Reverse-engineering an airspeed/Mach indicator from 1977 | Hacker News Reader

Reverse-engineering an airspeed/Mach indicator from 1977

righto.com

105 pointspicture27 comments

This is awesome. I've seen a much simpler one of these (and other things) behind the firewall and was pretty amazed at how much is really back there. To see that the systems are as reliable as they are, and that they tell you when they aren't (e.g. this ASI's fault checking), is really cool from my perspective and I believe has carried forward to more modern avionics.

I really freaking love avionics. I'd love to get into software development at an avionics company but I have no clue where to start with that.

E-mail me (@gmail) if you'd like to chat about this.

Avionics, sadly is not like this anymore. When I started in avionics (late 70s, when this instrument was state of the art) I was fixing some radios that still had tubes, a mechanical crystal carriers that got moved to change frequency. In 10 years it’s was all chips. In a few more years it was glass cockpits. I saw the future and moved into programming. But for a while at least I programmed flight simulators to use my avionics knowledge. But now MSFS is as good as the software in any of those sims. Of course, being in a cockpit built with actual aircraft hardware with a full motion base and a visual system focused at infinity is an experience I’m glad I had. And I repaired a lot of instruments like the OP has for the sim.

> now MSFS is as good as the software in any of those sims

Are you sure? I'd expect the primary purpose of a pro simulator is to be a training aid in which case the simulation needs to be extremely accurate especially at edge-cases such as near/at stall speeds, various combinations of faults, etc - the dangerous stuff you'd actually want to train for and can't easily/safely replicate in a real aircraft.

I'm sure consumer-grade sims do a great job at replicating routine flight, but I'd bet good money a pro sim is much more accurate at replicating let's see a stall condition.

The software is good, but your point is about the flight model. The flight model is all about tuning. That was not my area but I watched it being done. It started with curves from the flight testing program and then many hours of trial and error parameter tweaking until the test pilot said that it was right. I saw a session in the middle of the night where the engineer tweaked, the pilot said it needed a little more, the engineer undid the tweak and the pilot said it was perfect. I also saw some Air Force pilots in a C5 sim (they were commercial sims) who could tell if the model was right by judging when they passed over the (simulated) highway in Dover, DE for a given engine setting. They were right, according to the engineer.

The point is, if you spend enough time on [1] you can tune the model. No one spends that much time on the MSFS. But for logic based stuff, like the working title VNAV implementation in the CJ4 it’s just a matter of understanding the expected logic and implementing. It is very impressive stuff.

[1]https://docs.flightsimulator.com/html/Samples_And_Tutorials/...

I'm not claiming that MsFS is ready to train F16 pilots, but there is definitely an acceleration of convergence thing going on here. Engines like Unity, large scale mapping, pixel streaming, weather sims and the interest from high end defense contractors is going to make this a thing faster than you might think.

There is something about this era of higher-end electronic assemblies that just ticks boxes for me aesthetically- metal can transistors, raw fiber boards with no solder mask, axial lead film capacitors, perfect leaded (or sometimes silver) solder joints...

Some electronics are still made that way. Bartel Amps has created a new ultra-premium category that he calls "masterbuilt". Take a look at the construction details:

http://bartelamps.com/wp-content/uploads/2017/12/IMG_5727_cr...

http://bartelamps.com/signatureseriesdetails/

That point-to-point style is more reminiscent of 1940s-50s era radio construction than the 1970's post-transistor but pre-integrated-circuit PCB I was getting at. But very cool nonetheless.

They sure did make old electronics pretty. At least sometimes, as in the case of this device that probably cost as much as my car.

I'm curious about the soldered wires, though. Especially in a device intended to be used in an airplane. No concerns about brittleness? I was under the (probably mistaken) impression most vehicular wiring was clamped or crimped these days for that reason.

Maybe those aren't really soldered, but just look kinda like it?

A crimp is done to a connector so you have to consider the reliability of the connector+crimp. Those soldered wires are direct onto the terminal so no connector.

I was thinking it might be specific to the use case, for sure. And maybe these are solid core wires, I can't really make it out from the picture.

