Observing boundary layer evolution with the UW ultralight

With winter weather looming and my sabbatical break from teaching nearing an end, I watched carefully last week for conditions that would allow us to make at least one more flight with the Zigolo before the end of the year.  Among other things, I wanted to try to use up the remaining fuel so that the carburetor and fuel lines wouldn’t gunk up  from standing idle for up to several months.  But I also wanted to demonstrate the ability to make meaningful meteorological measurements that could not be as easily achieved with an unmanned aerial vehicle (UAV, or drone), conventional weather balloon, or conventional manned airplane.

The most important considerations in planning a flight were wind and precipitation.  With the Zigolo’s unenclosed cockpit, we certainly didn’t want to contend with rain or snow.  More importantly, the aircraft does not have the speed and mass required to handle stronger winds, especially when those winds are gusty.  And even apart from the fundamental limits on the Zigolo’s airworthiness in strong wind, I do not have enough experience flying it yet to risk more than a mild breeze.

So I considered what kind of interesting meteorology could be done on a calm day with no threat of precipitation and without the need to travel a greater distance from the airport.  My previous flight to 1,100 ft. AGL and the temperature profile we captured that morning provided the inspiration:  instead of measuring a single static profile, why not observe the evolution of the near-surface boundary layer in response to the first warming rays of the sun?  So I decided to aim for a flight at or soon after sunrise.

The forecast for Wednesday, November 29, seemed promising.  Skies would be clear, wind would be light up to several thousand feet, and we could expect a strong radiation inversion to form overnight and then dissipate soon after the sun came up. 

When Jonathan and I arrived at the Verona Air Park a little before 7AM, it was almost too perfect.  In particular, the temperature at the surface had dropped well below freezing.  Despite having multiple layers of clothing on, I wasn’t certain how adequate those layers would be when exposed to the 35+ mph relative wind.  Nevetheless, it seemed worth a try.

Technical glitches kept me from getting off the ground as quickly as I had hoped.  First, my goggles – and my glasses inside the goggles – fogged up so badly as to effectively blind me, and I ended up tossing them aside and vowing to fly without goggles.  Certainly I wasn’t worried about hitting any bugs, but I knew my eyes would probably water like mad in the cold wind.

The second technical glitch entailed the freezing up of the device I had strapped to my knee to monitor position and altitude.  Apparently the improvised clip I made for it had pressed on the factory reset button.  So it too was put aside.  Instead I would rely on Jonathan informing me of my altitude via radio, based on the telemetry data he would receive on his laptop.

As before, we strapped a Kestrel 5500 Weather Meter to the “nose” of the open-frame fuselage and set it to log temperature, pressure, humidity, and wind data at 20 second intervals.  The recorded wind, of course, would not be the actual wind but rather the airspeed of the plane

I ended up finally getting off the ground at 7:55AM, 50 minutes after official sunrise. I ascended steadily from the airport in a broad spiral; my climb rate was about 300 feet per minute.  The combination of the surface temperature of less than –5 C  (21 F) and the relative wind of 50 km/hr (31 mph) yielded a wind chill of –16 C (2 F).  I could tell that I wouldn’t last too long if it stayed that cold for any length of time.

Fortunately, although I could not view any measurements in flight, I could certainly feel the air getting a bit warmer as I ascended to a few hundred feet above ground level.  From there, the temperature declined again as I approached 1000′ AGL.  Above that altitude, I felt myself climbing through yet another pronounced temperature inversion.  By the time I reached my target altitude of 2000′, I was experiencing the warmest temperatures (relatively speaking) of any point along the ascent. No, temperature does not always decrease with altitude!

Upon reaching 2,000′ AGL, I immediately began descending again, and the air grew colder again.  I approached the airport from the east at about 500′ AGL and made a low overpass over an open pond, hoping to later see a humidity signature from that pass (I didn’t!).  Directly over the runway just a few feet above the grass, I added full power and began climbing again.


Over the course of the nearly hour-long flight, I made a total of three ascents to 2,000′ and back again, each time returning to within a few feet of the grass before ascending again.  By the third cycle, I realized that I could not continue – the cold wind was filtering through my layers of clothing, I was starting to shiver, and my hands were becoming numb inside my gloves.  

I landed, taxied back to the hangar, and shivered uncontrollably with honest-to-goodness hypothermia.  Jonathan let me sit in his car with the engine running to warm up; it was 15 minutes before the shivering fully subsided.

Lesson learned:  Get better cold weather clothes!  In particular, look into the kind of windproof winter clothing and gloves used by snowmobilers and motorcyclists.  Also, several experienced ultralight pilots suggested chemical warmers inside the gloves and boots.

In the end, the temporary discomfort was worth it: we obtained three perfect profiles of temperature and dewpoint (actually six if you count ascent and descent separately)  spaced about 14 minutes apart.  The plot below shows exactly the evolution we had hoped to see:  the strong surface inversion during the first ascent was already markedly weakened by solar heating during the second ascent (a mere 13 minutes later), and by the third ascent, it was gone altogether.   Also, convection from the surface appeared to erode the base of the higher inversion, which had started out near 320 m (1,050 ft) and then lifted to 420 m (1,375 ft) in just 15 minutes.

Temperature (solid curves) and dewpoint (dashed curves) from three ascents by the UW-Madison Ultralight on 29 November 2017. Displayed altitude is relative to ground level at the Verona Air Park (W19).

Key takeaway points:
  • To measure the same profiles, a traditional radiosonde system would have required three single-use balloon systems at a cost of $250 each, including instrument and helium, for a total approaching $800.  The UW-Madison ultralight, however, measured these profiles using less than two gallons of premium gasoline treated with synthetic oil, for a total cost of less than $10.  Also, it is normally not possible to operate conventional radiosondes so close to each other in time because of radio interference; instead, one would need to allocate three separate frequencies and three separate ground receiving stations.
  • A drone could have easily measured these profiles up to an altitude of 400′ AGL, but beyond that altitude would have required a waiver from the FAA.  The waiver process takes at least 90 days, and many waiver applications are denied.  In short, the full 2,000′ profiles we see here would have been possible with only significant advance planning and a successful waiver application.
  • Conventional manned aircraft could theoretically have also observed the profiles, but their cost of operation is typically far greater, especially when factoring ferrying time, intrument integration, etc., into the total cost.  Also, it is generally less safe for large conventional aircraft to descend to within a few feet of ground level unless the plane is actually landing.
At this point, the fuel tank, carburetor, and lines have been emptied, corrosion protection has been applied to the engine, and there is no urgency to fly again before spring.  Nevertheless, I’ll continue to watch the forecast, and if the opportunity presents itself, we’ll do our best to take advantage of it.

Testing and outfitting the UW ultralight – the view from above

In my previous post, I described our second off-airport test flight. Now I’m belatedly posting video taken from a GoPro attached to my helmet. Unfortunately, the camera was angled slightly too far downward so the horizon is rarely visible; also it quit recording halfway through the flight, just when things were getting interesting again!


Testing and outfitting the UW ultralight – October 26, 2017

This morning, I flew the ultralight for what is probably the last time in 2017 unless we have exceptional weather in December (I will be gone most of November). I took it to 1,100 ft. AGL – about four times as high as my previous flight, loitered for a short time, and landed again after only 8 minutes total time in the air. 

I had planned to stay up much longer and to try to burn through most of the rest of the fuel (about 3.5 gallons) before putting the plane away for the winter.  But the winds aloft were quite strong, and for a time, my ground speed (as measured with GPS) when heading directly into the southerly wind dropped to only 2 knots!  Given the 30–35 kt airspeed of the Zigolo,  the winds between ~600 and 1,100 feet must have been pushing 30 kts or so. 

In fact, I spent much of my time at altitude hovering almost motionless over a particular farm house about a mile southeast of the air park!   I realized that if I had headed downwind from the airport after takeoff instead of upwind, I might have had difficulty returning!  Something to think about when planning future science flights.

In addition, there was pretty obvious wind shear accompanied by gusty turbulence between 600 and 1,000 ft. AGL.  One more reason not to push my luck on this particular day, especially given the likelihood that that gustiness would migrate downward toward the surface over the coming hour as the rising sun heated the surface and eliminated the nocturnal temperature inversion that was, up until now, decoupling the stagnant surface air from the stronger winds aloft.

Jonathan had fastened a Kestrel 5500 weather logger to the front of the plane, and the figures here and below (click to enlarge) show the measured temperature and dewpoint profiles.

The temperature profiles tracked surprisingly well between ascent and descent, suggesting temperature sensor response time and repeatability aren’t problems for the Kestrel. We clearly see what was still left of the noctural inversion, which had probably extended to the surface a couple hours earlier when the surface air temperature was closer to 2-3 degrees. If I had been able to get off the ground a hour or so earlier and do multiple profiles, it would have been possible to track the evolution (this would be a great reason for a sunrise takeoff after a calm, clear, dry night!). 

The dewpoint was a full degree lower on an ascent than on descent, only minutes apart. At first I thought of sensor lag but then realized that the difference would be flipped (moister on ascent than on descent).  Then it occurred to me that my descent took me over a 500′-diameter pond that began about 800′ northeast of the approach end of the runway, and that could well have added enough moisture to the overlying air to increase the dewpoint by a degree (at that temperature, only about 0.1 g/kg of added moisture would be required).  [Edit: It’s also possible the humidity sensor simply drifted by that amount over 8 minutes.]

Flight track as viewed toward the southeast. The departure/ascent track begins on the right.

For future flights, it could be interesting to fly a low circle around a similar-sized pond under similar conditions (cool and dry) and see whether we can detected a plume of humidity on the downwind side.  

Total materials cost for this boundary layer sounding was about half a gallon of fuel plus oil, or maybe $2.  For comparison, a conventional radiosonde launch typically costs at least $200 in expendables.

PS: I captured GoPro video from the above flight and will post it at my next opportunity.

Building an airplane – a photo/video montage

I put together a quick-and-dirty montage of photos and videos covering the period from the arrival and unpacking of the Zigolo kit to its first off-airport flight on October 8. Total viewing time is about 13 minutes.



At the close of the building phase of this project, I’d like to thank Craig Oswald for his crucial assistance during the intensive first week of the build and Jonathan Thom for his substantial ongoing role not only in the build itself but also in subsequent operations. The quality and speed of completion of this phase would not have been possible without their involvement.

And of course we would still be scratching our heads over a large pile of random parts if Chip Erwin (Aeromarine LSA), through his Builder Assist program, hadn’t flown twice to Madison to work with us for several intense days each on crucial parts of the build.  His involvement also gave me considerably greater confidence that the Zigolo wouldn’t fold up and crash the moment the wheels lifted off the runway (or worse, five minutes later!).

I’d also like to thank Bob Paulos and Gary Anderson once again for providing the perfect space at the Physical Sciences Laboratory in Stoughton. I have no idea where and how we would have built the airplane without it.

Sue Foldy in AOS patiently handled all of my orders for exotic materials and tools, and Sonja Johnson continues to help me navigate the reimbursement process.

Throughout 2016, Ben Griffiths in the UW-Madison Legal Affairs office, Hartley Murray in Purchasing Services, and Jim Bogan in Transportation Services were critical in navigating the university and state bureaucracies,  aided by helpful individuals in the Wisconsin Department of Administration (e.g., Cheryl Edgington), and culminating with a signed memo from Gov. Scott Walker authorizing the University to acquire its own airplane.

Last but not least, none of this would have been possible without the Ned P. Smith Professorships that were awarded to Ankur Desai and me; these collectively provided one half of the flexible funds that allowed us (and me in particular) to embark on this exotic quest, with the remaining half divided equally between Prof. Tristan L’Ecuyer’s flexible research funds and from our partners in the Space Science and Engineering Center, especially Hank Revercomb and Fred Best.

And now, on to the testing and operational phases!

Building an airplane – leaving the nest, October 8, 2017

Unbelievably, it’s been almost six weeks since I made those first crow hops down the grass strip at the Verona Air Park (C29). First, I was away in Arizona for four weeks as a part of a sabbatical stay at Embry Riddle Aeronautical University in Prescott, where I’m working with colleagues in the Applied Meteorology program there to brainstorm meteorological applications of UAVs.  After my return, I was disappointed to learn that we still had not received the rocket for our BRS (ballistic recovery system) due to a licensing glitch with the Federal Bureau of Alcohol, Tobacco, Firearms and Explosives.

This meant that I would either have to continue waiting to fly the Zigolo to any significant altitude, or I would have to take my chances on the airplane we just built hanging together through whatever turbulence or moderately stressful maneuvers I might expose it to. 

Not that there was any particular reason to expect a problem, but we had paid for the BRS, and the first couple of hours of flight of an amateur-built aircraft were when the odds of actually needing one  seemed highest.  Also, as a novice ultralight pilot, I knew I’d feel less vulnerable at altitude in the completely exposed seat if I had that red handle dangling reassuringly above me.

On Thursday, October 5, I got the message from BRS that the license logjam had finally been broken, and they said they could ship it on Friday.  But it would then arrive in the middle of the following week, and I would only a have a very short time before I needed to leave town again.  What if the weather didn’t cooperate?  I might not get to fly until Spring!

In the end, I spontaneously drove to St. Paul on Thursday evening (four hours one way), stayed in a cheap motel, picked up the rocket in person the following morning at 8am, and then was back in Madison by early afternoon Friday. It turned out to be the right decision, because I was able to install the rocket on Saturday, and Sunday was forecast to be the last truly nice-weather day for some time to come.

On my way home from the air strip on Saturday, I picked up a good supply of Castrol TTS two-stroke oil and also filled the five gallon canister with premium gas, no ethanol.

On Sunday, I stopped at a bike shop to pick up a convex mirror that I could mount so as to allow me to see the gas tank behind the pilot seat.  After I got to the hangar, I carefully measured out one gallon at a time of 40:1 gas-oil mix and added it to the tank, and I marked the level of each added gallon with a Sharpie (I had propped up the tail so that the tank would more or less be in the position expected during level flight).  I now had a crude fuel gauge that was, most likely, more accurate than any Cessna fuel gauge I ever worked with.

Shortly thereafter, Jonathan arrived, and we both went over the BRS installation and the airplane itself in the most thorough preflight I have ever given any airplane.  Bolt by bolt, cable by cable, strut by strut, we confirmed that every one of dozens and dozens of critical joints seemed secure.  I was particularly concerned about the control cable connections from the joystick to the elevator – if there was one control I did not want to lose in flight, it was the elevator!

Next, I improvised a temporary cardboard holder for the ICom AC-14 handheld aviation transceiver, using cable ties and electrical tape to secure it to the forward tube that sloped upward and rearward in front of my seat.  I used the velcro strap to attach the push-to-talk button to the same tube, just above my head where I could easily reach it.  In the meantime, Jonathan secured a Kestrel 5500 weather logger to the tip of the fuselage so we could record temperature, humidity, and pressure (it could technically also record wind, but the main wind it would see was the aircraft’s own airspeed, which was not meteorologically interesting).

At around 5pm, after it felt like the breeze was dying down, we wheeled the Zigolo out onto the grass,  and I primed and started the engine.  I plugged my David Clark headset into the radio, put on my motorcycle goggles and my new Comtronics Ultra-Pro helmet, and strapped myself in. Jonathan set up near the runway so that he could take photos and video.  He also had the other ICom radio so that I could talk to him on the multicom frequency (122.9 MHz).  We don’t yet have a ground station license, so he could only listen.

It had been weeks since I last taxied the plane, let alone flown it.  So I started cautiously, taxiing up and down the runway, getting a feel for the handling.  I then repeated my crow-hopping exercise of several weeks ago, accelerating to takeoff speed and then flying 10-20 feet above the runway until the far end loomed, and I put it down again.

After the first crow hop, I touched down fine, but suddenly the breeze caught the tail and swung it one side, and I found myself careening off the side of the runway through some tall grass before I regained control and came to a stop.  Lesson learned – I realized I needed to remember my training in an Aeronca Champ a year ago and pull the stick sharply back after touchdown to hold the steerable tailwheel firmly onto the ground for better control.   The Zigolo is, after all, a tailwheel airplane, unlike the Cessnas I’m used to flying.

The remaining crow hops went well, and I persuaded myself that once I left the field for real, I would manage to land again in one piece.

It proved surprisingly difficult to commit to taking off and flying away from the airport. I knew that once I did that, the safety net would be gone; I couldn’t just put it right back down on the runway again, because I’d be over trees for a bit.  And I’d still be at too low of an altitude for the BRS to do any good if I suddenly found myself in a pickle.  In a Cessna, I would have given little thought to these things, because I know that Cessnas are built and maintained by people who know what they’re doing.  Here, I was contemplating flying to several hundred feet altitude or more in a relatively flimsy flying machine that I, a novice,  had helped assemble from a crate full of random parts just weeks ago.

Google Earth view of the Verona Air Park (W19). The UW ultralight is kept in the white hangar at the northeast end of the row of three buildings.


The feeling I had was remarkably like that of standing on the edge of a pool full of chilly water and indefinitely postponing that moment when you take the plunge – both figuratively and literally –  into the unknown.

Finally, I made a somewhat garbled announcement over the radio to let other pilots who might be in the area know my intentions.  I didn’t know what to call myself.  Zigolo N13UW?  I didn’t actually have the N-number assignment yet (I was still flying under Part 103).  UW Ultralight? UW Zigolo?  I don’t know anymore what I actually said, and it probably didn’t matter anyway, because there were no other planes far and wide that evening.

I squeezed the throttle trigger on the stick, and within around 100 feet, I was airborne.  This time, instead of letting up on the power and cruising level just above the runway, I kept full power and held the nose up, watching the Hall airspeed indicator carefully to make sure I kept to around 30-35 mph and didn’t let the speed drop toward the stall speed of 25 mph.  

I had also told myself on the ground that I would hold onto the red BRS handle with my left hand during climbout, but I completely forgot about that once I was on my takeoff roll.

Very quickly, I was over the tall trees, and I worried about where I would go if I suddenly lost power.  But shortly thereafter, I was back over open fields where the corn had already been harvested, and I allowed myself to breathe again.  

The view of the farms and fields from above was stunning – the sun was on the verge of going down, and everything was illuminated by a deep golden light broken by long shadows from the trees.  I continued climbing until I guessed that I was at something approximating pattern altitude – I didn’t have an altimeter – and I made a gentle 180 degree turn to return to the air strip, and I descend to land in the opposite direction from my takeoff (there was no wind component along the runway).

Whew!  That went okay!  Now let’s try making an actual circuit around the pattern.   I returned to the southwest end of the runway, just below the elevated Highway 151, and took off a second time on runway 04, lifting off quickly and establishing a satisfactory climb over the trees on the far end.   Runway 04 has right traffic, so at a comfortable altitude, I turned crosswind to the southeast and then downwind to the southwest.  I was nervous about descending too low over the busy freeway, so I turned base early and made an uneventful midfield landing.

By this time, the sun was officially setting, and I decided to call it a day. 

Postscript:  I later recovered the meteorological data from the Kestrel and determined from the pressure record that I had only climbed to about 270 feet above ground level.  It had felt so much higher in that open cockpit!  I resolved to buy an actual digital altimeter so I wouldn’t have to guess about that anymore.

Building an airplane – MY first flight, August 30, 2017

Recap: Yesterday, within minutes of tightening down the last bolts and setting the last rivets, Chip Erwin took the newly built Zigolo MG12 out of the hangar at Verona Air Park (W19), started it up, taxied it to the grass strip and did a few crow hops to verify that everything was in working order. It was a thrill to see it fly, however briefly, but the major milestone that was still missing was for me to fly it myself.

I wasn’t at all sure when that would happen. Not only did conditions have to be just right – which basically means calm winds (for now), but I also had to have confidence that I could safely get it down on the ground again.  As they say, takeoffs are optional, landings are mandatory.  And landings are by far the most demanding of flying skill, relative to other routine maneuvers.

While I have made many hundreds of landings in Cessna 152s and 172s, ultralights are fundamentally different in multiple ways.  First, they are so light and draggy that they carry almost no momentum.  If you get the nose too high with too little power, you quickly bleed off airspeed until you’re not flying anymore, which is a bad thing if you happen to still be 10 feet or more above the ground. Second, because they have such large wings in comparison to their weight, they get tossed around by every gust in a way that heavier aircraft do not.  Third, your sight picture from the “cockpit” is completely different; you’re not looking over the nose of the aircraft and using the horizon as a reference to judge your flare; instead, you’re watching for grass to come up just below your feet.

In short, I didn’t want to take anything for granted.  My first flight would be made only under optimal conditions, and I told myself I  would approach it with extreme caution.

So I was pleasantly surprised when, at about 5:00pm today, I noticed that the breeze had settled down to almost perfectly calm, and I decided that at the very least, I could practice taxiing the Zigolo on the ground.   Among other things, this would allow me to continue breaking in the engine, a process that is not complete until it has burned through a full tank of fuel (about 5 gallons).  I would leave open for now the question of whether any actual flying would take place.

