Ethereal Wake

End of the Blog

2011-10-08T09:31:45-04:00

I'm going to follow my normal pattern of behavior and abandon the blogging attempt. In the modern world, the traditional model of blogging isn't exactly useful where online personal journals have been absorbed into things like tweets, wall posts, and streams. Attempting to maintain a blog, especially within the inconvenient quasi-manual apparatus I set up, seems redundant and unnecessary. After all, who really cares what I type here that isn't already followed, friended, or circled?

Now, some things I'm working on may have general utility to others: describing the though process behind photographs, describing lens design, builder's log from an aircraft, etc. For these things, it makes more sense to use a hierarchical website layout instead of a temporal one. In the next couple months, I'll try to see what I can get together (with more automation this time).

That said, I did successfully solo yesterday.

Still At a Positive AGL

2011-08-29T14:50:49-04:00

I haven't really been up to much of late other than work and flying school. Ground school is held two nights a week (but is nearly over) and I've been scheduling flying lessons four times a week (still pre-solo, so I have a while left to go). For the math-impaired, that adds up to six days a week where I spend a considerable amount of time at the airport. Tack in work and there isn't time for much else right now.

Of course, in those few hours I do have, I'm busy planning my future homebuilt. The planning process consists of a few basic things. Before anything else, I first need to study the plans and understand how the construction process works. Without that understanding, I can't move on to later stages, such as design alterations and workspace considerations.

By design alterations, I'm not referring to changes to the airframe. I'm content to remain with a normal level of crazy, so let's not elevate this to bat-shit insane.

I am not referring to instrumentation and avionics, either. These are so user-specific as to not even be seriously considered part of the plans (and can be easily deferred to the end after another year or two of technological improvement has passed).

Some design alterations have become fairly standard, like replacing the Cozy's hand-cracked retractable nose gear and manual landing brake with electrically actuated systems. So while they need to be considered, I'm not worried about those yet (since there are off-the-shelf drop-in systems).

Right now, I'm thinking about precipitation static (or p-static). Basically, this can be thought of as lightning in reverse: charge builds up on the airplane as it passes through precipitation and when it builds up far enough, it starts discharging and can generate enough RF static to disable radio operation (if not zap the electronics or occupants directly).

For metal aircraft, this issue is easily resolved with the conductive skin and some lightning-rod like attachments; however, for composite aircraft, the situation is more complicated. Fiberglass is a fairly good insulator, so commercial aircraft (such as those by Cirrus) include a fine metal mesh in the top layer. The Rutan-derived composites (such as the Cozy) do no such thing by default. In fact, they take often advantage of the lack of skin conductivity and embed the antennae so they don't stick out into the airstream.

The vast majority of builders make no attempt to resolve this known risk and there do not appear to be a large number of reports of problems. I can speculate on a couple of possibilities for this: The majority of aircraft could be staying with the day VFR mission originally envisioned. The problem could be more common but simply not recognized or reported. The carbon black primer could be sufficiently conductive to alleviate the p-static issue. Alternatively, this problem could simply be an imagined one.

Of course, as an electrical engineer, I'm convinced I could spend some time to study this problem. For an embedded whip antenna, the ground plane is a necessary feature and can easily be integrated into the protective mesh. For half-wave antennae, the mesh shouldn't interfere if it is co-planar with the image charge plane. When it is necessary to protect the area over the antenna, a very coarse mesh could provide a level of protection while minimizing RF shielding.

Now, while I figure out where I'm going to build the thing, I can do some scale components to test the RF effectiveness of various p-static mitigation techniques: Lay up some small wings and hook it up to one of the Network Analyzers at work: do I still have a functional antenna?

Roller Coasters without Rails

2011-07-12T22:44:34-04:00

One of the most unnecessarily terrifying elements of flight is the stall. The concept is quite simple: normally, as you pull back on the yoke, the plane pitches up, oncoming air hits the wings at steeper angles, and you get increased lift. Eventually, the angle becomes too steep, the airflow separates from the wings, and the bird forgets how to fly.