Crimping on cars is probably just to save money, and some of the wire they use is garbage. Solder is pretty soft, especially lead based. Silver solders are more brittle and we don't use them for high reliability applications. Gold also has to be removed as it will embrittle the joint. For wires soldered to a PCB you can stake the wires with an electronics grade RTV, then bundle with twine, staked with epoxy. I have used low-outgassing RTV for space applications that costs $600 a tube.

One possible explanation: leaded solder is more elastic than non-leaded. (less brittle)

Modern mass produced products generally have had lead engineered out of them, including in the solder.

One popular product specification in this domain is RoHS, the EU Restriction of Hazardous Substances Direction, which largely bans lead and 9 other hazardous substances:

https://en.wikipedia.org/wiki/Restriction_of_Hazardous_Subst...

Although note that avionics fall under the "transportation" exemption, and RoHS does not apply.

That said, the supply chain has moved largely towards RoHS compliance, so some avionics inherit that for supply chain simplicity. Some don't; there's a cottage industry reballing BGAs from lead-free to leaded solder for aerospace applications.

Author here if you have any questions :-)

How would this unit be calibrated? It seems that wear an aging on the motors and potentiometers would change their characteristics slightly over time.

I assume there is a detailed testing procedure specified by the FAA, and they would replace any bad components.

But because of the servo loop feedback, most of the components aren't critical to accuracy. (Note that the internal power supplies are entirely unregulated.) If the motor slows down, for instance, it will still end up at the right location. The potentiometer is really the critical part, but it shouldn't change very much. And if it does change, as long as the resistance changes uniformly, it will still be okay.

https://support.cessna.com/custsupt/contacts/pubs/ourpdf.pdf...

This is an example of the pitot/static inspection procedure for a Cessna Citation Sovereign. The procedure itself begins on page 9 of the PDF.

Nice article and cool you are here for questions.

> The unit is powered by 26 volts, 400 Hz, a standard voltage for aviation.

Any idea as to the history as to why this is the case?

I assume higher voltage is useful for lower amps? But why 400 hz?

Seems odd to use such a high voltage if components such as this must have internal transformers?

Higher frequency allows transformers and induction motors to use less steel in their cores for the same power.

In aviation, that matters because weight matters.

The downside is iron losses become bigger (heat lost in transformers). In a plane that typically doesn't matter because you aren't worried about losing a couple of watts of electrical power.

In today's world, it is irrelevant because all voltage conversion is done solid state (which is easier from DC), and all motors (of new designs) are brushless and therefore prefer to run from DC.

Higher frequency means smaller and therefore lighter transformers, which is very important on aircraft. Nowadays a DC voltage supply would be better, but DC to DC voltage converters didn't exist when the 24V 400Hz standard was created.

For mains voltage we use 50-60 Hz because lower frequencies work better with very large AC generators in power plants and, and lower frequencies are more efficient to transmit long distances.

Others have described why 400 Hz is useful. But I'll mention that this frequency is very inconvenient for using the box on the ground. Fortunately CuriousMarc had some vintage HP boxes that we could use. We ended up hooking up an HP 3310A function generator to an HP 6824A DC power supply amplifier to produce the power for the indicator.

Ah good. I was going to ask that question (does anyone has a 400Hz AC power supply lying around? :) - well I guess the manufacturers have but not many more people)

I know on the ground, 400Hz was a better match for turbofan electric generators. Which were lighter and more efficient than normal ICE generators. And so used a lot when the equipment was mobile.

There's a long history of 400hz power in computing that goes back to early supercomputers like the CDC 6600.

https://en.wikipedia.org/wiki/CDC_6600?wprov=sfti1

CDC had a defense contract for early air data computers, used in the F-18. That business survives today as part of General Dynamics.

https://gdmissionsystems.com/airborne-systems

Great writeup! I didn't see it mentioned, but I assume the fast motor with reduction gears is helpful for a damping effect. To slow any twitching and show a sort of short term average speed.

Reply on news.ycombinator.com