0
$\begingroup$

I'm an electrical engineer and I'm interested in civil aviation, more specifically in fly-by-wire planes like the Airbus A320, Boeing 777 and others. But I fell in love with the Airbus A320 as it is the first civil fly-by-wire plane. I read a lot of its flight computers (FAC, SEC, ELAC), I saw its photos. But I want to have a look at their internal parts.

Does someone have photos of these internal parts? I want to have a look at those computer's internal boards. At last, I would appreciate any deep technical overview of how they work.

$\endgroup$

2 Answers 2

5
$\begingroup$

This is an indirect answer at best, but I think you'll get something good out of it.

I would definitely recommend you have look at CuriousMarc's series on YouTube where he and a couple other SMEs restore an original Apollo Guidance Computer to working order. You might think "that's not civil aviation, why?", but the AGC was the first digital fly-by-wire system. It was revolutionary both in terms of it's use of silicon integrated circuits and it's effects on software engineering. the AGC was also the basis for the first fly-by-wire systems installed on aircraft for atmospheric flight.

The other reason I'd recommend this is because it has all of the electrical cutting and shutting that you seem to be expecting. They spend a lot of time looking over original schematics and performing diagnostics with Marc's extensive personal collection of equipment in order to find individual failed circuits which require repair. They also had to find ways to interface the computer and emulate missing components, which involved a lot of bespoke design.

I'll be honest, I think there is a possibility you might find modern computers quite boring in comparison if your interest is limited to the circuits and does not include the software (not easy to tell from the wording of the question). This is for two reasons:

  • The AGC represents a time where the atomic components of the digital computer could still be seen with the eyes and cut/soldered out by hand. That is almost certainly not the case with modern computers, and I would also expect that the vast majority of people who work with the aircraft directly have never cracked open one of the computers for this reason. They are sealed (maybe even tamper-proofed), and are simply replaced as a unit and shipped back to the OEM when any invasive work is required.
  • A cursory search informed me that the original A320 computers actually used derivatives of the Intel 8086. If you're interested in early x86 micro-architecture (especially in fault-tolerant configurations), this is interesting. But, if you're expecting to open the computer and find the flight control laws to be implemented in hardware, you'll be a bit disappointed.
$\endgroup$
3
  • $\begingroup$ The big sea change in FBW has been the move of the analog-to-digital conversion of signals from inside the controller to out near or on the actuator. One of the big downfalls of analog FBW was the need to send sensitive analog low-voltage signals down long wire runs with numerous connectors, with the need to account for voltage drops in calibrations of signals. Anything that affects the line resistance, like bad connector pins, plays havoc with the system. Doing the digital conversion on-wing means the connection just has to be good enough for the signal to get there as an on-off. $\endgroup$
    – John K
    Jan 21, 2020 at 16:34
  • $\begingroup$ Thank you very much for your answers! $\endgroup$
    – mrhakerfox
    Jan 22, 2020 at 0:14
  • $\begingroup$ user44968, you gave me very interesting paper to read! What a pity that those who work with aircraft have never opened any computer. But the reason is very well understandable. $\endgroup$
    – mrhakerfox
    Jan 22, 2020 at 0:17
1
$\begingroup$

A little late but maybe I can help you.

The Spoiler Elevator Computer is based on, in total, four CPUs. Two 80186 and two 8086. Airbus computer in general, or at least in case of FBW computer, are based on a redundant architecture. They consist of a command and monitoring channel. As the name implies, the COM channel processes flight control laws and does servo control. The MON channel processes flight control laws, but there is no servo control, instead it monitors the position of the surfaces, namely spoilers plus in case of ELAC fail the THS and elevators.

Now some details: COM and MON are physically separated in the computer housing. In the center of the LRU are four slots where a PSU modul can be fittet. Two of them are used. One PSU modul for COM and one for MON. The slots separate the COM and MON Boards. COM consists of three boards: Main CPU 80186, Servo CPU 8086 and a Servo/IO board. MON the same but without Servo/IO. The CPU Boards are identical, there is no difference between COM and MON. Software for the Main CPU is stored in an OBRM (128 KByte), Airbus terminology for a ROM Module. When you saw outside pictures of the SEC: the OBRMs are the labeled cartridges. They are directly connected to the Main CPU board. Software for the 8086 is stored on two socketed ROMs (64KByte together) on the board. Servo and Main communicate via a dual port ram. Servo jacks or rather servo valves, related LVDTs and so on are connected via relays located on Servo CPU and Servo I/O. For THS and elevator, COM and MON relays are in series. For spoilers relays are located on Servo IO (if I remember correctly) and controlled by MON. Relays can be controlled by the CPUs or the Engage Logic. The engage logic is a set of PLDs with basic logics implemented, taking over even when all CPUs aren't working. For example, the Main CPU is monitored by a watchdog timer. It has to be periodically resetted, or it will trigger the engage logic, which in turn will switch the relays. I think even the SEC push button on the overhead panel is connected to the engage logic and the small light which will illuminate in case of failure. Arinc429 output is partly done by the CPUs, they generate four bytes with data and label included by software, and then push it into a PLD byte after byte, which then does parity and parallel to serial conversion plus arinc adaption (RZ Coding). One transmission PLD is fitted on Main and one on Servo. Main sends data to the FCDC and Servo should process the output to the opossite channel (Arinc429 Cross Talk Bus). Arinc input is managed by one set of PLDs on the Main CPU board. They validate the data and load it into a dual port ram. Not the same dual port ram as mentioned above for main-servo communication! The "hardware driver" software on Main loads byte after byte from the dual port ram into its own ram and does final depacking. (mostly just a few bit shifts in case of air data and so on and shifts plus masking in case of discrete words where each bit represents a state) For discrete IO adaption circuits are used (changing voltage levels) and then inputs are multiplexed on the 16th bit of the x86 data bus. Output the same just the other way around. All servo related hardware is located on the Servo CPU and Servo IO board. LVDT and RVDT signals are demodulated in hardware and then read in by the 8086 via a "diy" integrating ADC. Analog input from potentiometers like sidestick, thrust lever and speedbrake lever are processed by the 80186. They are multiplexed onto a single ADC. On the Servo IO board op amps for driving and regulating the current commanded by the 8086 through the servo valves are used. A set of four dual DACs drive the op amps. Momentary current of each servo output can be read back, also by the 8086, to verify correct operation of the analog servo current control.

Maybe I forgot something but for the beginning, that should by much information. Basically the SEC is "just" a big discrete build MCU, or let's say two MCUs (COM and MON). Hope I could help...

Simon

$\endgroup$
3
  • $\begingroup$ I always thought it was 80186 and M86000 (so it is dissimilar). $\endgroup$
    – Jan Hudec
    Sep 26, 2022 at 10:06
  • 1
    $\begingroup$ The Motorola 68k is used in the ELAC. So there is a dissimilarity between ELAC and SEC. $\endgroup$
    – Simon
    Sep 26, 2022 at 10:15
  • $\begingroup$ That massive paragraph in the middle could use some formatting to make it readable. $\endgroup$
    – FreeMan
    Sep 29, 2022 at 18:30

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .