Wednesday, March 17, 2021

Ghost in the Machine

“Look Dave, I can see you're really upset about this. I honestly think you ought to sit down calmly, take a stress pill, and think things over. I know I've made some very poor decisions recently. But I can give you my complete assurance that my work will be back to normal.”
HAL 9000, 2001: A Space Odyssey

2200+ Hours of Hand Flying

In the words of Hamilton's fictionalized King George, I am often asked, "What comes next?" 

"Moving up to a bigger, faster airplane?"
"How about something that pulls its gear out of the slipstream like a 'real' airplane?" 
“Are you seriously going to hand-fly everywhere, even in IMC?"

These are all good questions and I have been pondering them for quite some time.

Warrior 48's simple instrument panel in 2004. Half of these instruments sit on my desk now.

Even though I occasionally fly missions better suited to a Bonanza, I also like to putter along with my taildragger friends and take low and slow sunset flights that are poorly suited to a travelling machine. Flying the Warrior was always a compromise between these two extremes. While I appreciate the capability of Bonanzas and truly admire the lines of the elegant Piper Comanche, I do not have much interest in buying an aircraft with retractable gear. After 20 years of not contemplating the state of my landing gear, I see myself as being one minor distraction away from grinding the belly of a retract across the asphalt someday. Other deterrents include maintenance and insurance costs. Where these factors are concerned, simplicity is a virtue and the Warrior is simple.

I have often thought that an Archer II would be a good move-up airplane with more climb power and useful load from essentially the same airframe. My mission has evolved over the years and, if I knew in 2004 what I know now, I would have focused on an Archer II instead. Woulda, coulda, shoulda.

It is also not lost on me that an autopilot is an important safety tool for single pilot IFR operations. I remember becoming task saturated when assigned a new full route clearance while hand flying IFR  through the Washington DC Special Flight Rules Area (SFRA) in 2014. I spent roughly 20 minutes reprogramming the navigator in flight. Those knobs on the Garmin 430W must require a lot of torque to rotate because the Warrior invariably rolls off heading whenever I turn one. After many busy minutes of splitting my attention between holding a heading, climbing to an altitude, and reprogramming the navigator, I was allowed just enough time to take a deep cleansing breath before Patuxent Approach called with another new full route clearance and I started the process all over again. An extra hand would have been incredibly helpful that day. In-flight route changes, even relatively simple ones, are always tough to manage while hand flying.

Furthermore, given the lack of rest stops conveniently floating at cruise altitude, restroom breaks while hand flying on an IFR clearance are not only task saturating, but require a certain type of graceful dexterity that is not taught in flight schools. Beyond that, sometimes a pilot just needs a break. 

So...what comes next? I decided that automation might be a greater need than going faster. After all the hand flying I have done over the last 20 years, I have no macho need to convince anyone of my ability in that regard.

Aircraft Economics for Dummies

Conventional wisdom in the pilot community dictates that it is more economical to buy an airplane with the desired avionics already installed than to upgrade an aircraft without them. For a time, I contemplated Archers with autopilots already installed. Unfortunately, most of what I found was priced at a premium and lacked several of the amenities already available in Warrior 481. Many of the autopilots were old and dependent on analog gyro equipment of the sort that I discarded in the wake of my attitude indicator failure in 2017. Aircraft sales listings were often uninspiring. 

Then, three factors led me to question conventional aircraft economic wisdom.

First, I needed to put a new engine into the Warrior in 2020. The market value of Warrior 481 did not increase to match the full price of the installed engine, especially given a replacement case and new crankshaft that boosted the overhaul cost by 50%. Continuing to fly behind that nice new engine is probably the best way to wring the most value from it.

Second, the airplane market has gone absolutely crazy. I have watched several friends search for airplanes while navigating a labyrinth of overpriced junk, sketchy sellers, and rabidly competitive buyers. I think that at least one of my friends probably needs therapy for PTSD after his airplane purchase experience despite managing a positive outcome. Based on what I see in the market currently, I would be competing with other buyers for expensive used airframes with unknown flaws and ancient avionics. It's daunting out there.

Garmin's GFC 500 autopilot system consists of a G5 (bottom), servos (bottom left), and a GMC 507 Mode Controller (top). I already had a single G5.

Third, the FAA recently certified new autopilots from Garmin (GFC 500) and TruTrak / BendixKing under less onerous rules that significantly reduce equipage cost versus less-capable certified units of the prior generation. 

Taking these three factors together, I decided to buck conventional wisdom and upgrade Warrior 481 rather than trade-up to a different aircraft. I could install an incredibly capable autopilot for less than the hardware cost of legacy systems. An upgrade in 2021 made sense in a way that it did not previously. I also fully realize that there's an emotional component to this. Whether these specific upgrades truly make sense for this particular airplane or not, they make sense to me and after years of contemplation, this is what I want to do.

