Any Avionics Designers here?

[BombayDuck]

I'm not even sure if this is the right place to post this, but does anyone on this forum either design or test avionics for a living? I have a couple of questions to ask - academic in nature - about component selection. I'd appreciate any help I could get. You could reply in this thread or send me a PM, doesn't matter to me.

Thanks.

[Avtrician]

Ask the question, there is all sorts on here....

:E:E:E

(That can be taken in so many ways.):bored::bored::bored:

[BombayDuck]

You scare me, Av! :bored:

I wanted to know if there were any guidelines on how one would pick a (micro/computer) processor to base an avionic system around. I have to design one for a university course. I have estimated the amount of data it will have to handle (per second) but I am still hazy on *what* it will have to do with it; and I need a guideline on what assumptions I can make based on, say, if I had to do complex tasks like curve fitting, or Digital signal processing or making relational models.

I could easily pick the latest processor off the shelf and it is more than likely it would do the job with half of its transistors taking a nap, but I need to justify why I picked it; also I wouldn't want to pick something a thousand times as powerful as the task at hand - why waste money (rare these days) and power (you can never draw too little unless switched off)?

I know this may not be the ideal place designers hang out at (we all secretly dream of being in the hangar) but it is worth a shot...

[theavionicsbloke]

Hia Bombay.

Firstly aviation has avionics designers but the guys you may bump into here are 'Systems Designers'. ie We take equipment boxes / indicators and provide wiring diagrams and assembly drawings for a fitters to install a complete system in to an aircraft. These designs area approved by a 'Systems CVE'. CVE duties are mainly to verify the safe and approved manor of the system design eg installation materials and techniques. However sometimes we may need to consider component level operation but CPR performance is very rarely one of them.

Your question relates to 'Component Level' design of which is normally of concern of the system manufacturer such as 'Honeywell', 'Rockwell Collins' etc.

Without specific details of what you want this processor to do is a little bit of 'How long is a piece of string' question.

So, what specific application do you have in mind ???

[BombayDuck]

Hey, thanks for the reply. I know what you mean, I'm actually more or less a system designer too, but I'm putting together requirements for a system; specifically an aircraft condition-monitoring & management system. If it was just going to log data I could have done with any processor, but my system has to be able to resolve faults (root cause analysis and such) plus maybe do trend analysis to figure out if things are going to really go bad some time down the line.

And yes, I've had a gander around Honeywell and other such companies' sites and they (understandably) hold their cards close to their chest, especially for the aircraft being designed today. I don't want to reply on a 15-year-old standard (ARINC 624-1) built to certify systems from those days...

[muduckace]

I'm actually more or less a system designer too, but I'm putting together requirements for a system; specifically an aircraft condition-monitoring & management system. If it was just going to log data I could have done with any processor, but my system has to be able to resolve faults (root cause analysis and such) plus maybe do trend analysis to figure out if things are going to really go bad some time down the line.

And yes, I've had a gander around Honeywell and other such companies' sites and they (understandably) hold their cards close to their chest, especially for the aircraft being designed today. I don't want to reply on a 15-year-old standard (ARINC 624-1) built to certify systems from those days...

Most aircraft use processing power with the same reliability expectations as the rest of the system 150% of max load. At a minimum I would design one that is redundant to all functions EG: 2 processors capable of 150% data transfer comparing data from aircraft components "Monotoring". Pre determined operating thresholds are established and if "A" channel reports a fault but "B" is within limits, the aircraft is still capable of operating on the "B" channel.

Some modern aircraft are designed with up to 9 multi function channels, they do not report a failure until multiple lanes of redundancy are used.

To discuss root cause with a little more clarity, we could take a flight control actuator that has manual feedback (indication), and manual valve operating control on 2 channels, then seperately autoflight feedback and valve control split on the same 2 channels. 4 components supplying compared data, one goes out of limits and gets reported as a fault. Generally aircraft do not compile trend analisys but the airline or component manufacturer will. Unfortunately most of us are working with what we consider new aircraft on a ARINC databus standard.


Not sure what you are looking for or what exactly is expected of you, would be happy to share what I know about average aircraft avionics systems.

I will say though that the corperate world is on the leading edge of civilian avionics, alot more software based function, smaller system computers (cards in a card file). The G450/550 and CitationX are great examples, never got my hands on the Globemaster.

[theavionicsbloke]

Bombay

For you sort of application, a higher end PIC processor may cope but in all honesty, for what you are thinking about then I think you will need to go down the route of a conventional pc based system.

I know this may be hard to believe but, Most role equipment based avionics navigation systems are now basically a pc using a standard Intel Processor. Some are even running a standard Windows XP Pro operating system.

I have recently put in to service an aircraft with 3! PC systems running the video & mapping systems.

Once you have this in place you can then employ standard or advanced database technology were data analysis becomes relatively simple.

Modern .net framework CLR (Common Language Runtime) software eg VB 2005 makes develpoment easy. Production and modification management becomes efficient.

Data transfer, well this then becomes an simple and obvious fibre optic solution.

So as you can see, this has got to be the future.

In reality though, the difficulty with the above is certification, as RFI hardware and software reliability becomes difficult to substantiate and demonstrate because,well franky, It's a new idea! However, I have so far managed to achieve this in none safety critical systems.