I drove out to the air strip, which is about 20 minutes from campus, opened the hangar door, and wheeled the Zigolo out onto the grass.  After checking that no nuts or other components had mysteriously worked themselves loose overnight, I fastened myself in with the harness, started the engine and let it idle until the cylinder head temperature was above 100 degrees Celsius.

The Zigolo has an unusual throttle arrangement (for an airplane) in that you squeeze and release a trigger on the joystick to add or reduce power. The engine responds immediately, very similar to squeezing the throttle on a chain saw. I carefully increased the engine RPM until the Zigolo began rolling through the fairly deep, damp grass and then used the rudder pedals to steer.  The aircraft was pretty responsive to rudder, especially if I goosed the engine to let the prop wash help move the tail around.

I taxied carefully to the southwest end of the grass strip and used the technique Chip explained to me for making the sharpest possible turn: hold the stick fully forward to take weight off the tail while adding a burst of power and giving full rudder deflection.  There are no brakes on the Zigolo, differential or otherwise, and the tailwheel doesn’t do much in the wet grass, so the rudder is all you’ve got.

Once turned around, I aimed the plane down the grass runway and cautiously added power.  I taxied slowly at first getting a feel for the rudder and for the handling on the somewhat uneven and, in some places, soggy sod. Reaching the end of the 1900 foot runway, I turned around and repeated the process in the other direction.  As I gained confidence with each pass, I increased the speed to the point that I could lift the tail with forward stick pressure and roll along on just the main wheels at just below the stall speed of about 25 mph, holding direction with the rudder.

Pretty soon I got comfortable with the handling on the ground and experimented with giving full throttle, backing off just as the plane seemed to want to fly.  I was holding the stick forward, so it wouldn’t have flown even well above stall speed unless I pulled back some, and I made numerous runs in both directions at what felt like a pretty good clip, bouncing over hummocks and splashing through soggy areas (there are now some mud splatters on the lift struts).

After at least an hour of this confidence-building exercise, I decided to accelerate to 50 km/hr according to the metric-calibrated Hall airspeed indicator that the Zigolo is equipped with; this translates to 31 mph, which is more than enough speed to get airborne (and about as fast as my bicycle gets going down Seminole Highway northbound from where it crosses the Beltline)! 

The first time around, I kept it on the wheels just to feel how it handled on the ground at that speed.   The second time, I eased back the stick, and the Zigolo immediately lifted off the runway. I reduced power and settled right back down again, very smoothly and with no obvious tendency to yaw.  My first flight in the Zigolo! It probably reached an altitude of 1 foot and covered a distance of 30 feet, but it still qualified as the big milestone in the nearly two-year process of acquiring and building the Zigolo.

After a few more very tentative crow hops, I began to leave the power in longer and travel further down the runway before settling down again.  I also allowed the plane to climb higher. In all cases, I was extremely attentive to airspeed.  Keeping it near 50 km/hr using pitch while using power to control whether I climbed or descended at that speed seemed to work well.  In no case did I make a “bad” landing; as long as I kept the airspeed up around 45-50 km/hr and flew it back to the ground before leveling off and reducing power, it always touched down gently and under control, and it rolled to a stop fairly quickly on the tall grass.

In all, I probably made a dozen honest-to-goodness flights, the last few of which extended most of the length of the runway and reached the modest altitude of 30-40 feet before touching down again.  Had I gone higher, I would not necessarily have been able to land again without leaving the airport and flying out over forested areas, which I was not ready to do.

Around this time, I noticed dark clouds on the horizon about 10 miles to the north and simultaneously  realized that wind had abruptly come up – the outflow from what turned out to be an approaching (but still relatively distant) thunderstorm.  This was not part of the plan!  My final landing was made with a significant buffeting crosswind, and I had to fight a tendency for the wind to pick up a wing and push me off course as I tried to line up with the runway. The fact that the aileron action is a bit friction-y right now didn’t help; I’ll have to do something to loosen that up, maybe some lubricant in the hinges.

  I got safely onto the ground but had to work hard to maintain directional control and to keep the wings level as I taxied back to the hangar. I was worried that if a stronger gust came along, there would be nothing I could do to stop the plane from tipping to the point that a wing tip would hit the grass (but better tall grass than asphalt or gravel)!

Fortunately, I made it to shelter without incident, but there was an important and timely lesson in this:  no matter how benign the wind for takeoff, be aware of the possibility that everything can change with little or no warning, especially if there is convective activity anywhere within a 20 mile radius. Weather that poses comparatively little risk for conventional airplanes, like the Cessna 172 that I regularly fly, can evidently be significantly more problematic for an ultralight.  It’s one thing to hear that message from someone else; it’s another to experience it firsthand.

In the future, I plan to always have a second person present at the airport whose job includes monitoring the area weather carefully and being ready to signal to me either visually or by two-way radio when its advisable to come down, since the signs of changing wind in particular aren’t necessarily obvious when you’re in the air.

All in all, an exhilarating and educational first experience with flying the Zigolo.  This calls for a glass of champagne.

Building an airplane – first flight, August 29, 2017

Two months ago today, the Zigolo MG12 kit arrived as a 200+ lb. crate of parts (plus a bundle of long aluminum tubes) packing hundreds of seemingly random plates, bolts, rivets, cables, brackets, and fittings.  Today, with Chip Erwin at the controls, it flew for the first time.

