;

12. November 1978: Air Force Systems Command (AFSC) Multiplex Data Bus Conference;

13. May 1, 1980: Mil-Std-1553 Multiplex Applications Handbook published;

14. November 1980: AFSC Standardization Conference;

15. Fall of 1981: French translation of Mil-Std-1553B for NATO Military Agency for Standardization (MAS);

16. Dec. 3, 1981: NATO MAS STANAG-3838 promulgated;

17. November 1982: 2nd AFSC Standardization Conference.