The whole concept seems frightening but really isn't (assuming you have altitude to spare). Lift may have decreased, but it hasn't become zero. Most importantly, the plane will naturally pitch forward as it begins to fall, decreasing that problematic angle of attack and restoring lift. If you simply let go of the yoke, the plane will probably do a good job of righting itself. There are a few more details, such as avoiding altitude loss and spins, but it isn't that bad if you respond appropriately.

The truly frightening aspect, for those of us who hate roller coasters, is the sensation. Of course, seeing my obvious discomfort, the instructor feels we will keep coming back to stalls until I am no longer bothered by them. The thing is, for that second or two while in stall, you briefly experience negative g's. It's just like cresting the top of a roller coaster, a crazy roller coaster without rails carrying several hundred pounds of gasoline.

High As a Kite, Maybe a Little Higher

2011-07-12T22:44:34-04:00

This past weekend, I completed my first flying lessons. (In a real airplane. Off the ground. For real.) Being a rational, cautious individual, this was approached with some anxiety, but not nearly as overwhelming as I expected.

The most immediate thing you notice is how overwhelming the initial lessons can be. Even before you start yanking and banking, you've already gone through an hour of pre-flight, flight clearances, and the like. Sure, it's important but the numbness from information overload is already setting in. I could probably have stopped after pre-flight and still felt like I learned more than enough for one day. Instead, I'm (poorly) taxing from the tie-down to the runway where the instructor points the nose down the runway and into the air. During this initial lesson, when I'm still trying to remember at what RPM we tested the mags, I'm already trying to keep straight and level flight, make thirty-degree banks, and make altitude changes.

Of course, this being my first time in a small plane (without all that many trips in an airliner), I have to deal with the new physical sensations. Assuming you are in good health, your body maintains balance by monitoring the direction and magnitude of the forces exerted on you (namely gravity). In a plane, especially while maneuvering, gravity is pulling at odd angles and at strengths other than 1 g. In the first several minutes of flight, my instructor felt the need to comment on this with such things like “don't worry, you aren't going to fall out of the plane.” Sure, I know that, but please explain it to my completely confused vestibular system. Thankfully, the non-debilitating panic did not survive the first flight.

So, thus far, I've learned the basics about keeping a small plane from impacting the ground. And I've learned that I can get in a small plane without throwing up. At this point, the only impediment to earning my wings will be my patience, determination, and continuing lack of a major screw-up.

Washington Goes to Google

2011-06-23T22:04:23-04:00

The big tech news for today comes from the Wall Street Journal: the FTC is prepared to serve Google with subpoenas as part of an antitrust probe. The amusing part is not so much the legal aspects, but how this issue turns on-line communities (such as those at Ars Technica) with a record of general hostility to free markets suddenly turn into an extended wing of the Libertarian Party. Arguments deemed inadequate for Microsoft or Apple are now suddenly appropriate in the case of Google.

The most common argument in defense of Google is the ease by which users can switch search engines. They argue that if Google's coercive actions result in an inferior product, users will simply switch to Bing, Yahoo!, or the myriad alternatives and the situation will resolve itself. There is a fundamental flaw in this argument, an attribute of the knowledge market: the user does not know if Google is providing the optimal answer or not. A user's only metric is whether they receive acceptable answers to their queries.

Due to inertia, Google will remain dominant so long as users receive a experience sufficient to their needs and expectations. When a user searches for “photo sharing,” they don't know that Flickr should be the top link or else they would never have asked Google. If Google's own Picasa is provided in the number one slot, the answer is acceptable as judged by the user since it fulfills their need and expectation. This means that should Google give preference to its own products in search or advertising placement, it will unfairly benefit Google's own products without negatively impacting users' perception of Google.

This ability is certainly not unique to Google. Microsoft could easily give preference to SkyDrive over Flickr in Bing; however, Bing's market share is a small fraction of Google's and thus would not significantly impact Flickr's overall market share. In contrast, Google's dominant position in the search market, the fundamental means by which users discover content and products, gives it the ability to produce significant barriers to entry in sectors where it wishes to compete, a key concern of Antitrust law.