Chip Erwin “crow hops” the newly completed Zigolo MG12 down the grass runway at the Verona Air Park (W19).

Last Thursday, Chip had found an unexpected break in his schedule that allowed him to come to Madison on very short notice for the final push. The five days since then have been intense: Chip, Jonathan, and I worked together through Thursday and Friday, Chip and I worked all day Saturday, I worked alone on Sunday from noon to midnight covering and painting the wings, and yesterday Chip and I madly worked from 8am to 7pm to complete everything on our list in time to cart the aircraft components out to our rented hangar at the Verona Air Park (W19) for final assembly.

It turned out that the bed of the University Fleet’s Ford F350 pickup was too short to be usable for our purposes, so we rushed out to find a 20′ flatbed utility trailer we could rent for the afternoon to transport everything in four trips between Stoughton and Verona. Paint was literally drying on one wing while we were driving the other one down the back roads.

This morning, the push continued starting at 7am at the Verona hangar; we needed to finish a number of items that had had to wait for resupply of the appropriate rivets; we also installed the ailerons, and Chip rigged the aileron cables. At 10:30am, we gave the plane a thorough preflight, checking for bolts that might still be loose and rivets that might have been missed. 

With everything looking good, Chip grabbed the Zigolo without ceremony and pushed it out onto the grass and started the engine. Wind was light and more or less straight down the grass runway.  Chip throttled up from mid-field and was airborne fairly quickly.  He stayed within a few feet of the grass and touched down again after maybe 100 yards. Turning sharply at the end of the runway — using a burst of power to bring the rudder into play — he repeated the crow hops in the downwind direction.

After a couple more passes, Chip taxied back to the hangar and declared himself extremely happy with the flight characteristics.  Unfortunately, that was it for testing this morning,  as Chip had already generously extended his stay through this morning after we were unable to complete all the necessary work in time for a test flight yesterday, and now he was facing a 7-hour road trip back to Ohio.

The next step will be more me to repeat the crow hops on a day with little or no wind, possibly as early as this evening or tomorrow.  This will allow me to get familiar with the handling during the critical phases of takeoff and landing.  I’ll be low and slow enough (maybe 10 feet at 35 mph) that the risk of serious miscalculation will be low — again, assuming little or no wind.

On Chip’s recommendation, the Zigolo won’t be taken to higher altitude until we have received and installed the BRS rocket-propelled parachute system  (we were supposed to receive it months ago, but a supply chain problem has held up delivery).  After that, I will be guided in part by the recommendations of Facebook friend (and experienced ultralight pilot) Bill Esker, who wrote 

Remember three things while flying lightly loaded wings.
[1] Airspeed is everything. Even more so than GA. When you pull power get that nose down and maintain airspeed.
2. Forward penetration inertia is almost zero. You have no weight!
3. You will be far more subject to turbulence. No big deal, but you will be bounced a bit more.Your takeoffs will be short, your landings will be short!
Enjoy because you now have a “motor-floater” and can go play with the thermals! … Even though you are a GA pilot fly only calm days for the first ten hours. Very light aircraft are a little weird to get used to, and no pulling the power half a mile from the numbers like GA , keep power on until you are 50 yards out! Once you get some circuits under your belt, then do some deadsticks right over the runway at first. You need to know what it feels like to.fly gravity only for your motor. It’s fun, you will enjoy it, but remember airspeed!

Incidentally, I had fully intended to blog more regularly about the build process leading up to this point, but that proved impractical. On the one hand there were sometimes week-long periods of time where nothing much was happening at all with the build, either because we were waiting on parts or because we were tied up with other things. On the other hand, the busy periods were so busy that I not infrequently worked 10–14 hour days at PSL and had no energy left for blogging by the time I got home at midnight! I apologize to those who might have been hoping for blow-by-blow commentary on the specifics of building this plane.  If you’re building a Zigolo kit, I’ll be happy to answer any questions that come up.  I had plenty of questions myself, and Chip’s on-site oversight and assistance saved us countless hours of head scratching and likely mistakes!

PS:  In the excitement and haste of the moment, I failed to get decent photos or any usable video, despite having good equipment with me (most of it still in the car).  There should have been a third person along whose sole job was to capture the moment. Or, in any case, more planning.

Total eclipse 2017

Antje and I pose with our eclipse glasses during the early partial phase of the eclipse.

Yesterday, my wife Antje, older daughter Annika, and I had the once-in-a-lifetime experience of being at the center of the path of the August 21 total eclipse. Our location: the Crab Orchard marina just east of Carbondale, Illinois.

The partial eclipse began at 11:52 AM and progressed slowly. Wearing our eclipse glasses, we could see the growing bite taken out of the sun’s disk, but there were no remarkable changes in the ambient lighting until the degree of coverage was well past 50%.  Only at that point did it start to become noticeable that the light was unnaturally muted relative to normal noon sunlight – but without either the warmer colors or the long shadows of late afternoon. 

(Interestingly, the effect of the reduced light during a partial eclipse is completely undetectable in photographs using automatic exposure metering, because the camera compensates by increasing the shutter time or opening the aperture by a stop or two, and there are no other visual clues in the light that the sun is partially blocked.)

