The end of a once great company - Honeywell / Bendix-King?

Certification is a consequence of ICAO.

So it cannot be avoided - except within a sub ICAO national regime which is not valid outside that country; e.g. the US Experimental Category. If you want unhindered international flight (noncommercial) then you have to accept this. It's not ideal but it's the best we have, and if it wasn't for ICAO, GA would be banned in most of the world.

But as with any certification e.g. CE, ROHS, ISO9000, you name it, there is a gravy train on every corner, and people will use it to inflate prices, protect a business sector, etc. That is where it gets cynical. Anybody tried BASEEFA? It is usually easier to protect one's revenue stream by manipulating certification than by bringing out some amazingly good product. I don't see human nature changing...

The reason why one can get partly defective products like the KFC225, which is a great autopilot and IMHO the best performing autopilot in existence apart from far more recent stuff like the GFC700, is that there are no controls on reliability. Certification of avionics is little or nothing to do with reliability. Engines are a bit different... So nobody from the FAA is going to look at the design and spot (in 5 minutes max if any good) that the current limit of the servo amp is 2A whereas the motor stall current is 1A, and with no slew rate etc protection / no input filtering you can burn the motor out.

I was able to work all this out only because I managed to obtain the maintenance manuals, with circuit diagrams. It wasn't easy...

but due to the substitution of "equivalent" parts over the production life turned into dodgy unreliable stuff. This problem is endemic in some sectors of manufacturing in China;

This is why I build all PCBs in the UK, populate and sample test them, and send them out to China for the labour intensive bits like cable moulding. But the Chinese are continually increasing prices... however I don't see Chinese being a player in avionics because avionics is the very last area where labour costs are significant. When it costs say $200 to make a GNS430W and stick it on the shelf, and the trade price is say $4000, no way is Garmin going to do a Steve Jobs and get Foxconn to make it for $100, with the massive amount of oversight needed to maintain QA and kick ass when necessary.

China is the Wild West. We had a contractor out there go bust, costing us about $10k in lost stock etc. The managers were all Taiwanese and they escaped rapidly to Taiwan because, apparently, they owed some unpaid wages to the workers and thus were liable to be terminated (I mean killed).

That said, the PCBs in the KFC225 servos are made in places like Indonesia, so it looks like some players at least are making finished PCBs out there. That I find really suprising, with the cost of the PCB being about $10 and the list price being over $2000. The whole servo sells for about $3000. It seems crazy for an American company to make trivia like that overseas, and worry about component substitution and 1000 other things.

Yes; this is the issue with delegating design authority.

It's been in the press lately with Boeing and Airbus, too.

But the certification authority will never be able to exercise "physical" quality control over designers and manufacturers. The most they can ever do is require a specified weight of documents to be dumped on their desk, with the right boxes ticked.

It would be entirely possible to submit false flight test data for a FAR23 aircraft and get it certified. You would need only 2 or 3 people to be in on it. Until you get some "unusual" crashes, perhaps...

My estimate, based on servo serial numbers, is that King made some $10-20M selling replacements for burnt out KFC225 servos, over the past 11 years during which there was zero OEM business. You might call that a disincentive to fixing it. An accountant certainly would. A lawyer probably would not, but there will be very few crashes in which the AP might be implicated, and any properly fatal one will never leave the required evidence.

The certification process is pretty perverted. If you can find an FAA FSDO in which everybody in the office says that an EHSI is "EFIS", what chance do you have of finding somebody who might spot that putting 1A into the base of a 2N2222 is not a good idea. Or in EASA. So the process has to be rigged to work on box ticking. There are a few people who work out basic rules e.g. control forces have to be progressive (monotonic) but after that it is nearly all box ticking.

The old ICOM radios were UK CAA approved. Presumably this meant they could be used in the air openly. Later, the CAA stopped doing approvals and delegated them to some govt agency. But nobody bothered about the approvals after that, anyway...

A lot of "vintage" planes have a handheld radio mounted somewhere. It's their only bit of electrical equipment.

As for crap design leading to intermittent connections, reading some of the tips in this repairman's handbook is quite revealing.

https://www.bendixking.com/en

Here’s a short “dictionary” that explains why many of those boxes look somewhat familiar…

AeroVue (Honeywell APEX)
AeroVue Touch/xVue Touch (genuine BK, nonexistent?)
AeroFlight (Sandia SAI 340)
AeroWave (Honeywell?)
AeroPoint 200/300/xPoint 100 (JP Instruments)
AeroCruze 100/xCruze 100 (TruTrak)
AeroCruze 230 (genuine BK, nonexistent?)