Does this mean that Google is guilty of antitrust and corrective actions should be taken? Of course not.Natural monopolies (which Google arguably is) are neither illegal nor bad. Examples include utilities, sports leagues, and operating systems. However, each of these has been investigated by antitrust bodies and subjected to oversight and regulation due to their ability to create economic damage. Google should be no different.

Kitplane Safety

2011-07-12T22:43:00-04:00

So, as mentioned a few days ago, I've been obsessing over airplanes. Of course, when it comes to aircraft, ‘failure modes’ are a topic of discussion. The natural tendency for anything in the air is to come tumbling down, so an aircraft needs to be doing something to remain aloft. When that something fails, you have developed a bit of a problem.

For most people who want to own their own plane, self-built becomes the only affordable option. However, given the safety-critical nature of aviation, the idea of an airplane being built in the garage of some random scrub is rather terrifying. This thought crystalized in my head when someone on sonextalk mentioned that approximately 1 in 46 completed RV kitplanes end in a fatality (what?!).

The FAA reports that experimental aircraft (includes home-built) have a fatality rate significantly greater than that of the general aviation fleet [LA Times, 9 April 2011]. While there has been much debate over the accuracy of there figures, it would be difficult to argue that experimentals have a better overall safety record. The real question becomes how much of this is due to structural deficiencies of the planes versus deficiencies in the pilots.

Normalization by flight hours, as the FAA did, is biased towards pilot error. The more time you are in the air, the more likely you are to take off without refueling, fly in marginal conditions, or otherwise get unlucky. A structural defect resulting from shoddy workmanship is probably going to exhibit itself rather quickly. I postulate that normalizing by fleet size will be biased more towards reliability of the airframe and less on the pilot's judgement.

Van's aircraft and their RV line is generally accepted as the market leader. According to their web site, 7,321 are known to be completed and the NTSB lists 154 fatal events [1]. That means approximately 1 in 48 RV's have met their end in a fiery inferno. Given that the completion number is purely an estimate, verify with FAA registrations returns a similar 1:46[2].

To add another data point, Sonex is an up-and-coming manufacturer that emphasizes cost and completion time. An estimated 340 completions are reported on their website while the FAA counts a much higher 395 registrations [3]. With the NTSB only listing 3 fatal events, this gives a significantly better rate of 1:130. Their emphasis on minimizing cost and build time does not appear to have impaired safety.

How does this compare to type certified craft? The venerable Cessna 172 has a ratio of 1:27 (43,000+ produced with 1,565 fatal events) and the Cessna 150 has a ratio of 1:20 (23,949 produced with 1,171 fatal events). Presumably, this is a result of the greater number of cumulative hours flown by these airframes and the extensive use of Cessna in flight instruction.

So, what happens when we try avoid the skewing effects of fleet age and flight instruction? Cirrus Design, producing aircraft since only 1999 (shorter than Van's and about as long as Sonex), has 3857 registrations with the FAA and an NTSB count of 67 for a ratio of 1:59, close to that of the RV's but well below that of Sonex. They even include a parachute on every plane they sell.

When non-fatal events are included, Van's drops to 1:15, Sonex to 1:21, Cirrus to 1:23, the Cessna 172 to 1:4, and the Cessna 150 to an unbelievable 1:2 (better expressed as 42.6%). In part, these numbers may be biased against the type-certified aircraft as their stricter maintenance requirements make it infeasible to hide minor incidents (let alone the considerable use of Cessnas in flight instruction).

Barring a severe misreading of the reports on my part, I don't see that type-certified aircraft illustrate a considerably improved safety record over the home-builder. I can comfortably accept that the pilot is significantly more important on safety than the origin of the plane. So, can someone point me to the nearest transition instructor?

