Menu Sign In Contact FAQ
Banner
Welcome to our forums

A kit is a kit

Peter wrote:

When you are making a $1000 piece of kit (which does the same apparent job as a $5000 certified version) you just can’t use $100 milspec circular connectors. Something has to give.

Non-certified EFISes probably have a life about 10 years or less. Then they are replaced by something better/newer, not because they are broken, but because they are old fashioned. You can do that when the prices don’t go through the roof and you can do it yourself. One would also believe that panels going black would be a common problem, but it’s not. I have never read or heard about a single case, other than with certified G1000s. Not to mention certified transponders and radios. At my club there are problems all the time. Talked to the guy in charge of those things the other day, and the latest he did was to send away the brand new Becker transponder and purchase an ancient mode C on e-bay and mounted it in one of the Cessnas. I just hope my Trig units will hold up, and he was very interested in them also.

I think you are a bit too negative. The non-certified gadgets just keep on working, against “all” odds. The producers themselves say it’s because they use “certified” principles and design criterias, but take advantage of newer, better and cheaper components. I cannot possibly verify this, but the statistics speak for itself IMO. Also, the ECB unit has been on the market for over a year now, and no complains to be seen anywhere.

The elephant is the circulation
ENVA ENOP ENMO, Norway

take advantage of newer, better and cheaper components

There are almost no “newer, better and cheaper components” that have come out in the last 30 years or so. In fact components have become more expensive as manufacturers realised that making commodity parts (e.g. a LM358, from c. 1972) is not profitable whereas making specialised parts (much of what e.g. Linear Technology makes) gets you 5x the price. So the designer’s challenge today is how to make a product using 30-40 year old parts

There are exceptions for sure e.g. 16-bit A-D converters for €1 which didn’t exist till recently…

However, the electronic component cost of say a GTX330 (€2000+ list price) is about €50…

Non-certified stuff is much cheaper because

  • the barriers to entry are low
  • it is made by small lean companies, not by old once-huge companies with loads of old guys looking forward to playing golf in retirement (Honeywell, anyone?)
  • no fat old parent company or investors looking for a return
  • the mfgs don’t need to play even the simplest QA scams (e.g. ISO9000) so no unproductive “QA” staff
  • the stuff is sold direct so no dealer margin
  • the average customer is heavily self-selected on tech expertise (high anorak score ) so is probably more clued-up than the average dealer/installer most of whom are just wiremen and when something doesn’t work they phone you up and waste the time of competent (=expensive) staff
  • selling direct = good cash flow (everybody pays with the order) so no need for banks, investors and other sharks; the business can grow by self-financing
  • selling direct = good input on new product features and functionality (less needs to be spent on R&D because customers tell you what they want)

The above pretty well describes my business too…

For sure a lot of certified stuff is crap (Narco, anyone?) but I suspect there is a correlation between the population of certified aircraft and the population of neglected aircraft (parked outdoors all year, abused by syndicates/renters). Whereas most homebuilts I know of are operated by people who are not tight on money, and they are nearly all hangared and generally looked after, and taken out on nice days. Also I bet you homebuilts fly far fewer hours than club-level certifieds. Anybody with the FR24 app will immediately see that, the lack of Mode S use notwithstanding. All this will result in a poor reliability figure for certified avionics in club-level aircraft.

Administrator
Shoreham EGKA, United Kingdom

Peter wrote:

There are almost no “newer, better and cheaper components” that have come out in the last 30 years or so

Oh really? Like the chip I have on my table right now, a complete ARM SoC that costs around $3 even if you’re just buying one – something that didn’t exist even 15 years ago let alone 30? Or even the improvements in power transistors (ever lower Rds (on), ever higher current/voltage capability) over the last 30 years.

Sure on the analogue side, the old familiar part numbers are still used. But there have been enormous strides in microcontrollers/microprocessors/memory. Whenever I’m doing my nerdy stuff at the University of Zaragoza, I point out the Convex supercomputer that’s in the museum (which cost the equivalent of 1M euro 30 years ago) and hold up a Raspberry Pi, which performs better, has more memory, has more mass storage, and fits in the palm of my hand and costs 30 euro.

Much of today’s avionics will be on the digitial side. A typical EFIS display is likely to have half a dozen ARM cores in it.

Last Edited by alioth at 17 May 15:46
Andreas IOM

I agree, but you don’t need a dozen ARM cores to run an EFIS box for a homebuilt.

You could do it with any 80×86 board i.e. some PC motherboard and a low-end VGA card. You aren’t doing 1080P 60FPS video…

And the uncertified EFIS boxes on show at EDNY aren’t exactly perfectly slick in the graphics implementation.

To give an extreme example, in the mid 1980s I was doing graphics with an 8MHz Z80 driving a dedicated graphics controller from NEC – the UPD7220. The graphics performance, say 1024 pixels wide, would be easily enough for a smooth EFIS screen.

Today you would buy an ARM board for 100 quid, with a massive LCD on top, and do it all with that.

So it has become easier and cheaper but I don’t think that costs have come down enough, in the context of a four digit priced product.

MOSFETs are better now too but you don’t need a 0.001 ohm Rds unless pushing the boundaries.

