How many degrees in a circle (GTX345 crashing, avionics manuals)?

Michael_J wrote:

Reminds me of a story from a controller at Copenhagen ACC:

The computer at Stockholm ACC crashed some years ago after someone doing data entry had entered a zero in a field where zeroes were invalid. The computer divided by this number and…

Update to this post:

It took me awhile but I finally got my hands on the GTX345 manual (I despise the fact that the FAA allows these companies to hide install manuals).

As it turns out I was told wrong by my shop, the heading does go over a 429 interface.

In my case, they went Aspen → CX80/GNS480 → GTX345 for the heading all over 429. This should be fine, unless there is some limitation to daisy-chaining 429 data that I am unaware of. It breaks down at the 345 only on heading 360. The Aspen and CX80/GNS480 get along just fine together for heading purposes.

So that tells us that Garmin has likely gotten the 345 certified with an ARINC429 bus that is not compliant for heading input and is trying to lie about it.

I agree with your assessment and understand the differences, Peter. It’s a matter of the small detail that ARINC429 has hundreds of pages of specs that should be adhered to and RS232 is just pins that they can put whatever on. I also agree that it’s sloppy software development. Garmin admitted to the first customer who had this issue that it was a div0 error on the 429 heading input. I also would point out that the final fault lies with the FAA, for collecting tax dollars to certify these devices and doing no such thing. If an avionics maker is allowed to ship a certified device that will take ARINC429 inputs that is not compliant with the ARINC429 spec, what’s the point of having the FAA involved at all?

The status of avionics in terms of the FAA and general aviation in the U.S. is that the FAA exists to service their friends in the avionics business, their brother-in-law with the consulting firm that they herd avionics makers into hiring, and their other brother-in-law the DER (designated engineering rep) that gets paid by the page to autograph things. The end user is denied documentation, denied a clear source of recourse in case of false marketing and failed units, and denied access to the certification data. The only purpose I serve as the end user is the relatively insignificant role, in their minds, of paying for all of this.

No one is talking about it much, but the recent legal reform of the FAA that passed in the U.S. had a much bigger change than allowing people to fly pistons without a medical checkup. It also, as of next year, gives our local district courts legal jurisdiction over the FAA. The FAA has been hiding what they do from the public for decades, and that will come to a halt when the new law takes effect. Considering the nature of an injury attorney is to seek the path of greatest pocket depth, I suspect it won’t be long until they stop taking aim at insurers and the estates of dead doctors who crash Bonanzas, in favor of the much deeper pockets of the U.S. Treasury.

The FAA may be inept but they’re not stupid, they are engineers and lawyers, after all. They have been promoting a culture of taking all of the authority while accepting zero liability for the last 40 years. That culture is promoted by instructors who wish they were check ride examiners and maintenance/avionics personnel who wish they were DERs at the FAA’s behest. When they can be compelled by any federal district judge so produce evidence in a civil suit I suspect their cushy jobs won’t be so cushy anymore. We shall see.

What a wonderful manifesto to destroy the best aviation regulatory system in the world. Just what we need loads of frivolous lawsuits against the FAA….

Just a tip, don’t move to Europe. If the FAA upsets you this much you don’t want to start dealing with EASA.

RNCTX wrote:

I also agree that it’s sloppy software development. Garmin admitted to the first customer who had this issue that it was a div0 error on the 429 heading input

Here’s what I’m struggling with – maybe I’m missing something, I probably am missing something! – but why would anything, let alone a transponder, want to divide something by the heading or anything derived from it? It seems bizarre. Consider the difference you’d get by dividing some value by 360 and dividing the same value by 1 – a trivial heading change would result in a huge difference in the result of any calculation that divides by the heading.

I can imagine the heading being divided by something else, and 0/x is perfectly well defined.

RNCTX wrote:

It took me awhile but I finally got my hands on the GTX345 manual (I despise the fact that the FAA allows these companies to hide install manuals).

Would you expect Mercedes or BMW to give you unrestricted rights to their engine management software or their dealer-only maintenance manuals?
Incidentally, only those functions which form part of the Avionic equipment TSO need to be certified to an appropriate software design assurance level. Other non-TSO functions will often have level D/E software and is not subject to any regulatory verification.

alioth wrote:

but why would anything, let alone a transponder, want to divide something by the heading or anything derived from it?

We don’t know if the crash/reset is the result of a division by zero. I think (don’t want to re-read the whole thread now) that the division by zero was just given as an example for a larger computer failure. But a value of zero in a place where only values of one or greater are acceptable can cause an exception in the program. If this exception is not handled well some strange behavior like this constant rebooting can result. On the other hand: The Apollo 11 mission was saved because the guidance computer did exactly that: After an exception, it restarted itself and thereby continued to function well enough to enable the landing as we all know.

wigglyamp wrote:

Incidentally, only those functions which form part of the Avionic equipment TSO need to be certified to an appropriate software design assurance level. Other non-TSO functions will often have level D/E software and is not subject to any regulatory verification.

However, it shouldn’t be acceptable for an ancillary function (transponder doesn’t need heading for it’s primary function) to crash the whole thing and interrupt the device in its primary job for which it’s certified. Especially if the issue is something so trivial. It shouldn’t be possible to get away with such a stupid design. That’s my opinion.

wigglyamp wrote:

Would you expect Mercedes or BMW to give you unrestricted rights to their engine management software or their dealer-only maintenance manuals?

I certainly wouldn’t mind it. Their competition will take those cars apart anyway. The tools and manuals should be available (as the cars can be serviced outside the authorized network) but it’s not convenient if you just want to look around or tinker (i.e. you’re not a professional).

wigglyamp wrote:

Would you expect Mercedes or BMW to give you unrestricted rights to their engine management software or their dealer-only maintenance manuals?

Not the internal software – but maintenance manuals, yes.

Only selling installation or maintenance manuals to a limited audience, voiding warranty of a car if it is serviced or repaired anywhere else than authorised dealership, or if generic replacement parts are used, are all practices that are designed to restrict competition.

In the automobile industry, many of these practices are – quite rightfully – illegal.

I don’t know if car maintenance manuals must be sold to anyone, but diagnostic tools and original manuals are all widely available.