Footnotes

Requested from the NTSB by Model "RV" on 21 June 2011. All queries listed on this page were made from the NTSB or FAA database on 21 June 2011. The total number of incidents over the history of the aircraft are considered.
A query was completed on the FAA's Make/Model Inquiry form on 21 June 2011. Being kitplanes, the ‘Manufacturer’ is listed as the builder and the model does not follow any consistent naming pattern. The query was made on ‘RV’ and then non-RV's were filtered out by hand. Ambiguities were counted to the benefit of Van's.
Includes the 344 Sonex, 5 Sonex/Waiex, and 48 Waiex registrations.

Into the Wild Blue Yonder

2011-07-12T22:43:00-04:00

A couple weeks ago, I tried to figure out how much a personal spaceship (think science fiction) would expect to cost (in today's dollars). I made the following postulate: whatever an airplane costs over a car, a spaceship would cost that much over an airplane. I then set upon the Great Oracle [1] to figure out what the least expensive new airplane would cost. Checking the traditional General Aviation channels, such as the venerable Cessna, I find that a new Cessna Skycatcher is roughly $125k [2]. Assuming a new car to be roughly $10k, that's going to be roughly $1M for a cheap spacecraft. Both are well outside the price of what a mortal could spend money on (let alone frivolously), so I was about to let it end there…

…until I remembered 'kitplane.' There are crazies out there who build their own aircraft [3], so how much money does this save you? A lot, apparently. Assuming you are invest to spend the 800-2000 man-hours [4] required, you can launch yourself into the air for the price of a new car. Hell, I can afford that right now, in cash (I've been saving up for the down payment on a house).

People who know me will understand the inevitable: I suffer from a severe case of "oh, shiny!" This has been manifest in things like rapid prototyping, photography, and Teddy Roosevelt reenactment. Yes, one of those is a lie. Now, I spend a lot of my free time ogling kitplane and avionics websites. But before I do anything stupid (like buy an airplane), I've taken a reasonable preventative step: sign up for flight lessons.

Footnotes

The Great Oracle of all that is True and Untrue, sometimes referred to as the Internet.
The Skycatcher is a new airframe specifically targeted at the Light Sport category (unknown to be at the time), so it isn't entirely appropriate for this comparison.
Formally known as Experimental, Amateur-Built on the Special Certificate of Airworthiness.
4.5-11.5 months as a second job (40 hours/week), 1.5-4.8 years at a more leisurely 10 hours/week.

Right for all the Wrong Reasons

2011-04-25T23:55:10-04:00

Maintaining a blog is hard…or, at least, one that someone would have an infinitesimal interest in reading. Unless you are a naturally insightful individual (e.g. legitimate philosopher), worthwhile content will either be a subject in which you have domain knowledge (such my article on Signal Processing) or one where you are willing to spend considerable time on research (such as my initial trip through photographic optics). Add to this my lack of a legitimate CMS and it isn't hard to see why I've become delinquent. So, in the interest of becoming a bit more active, let me write something I hope will be moderately insightful.

At work, I had a conversation with “new guy” a few weeks ago regarding his performance on a graduate school exam. He was upset in a couple cases for being docked points despite having the correct answer. As I pointed out, the mathematics of his approach was flawed. The correctness of his solution was pure chance, so it unreasonable to expect full credit. When it comes to engineering, it is important to be correct for the right reasons: luck cannot be relied upon in the field nor does it convince your customer.

This episode returned to my memory when reading The Science of Why We Don't Believe Science on Mother Jones [1]. The underlying explanation of why individuals reject scientific consensus is neither controversial nor surprising; however, the author makes the same basic error as “new guy” when attempting to apply this new-found knowledge to real world situations. He concentrates too much on people who don't agree with the consensus (the “wrong” answer) and gives a free pass to people who agree with the consensus (the “right” answer). Expanding their example of climate change, many people on the left agree with the consensus not because they are more “logical,” but because the presentation is compatible with their ideology. The same group of people will reach for all kinds of crackpot theories when it comes to specific, inconvenient cases of vaccination [2], economics [3], or alternative energy [4].