I guess the one item which is definitely easier now is solid state AHRS. No need for the KG102A

Administrator
Shoreham EGKA, United Kingdom

Peter wrote:

you don’t need a dozen ARM cores

They might not be really required, but they might well mean the most cost-effective solution. Well, ok then, half a dozen, perhaps.

EBZH Kiewit, Belgium

Peter wrote:

So it has become easier and cheaper but I don’t think that costs have come down enough, in the context of a four digit priced product.

I believe they have. I am heavily involved with medical devices and I know certainly that most products with 4 or somtimes 5-digit EUR prices consist of parts worth less than 100 EUR. Complexity-wise IMO these device are in the same range as avionics and there are normally safety implications as well.

Hajdúszoboszló LHHO

Peter wrote:

I agree, but you don’t need a dozen ARM cores to run an EFIS box for a homebuilt.

You don’t but they will already come with the display etc. (for many off the shelf displays). ARM is very scalable, you can start off with the ARM M0 (up to about 50MHz, no cache, simple 32 bit address space, Thumb instructions only, no supervisor mode or MMU or FPU, simple 3 stage pipeline) up to the high end 64 bit stuff you find in a tablet or phone which will be on a par with some desktop CPUs, and it’s generally cheaper than anything in the x86 world and consumes a lot less power for the same performance (and for embedded, which avionics is, there are other advantages to ARM such as the very low interrupt response latency even on the very low end). You can also find ARM SoCs which will do everything an EFIS needs in terms of CPU, peripheral interfaces etc. in a single chip that costs buttons. Forget low end VGA cards etc. – the whole thing is in the ARM SoC and you don’t need very much on the board it’s soldered to to make a complete computer.

Last Edited by alioth at 17 May 23:41
Andreas IOM

I know… we are using an ARM board at work also.

I think my point was that this €5k (say) product might have contained €300 in parts in 1985 and €100 in parts in 2015, so it makes little difference. Except of course €5k was a lot more money in 1985, but that’s a different point

A lot of stuff was much harder in 1985 e.g. precision analog e.g. voltage references with X ppm/degC drift (I was doing that in the 1970s but it involved pretty weird stuff) but there is nothing applicable in avionics I can think of except AHRS, and hi-res LCDs. Flow totalisers still squirt out the data at 4800 or 9600 baud on RS232

IMHO the avionics technology scene is driven by factors which are little to do with what is technically possible and what it costs to do it. You have

  • a very slow moving market, with very little need for functional innovation (VFR and IFR has not really changed for decades)
  • the scene is dominated by big players who run everything they can as a cash cow, preferably for ever
  • most installs are dealer-only and since most dealers are not too technically competent, and have to provide a warranty to the customer, they are highly loyal to the dominant vendor who supports them technically and with a warranty
  • products have to be built with parts which will be available for many years

Looking at the homebuilt scene, even if people chuck the stuff out every 5-10 years, you still have the lack of a need for new functionality. Take a walk around Aero Friedrichshafen and ask yourself how many more lookalike EFIS boxes can sell into the marketplace…

Administrator
Shoreham EGKA, United Kingdom

Peter wrote:

with very little need for functional innovation (VFR and IFR has not really changed for decades)

VFR has indeed changed. The rules are the same, and you can fly using a compass and a map today as well. But, first came the GPS, and things started to happen. Then moving maps, and now EFISes that can do 3D nav on autopilot. Compass and map will soon disappear, not because it doesn’t work, but because obtaining updated paper maps for an arbitrary piece of Europe is a PITA today. In moving map app i’t a click on an “OK” button. This is transferred directly to the EFIS also. I believe we are only at the very start of this change. Moving maps on pads will also eventually disappear, or become merged with EFISes somehow. Lots of things are happening under the hood of these EFISes, that’s where things are really changing right now.

For “digital” VFR navigation there are nothing of old procedures and regulations that holds the development back. IFR still uses procedures and solutions that were made for navigation technologies from the 60s and earlier.

MGL has this thing they call “GLS” that work on any air strip. Its an imaginary “ILS” shown on the HSI and in 3D (synthetic vision), and is dead simple and intuitive (except for IFR rated pilots I guess ) They also have a gliding thing that helps you in the event of engine failure, and because of 3D map it really works.

We don’t need any of it, it’s just that it’s so much simpler to use than compass and map, that it creates a need just because of that. The prices are such that people can afford to upgrade every 10 years or so, and it get’s a life on it’s own, not unlike PCs and mobile phones. Myself, I could get by with just a compass and map just fine, but when it’s there, and it’s simpler to install and use than old steam gauges, and and it doesn’t even cost more.

The elephant is the circulation
ENVA ENOP ENMO, Norway

The real IFR revolution will be when the ILS cross pointer gets replaced by a SynVis runway, and path guidance by more intuitive “tunnels in the sky”, arrows etc instead of HSI needles.

In the current world, this is a bit dodgy except for approaches where both lateral vertical guidance is provided by the GPS. But in the long run, everything will be LPV, so the needles will be obsolete.

And it will be SOOO much easier to fly IFR.

Biggin Hill
Sign in to add your message

Back to Top