Now, I have to admit, I have seen a number of partial eclipses in my lifetime, and I don’t find them terribly exciting. Unless you view the sun’s disk through a filter, it could easily escape your notice that an eclipse is happening at all.  Indeed, the only real clue might be the crescent shaped shadows that commonly occur where sunlight is filtered through a tree or other small openings.   I would not drive hundreds of miles to catch a glimpse of a partial eclipse.

That said, the progression of the partial eclipse was not without its excitement. We warily monitored the increasingly widespread cumulus congestus clouds that had begun developing throughout our area starting around 11 AM.  These were large and slow-moving, and I worried that one might block out the sun during the critical 2 minutes and 38 seconds of totality. Indeed, as we got to within 10 minutes of the start of the total eclipse, our tension was heightened as one towering cloud edged ever closer to the sun’s disk.  At that point, I sadly reckoned that we had at best a 50/50 chance of directly observing the total eclipse.  Would the 950+ mile round trip have been worth it? But fortune was with us, and the clouds remained more or less clear of the Sun’s disk.

As we got to within a couple of minutes of totality, the reduction in light accelerated rapidly.  Recall that the Moon slides over the Sun’s disk in linear fashion with time, while the eye responds logarithmically to light and thus to each factor-of-two change in the amount of uncovered area of the Sun’s disk.  The first factor-of-two took around 45 minutes, the next factor-of-two required maybe 20 minutes, and each successive factor-of-two required about half as long as the one before (these are crude estimates; the exact math is complicated).  Only on the brink of total eclipse were we able to perceive the rapid change of  ambient light as if someone was working the slider on a dimmer switch.

Through our eclipse filters, we watched as the uncovered portion of the Sun’s disk became an impossibly thin sliver and then, briefly, a residual bright spot known as the “diamond ring” (more on that below).

Seconds later, even that was gone.

As we removed our eclipse glasses, a collective gasp of appreciation went out from all those in our vicinity as we all beheld a sight that I have  never seen adequately capture in either print or photograph. This blog post will be no exception. 

The sky was dark, though not the dark of the midnight sky; more like pre-dawn.  The corona of the sun was suddenly visible in all its other-worldly glory.  This photograph shows the exact same corona we saw, but without capturing the fine fibrous texture or subtle coloration we saw with the naked eye. I think no photograph can.

Looking around us, we saw that the towering cumulus clouds and sky low on the horizon were illuminated with an orange glow – in all directions. It was as if the sun had just set to the north, east, south, and west, and all points in between, all at the same time.

At our location at the center of the path, we had a full two minutes and 38 seconds to appreciate the spectacle of a total eclipse, and I was grateful for every second. For a brief period, it partly disappeared behind a shred of cloud, only to reappear through a gap.  If anything, the proximity of clouds only enhanced the beauty of the scene.

I allowed the sun to begin reemerging from behind the moon without putting on my eclipse glasses again.  I did not look at the sun itself during this time but instead took in the environment around me during this critical transition back to a partial eclipse.  And that’s when the next surprise hit me.

When the “diamond ring” phase occurs, the intense light of the sun comes from an exceedingly small, almost point-like source rather than from the  half-degree diameter disk that we’re used to. Why is this significant?  It’s because the “large” unobscured disk causes shadows to have blurry edges. Look at the shadow of  the leaves of a tree on an ordinary sunny day, and you’ll have a hard time making out the individual leaves; the shadows are all smeared by that half-degree variation in the direction from which the light is coming. 

During that “diamond ring” phase, however, that blurring effect is almost entirely eliminated – shadows are exceptionally sharp and detailed.  A co-worker who also traveled to see the eclipse told me that he was able to discern the shadows of the hairs on his own head.
In fact, the lighting during the “diamond ring” phase reminding me of that cast by someone operating an arc welder during my shipboard days – the light is unnaturally white and the shadows are unnaturally sharp.

But what really drove that observation home for me was the shadows cast by the cloud wisps that passed in front of this point-like source of intense light: I could see astonishingly fine-scale texture in the clouds’ shadows as they moved across the ground before me, even though the clouds casting the shadows were some 2,000–3,000 feet above ground level!  I don’t know why I had never previously seen mention of this truly striking phenomenon.  I did retroactively find this page, which describes “shadow bands,”  which I believe are closely related to what I saw, but the latter apparently result from turbulent inhomogeneities in the cloud-free upper atmosphere.  The thin sliver-like source of sunlight is central to their occurrence as well.

You’ll notice that I have not posted any eclipse photos of my own.  I did not believe I could capture the experience in photographs well enough to compensate for the distraction. If you want to see images professionally captured by NASA from approximately the same vantage point, you can check out this YouTube video.

Rainbow photographed from the car as we headed back north through Illinois on the way to Madison.

Building an airplane – August 1

Jonathan and I essentially finished the wing assemblies today (sorry, no pics yet!) after finally getting a packet of more 3×16 mm rivets (actually 1/8″ x 5/8″ according to the U.S. system). Just a couple of significant loose ends in the wings:
  • We need to chase down some slightly longer 5-mm bolts (not easy to find a wide selection of metric bolts locally) to anchor the internal wire bracing, as a couple of the wires got cut a few millimeters too short.
  • We found that the first ribs on the inboard end collide with the bracing cable as shown below.  We believe that the ribs were placed exactly as specified in the plans; nevertheless, a mere 1/2″ displacement in the outboard direction would have avoided the collision.  Still pondering what to do – the ribs are firmly riveted and epoxied into place so can’t easily be moved; at the same time, we’re loathe to cut into the end bracket to create a notch for the cable to pass through. We’re waiting for advice from Chip.
  • We want to add custom hard mount points near the wingtips before covering so that we can conveniently and securely mount instruments there, if desired.  This customization would be far harder to add after the wings get covered.
  • Oh, and we haven’t actually secured the curved wingtips themselves yet – OK, there’s more left to do than I immediately remembered!
But apart from those things, the lift and tail strut assemblies, and a few other odds and ends on the fuselage that we’ll probably want help from Chip with on his next visit (TBD), we’re done with the bulk of the mechanical assembly phase.