The problem with complicated subjects like climate change is that the vast majority of people will have nothing more than the most cursory understanding. This means that most of us (myself included) rely on the opinions of experts: ethos, not logos. Many physical sciences have the advantage of experimental validation. We accept quantum mechanics as valid because of its experimental history and the modern technology derived from this knowledge (e.g. microchips and lasers); however, no such experimental validation is possible with climate sciences. At best, they can make predictions while we wait to see how close reality matches their guesses. Unfortunately, activist members of the climate community have a record of producing dire predictions which fail to materialize, accomplishing little beyond undermining trust in their conclusions.

Without the ability to verify their science in a laboratory, the emphasis must be the integrity of the scientific process. The presentation of the science must be so unequivocal that informed criticism is impossible. While the denier's obsession with the climategate e-mails as proof of a global green conspiracy are misguided, the e-mails clearly describe an academic monoculture^[5] lacking the necessary scientific skepticism and exhibiting undue hostility to external criticism. Capable individuals, such as Steve McIntyre, are actively obstructed from obtaining the necessary data and publishing the results of his study. Worse yet, the original message from Mother Jones comes true when easily dispelled excuses are accepted as fact by the activist community [6].

What I find most objectionable in the entire discourse is not that people believe differently than I or even that people hold such beliefs strongly and blindly. Such things are expected, even necessary, for reasoned debate. I take issue with the people who condemn their opponents as ideologues while remaining ignorant of their own hypocrisy. An extraordinary sacrifice is being demanded of the world and it's only natural to demand extraordinary proof in return. The apocalyptic hyperbole, conspiracy theories, straw man arguments, and ideological battles [7] characterizing both sides of the debate serve no purpose other than making sure we make no progress on what may be an essential human problem.

I thought this would be a simple, quick essay as I stand here after over ten hours of writing…

Footnotes

As a moderate libertarian, Mother Jones is not in my standard reading list. This is a link I found through a third party.
The unsubstantiated link between autism and vaccination was another example referenced in the Mother Jones article.
Liberals have a tendency to reject basic principles of economics when it comes to government interference in markets, such as price controls and the economic impacts of taxation.
A conservation group in the UK found that wind farms averaged 30% of their rated capacity. Over half the time, they are operating below 20% and a third of the time they are operating below 10% [BBC; Accessed 25 Apr 2011].
The much-praised peer review process only functions when there is enough diversity of thought within the community such that the members are willing to challenge each other's conclusions. How many climate scientists lack a strong sense of environmental responsibility?
“Harassment” by FOIA requests is commonly cited as an excuse for the CRU's behavior, even though it is not supported by the evidence [Steve McIntyre; Access 25 Apr 2011].
Proof of the climate changing is not sufficient as few reasoned skeptics reject such observations. Most reasoned skeptics even accept the possibility of a human contribution to such change. The primary point of contention is the basis for apocalyptic predictions requiring a drastic human response.

If You Want It Done Right, Remove the Automation…

2011-04-25T23:55:10-04:00

In the most recent article, I chronicled my experience in attempting to coerce a Thing-O-Matic into working. Since then, I have accomplished that goal, going so far as making complete sets of Mendel and Prusa Mendel parts. For this most recent contribution to The Great Oracle of All That Is True and Untrue [1], allow me to share the secrets to my current level of success.

Starting with a recap, the problem I was facing is that prints would inevitably fail upon reaching a certain height: An inch or so beyond the build surface, the extrusion head would inevitably grab the top of the object and chaos would ensue. When the endless fiddling with Skeinforge accomplished nothing except trading this problem for others, I sought consultation with the Oracle. As it appears the Automated Build Platform (ABP) is the source of many troubles, I disassembled the ABP and replaced it with the simpler Heated Build Platform (HBP). Perfect. I was even going raftless with minimal effort.

Both the HBP and ABP are intended to heat the deposited plastic, improving adhesion by minimizing the thermal shrinkage should the plastic be allowed to cool. The ABP improves upon the HBP by adding a conveyor belt above the heated surface. Once a print is completed, the belt can advance to eject the finished part and immediately begin construction of another. However, with added complexity comes added possibility of failure. In this case, the belt (deformed from heat) had too much give.