Useful trick: if you have  half-full squeeze bottles (with tapered nozzle) of hardener and epoxy, just mix the two in one of the bottles, and then use it to squirt the epoxy mix directly onto the parts being joined. Much faster and more accurate than applying with a brush, though a brush can be used for spreading and for picking up excess. Regardless of how you do it, be sure to use 30-minute epoxy, otherwise you’ll never get done before it starts to set up!

We’re now looking ahead to covering the control surfaces and horizontal stabilizer (the rudder and vertical stab have already been covered, though not shrunk or painted yet). We’ll cover the wings last, not only because it’s the biggest and most important covering job but also because of the aforementioned customization we’re still thinking about.

For those interested, we’re using Stewart System EkoBond adhesive to attach Dacron fabric which is subsequently shrunk with heat and then coated with Latex exterior primer and gloss finish paints. You can find good videos online of how the covering process works using EkoBond, which makes the covering process almost miraculously easy compared with more traditional adhesives.

In the meantime, there’s also paperwork to fill out and file with the FAA.  We will be registering the plane as an amateur-built experimental and have requested N13UW as the registration number: ‘UW’ for the University of Wisconsin, and ’13’ because it was the first free number available with the UW suffix and also happens to be the floor I work on in my building (no, I’m not superstitious!).   I have also begun making contact with an FAA-designated inspector to sign off on the construction.

Building an airplane – July 21–25

I’ve gotten way behind on my blogging about the airplane build, in part because on the days I’m doing it, I tend to work until late (after 10pm in a couple of cases) and on the days that I’m not, I’m too busy catching up with other essential business, like reading student theses.

The short version is that I worked Friday, Monday, and Tuesday (yesterday) for anywhere from 4 to 10 hours each, with Jonathan helping out for several hours each on Monday and Tuesday.

The completed wing ribs. It took me about a day and a half to make all of these, with the first few taking the most time.

We are now most of the way through the fuselage and wing assemblies, and Jonathan has done most of the engine mount assembly and has started to look at the struts. Our plan is get as far as we can on our own before Chip rejoins us to help wrap up.

The wing assemblies with ribs attached. Metal straps on each end were wrapped around the spars, epoxied, and riveted at the top of the ribs. Half-ribs still need to be attached between the main ribs.




A couple of technical notes for others who might be tackling a Zigolo build:

  • The wing ribs are trickier than they look, at least until you get the hang of them. The pre-drilled holes in the ends of the top and bottom tubes don’t perfectly line up with the pre-drilled holes in the steel end brackets, which means you often have to drill through to get the 3×16 rivets inserted. But those 3 mm diameter rivets are extremely sensitive to the size of the hole on the far end – if it’s even a little too large, the mandrel pulls all the way through to the near side of the bracket before breaking off rather than on the far side where it should. You can tell when this happens, because the rivet “pops” only after the second or third pull rather than on the first. Sometimes a pulled-through mandrel doesn’t make a noticeable difference in the strength of the joint; sometimes it does, because it can break up the aluminum sleeve between the bracket sides. In any case, the way to avoid this problem is to first start with a true 3mm drill bit, or at worst a #31 bit, rather than the more readily available (in the U.S.) 1/8″ bit, which is about 6% too large. Then, be very careful not to widen the hole more than necessary once you get through to the far side of the bracket.  It took me several ribs and a few ruined rivets that had to be drilled out again before I figured out the right technique.

  • Another “gotcha” with the wing ribs:  there are sheet metal straps that get riveted to the bottom of the ribs on the leading and trailing edges – these eventually get wrapped around the spars and epoxied into place. I made the mistake of thinking the pieces were all the same for both leading and trailing edges, not noticing that some had center holes and some didn’t.  The ones with the extra hole go on the trailing edge. Don’t do as I did and attach them indiscriminately!  Naturally, about half turned out to be wrong, and I had to drill them all out and re-rivet the correct pieces to the correct ends, all after I thought I was completely done making 26 beautiful ribs!

  • If your riveter stops pulling the rivet mandrels for no apparent reason, you probably have a piece of debris stuck between the internal jaws, preventing them from clamping down.  Simple enough to cure once you know to partly disassemble the head, but puzzling and frustrating for a newbie when you don’t know what’s going on.  It happened to me with two riveters in the space of a half hour, and only the next morning did Jonathan and I figure out how to fix them.

Before we can do much more work on the wings, we need a fresh supply of 3×16 rivets, as we ran out prematurely and need them to attach the half-ribs. Still looking for a U.S. supplier!  In the meantime, we’re on pause again, and I’ll take Thursday and probably Friday to catch up on other things.

Note: I originally had some incorrect information about bit sizes in the item above about 3mm rivets – thanks to Clay Stuart for pointing out my error!