While I plan to eventually reassemble the ABP, I've learned that it is better to get your machine calibrated on the simpler HBP (or even the ABP sans belt) before burning a lot of plastic wondering what's wrong.

Venus de Milo by Jonathan McGee

Footnotes

More commonly know as the Internet.

Why Do Things Never Work Out of the Box?

2011-04-25T23:55:10-04:00

I've been negligent in the posting of new subject material on my web site. In part, because my roommates conspired to involve me in a gigantic time sink, but also because my rapid prototyper arrived [1]. These devices, also known as 3D printers, take a 3D design and quickly generate a facsimile in plastic, resin, or a dozen other possible materials. This allows the designer to quickly test out a design before spending the money on having it manufactured for real.

While a mainstay in mechanical engineering for years, the considerable price of these machines have left them out of the outs of hobbyists (the cheap ones tend to be multiples of $10,000). That is, until a couple years ago when the RepRap project started up. Started by an academic (of course), this project's goal was to design a rapid prototyper that could be manufactured using a rapid prototyper. Once this goal is accomplished, one generation of machines could manufacture the next at low cost.

By virtue of being overly idealistic, there exist a few major problems. The first, most obvious problem is reminiscent of the chicken and the egg: where do the parts for the first generation of machines come from? While some people have access to commercial rapid prototypers, this is an expensive option for the majority of people. To address this problem, several companies started up using more accessible tooling (such as laser cutters) to manufacture machines of similar design which could be used to manufacture a first generation RepRap (or “bootstrap” the process).

This particular construction project has been my third attempt to get a working prototyper, and the first two attempts both illustrated the second problem: a self-replicating machine can be only as sophisticated as its ability to manufacture. Add to this, the limited engineering experience of the project initiators and unique limitations of each manufacturing method, the first generation of these machines were far from optimal. My first attempt came when a friend purchased a laser-cut derivative of a Darwin (version one RepRap). While we got the machine assembled, it had a habit of rattling itself apart and the extruder (the heart of the machine, responsible for laying down the plastic) refused to work but once.

Realizing the goal of a self-replicating rapid prototyper might be a bit overly optimistic, several companies started up product lines which simply sought to make a low-cost rapid prototyper. The ability to replicate itself was simply not a concern; instead, the focus is on optimizing the design for the manufacturing capabilities available to them. The most well-known manufacturer here would be Makerbot Industries, the source of both my second and most current attempts.

Makerbot's first product, the Cupcake, improved upon the performance of other laser-cut RepRaps, but still suffered many of the same problems. Most critically, the extruder, while improved from the original Darwin, still suffered significant reliability problems that prevented me from ever successfully manufacturing an item. Their newest generation of extruder, included with my recently purchased Thing-O-Matic, appears to largely resolved the problem. So, here's my assembled machine:

Thing-O-Matic by Jonathan McGee

Before people start asking me to make random things, take notice of my choice of words: assembled, not finished. After the device was fully assembled and happily throwing the stage to and fro, extruding plastic and the like, I tried to print a shot glass to celebrate my success...failure. The print can't make it much more than an inch vertically before the extruder head embeds itself into the top of the object and otherwise begins misbehaving.

The instructions from Makerbot have you use a default configuration for skeinforge (the program that converts the 3D drawing into machine commands). What I hoped would be an acceptable starting point turned out to be non-workable. So for now, I need to figure out what combination of settings will resolve my problem.

Given the primary symptom (the print head catching on the building object beyond a certain height), it appears that more plastic is being extruded than the default calibration expects. This is unexpected, given that MakerBot's current extruder runs on a stepper motor, so the extrusion rate should be the same for everyone. So far, I've been able to get a tower to successfully complete by increasing the measured layer height, but not without side effects. I'll follow this post up later once I have a better idea of what's going on and how to resolve it.

Footnotes

The actual reason would be that my attention span is shorter than the lifespan of a mayfly.