Nuts and Bolts

I conferred with Jake from Boshart Enterprises in Batavia, NY to make a plan. It is always a pleasure to work with people you know well and I have 15 years of experience with this shop. With Jake's assistance, I finalized an upgrade package that would enhance the safety and utility of my airplane:
  • A GFC 500 configured as a two-axis autopilot (roll and pitch)
    • I did not spring for the third servo allowing the autopilot to drive the trim as I did not think that the cost/benefit ratio was in my favor.
    • Unlike a simple "wing leveler" (single axis autopilot) or a two-axis autopilot with simple altitude hold capability, the GFC 500 can fly instrument approaches right to minimums, leaving the pilot to manage power and trim. The less expensive TruTrak / BendixKing unit is not certified to fly coupled approaches in this manner.
  • A second G5 unit replacing the mechanical directional gyro (DG) that would:
    • Provide redundancy for the existing G5 attitude indicator. Although the units are connected and pass data back and forth, they are completely independent and use separate back-up batteries.
    • Eliminate the airplane's balky vacuum system entirely. I never really trusted it. Now I won't have to.
    • Include installation of a magnetometer that would eliminate the need to constantly reset the mechanical DG due to gyroscopic precession. 
    • Add horizontal situation indicator (HSI) capability and eliminate a Garmin GI 106 mechanical course deviation indicator from the panel (sold back to the shop).
    • It was already my plan to add the second G5 in 2020, then Dansville happened.
Interestingly, these upgrades will also serve to classify Warrior 481 as a Technically Advanced Aircraft in the eyes of the FAA, a status I never expected my airplane to achieve. All I need is a parachute, more power, a really good waxing, and I'll practically have a Cirrus! This is not a terribly important classification unto itself except that, if I choose to pursue a commercial rating, I will be able to do it in the Warrior under the new rules.

Work in Progress

"Before" - Photographed 13 Oct 2019

Ed C followed me to Genesee County Airport in his Archer on the afternoon of February 17 to drop off the Warrior. Jake provided routine photographic evidence of his progress.

26 February 2021, photo by Jake

In just over a week, the panel was well and truly gutted except for the primary pitot-static instruments.

26 February 2021, photo by Jake

26 February 2021, photo by Jake

26 February 2021, photo by Jake

The roll servo was installed in the spar box under the back seat. It simply clamps directly to the existing cables that drive the ailerons.

26 February 2021, photo by Jake

There was a delay receiving the correct bracket for mounting the pitch servo. That would eventually mount all the way back in the empennage next to the emergency locator transmitter (ELT, the orange box in the photo above). I teased Jake about having to crawl back in there to work, but he revealed that it was actually Jeff, the owner of the shop, who undertook that particular chore. Hats off to Jeff!

09 March 2021, photo by Jake

Sometimes, things look worse before they get better.

09 March 2021, photo by Jake

By March 9, Jake had all the trays positioned and wired for the radio stack. Completion was nearer at hand than this photo makes it seem.

09 March 2021, photo by Jake

Later that same day, Jake had the center stack complete and the 48 year old King KX-170B radio slid into place. I have literally waited 17 years for this thing to die. It simply refuses to acknowledge its own obsolescence. Why would I spend a few thousand dollars on a new navcom radio when this one works perfectly well? 

I used to boast about having the ugliest KX-170B in operation, but Jake exercised his artistic side and repainted it. It actually looks pretty good now and the new coax connecting it to the antenna has eliminated background noise that has been apparent for the last five years.

By March 14, the work was finished except for the cosmetics. Jake was kind enough to grant a visitation.
The Biggest Loser

All set except for some black touch up paint needed on the instrument panel

I took a vacation day on Monday, March 15 to do the test flight and (hopefully) bring Warrior 481 home. Tom flew me over to Genesee County in his beautiful new-to-him Cherokee 180, Two Six Romeo. Tom's airplane has the same GFC 500 system and dual G5 setup that I installed. Seeing Tom's capable autopilot in action whetted my appetite to test mine.

 As part of the upgrade, Jake performed the first direct measurement of weight and balance on the airplane in the 21st century. With the vacuum system removed, Warrior 481 lost weight and picked up about 17 pounds of useful load. Not bad! I could stand to lose about 20 pounds myself.

Beware the Ides of March
Date Aircraft Route of Flight Time (hrs) Total (hrs)
15 Mar 2021 N21481 GVQ (Batavia, NY) - local flights - SDC (Sodus, NY) 1.6 2216.1

It was an auspicious date for a test flight, but what was the worst that could happen? It was not as if my airplane could literally stab me in the back. 

After a thorough preflight in the Boshart Enterprises hangar, we pushed Warrior 481 out into the cold. With Tom in the right seat, I ran the engine start checklist and discovered that the fuel boost pump was inoperative.

The massive hangar door was lifted again, flooding the shop with cold air. Inside went Warrior 481. Jake climbed in, energized the master switch, and the boost pump ran as expected. After some group speculation about my competence as a pilot, closer examination revealed that the rework on the breaker panel left a loose connection for the boost pump. Jake found it immediately and fixed it. He second checked the remaining breaker connections.

Up went the door and out went the Warrior again. This time the boost pump ran as expected and I fired up the engine. Before Tom and I launched from runway 10, I pushed the TOGA (take-off / go around) button on the panel, then firewalled the throttle, and climbed into the cold March air. About 800 feet off the ground, I engaged the autopilot to continue the climb in take-off mode.

Warrior 481 immediately attempted to kill us.

Et tu, Garmin?

Activating the autopilot resulted in a violent pitch skyward toward a stall attitude. As the airspeed dropped precipitously, I pushed back on the yoke and easily overpowered the autopilot. Then I remembered the autopilot disconnect switch (what I now call the "nope button") on the yoke and mashed it with my thumb. 


After sounding an alert tone, Warrior 481 immediately relented in her murderous intent. I pushed the nose down to a normal climb pitch and watched the airspeed needle reassuringly sweep clockwise. Airspeed is life.

Surely that was user error.

Trying a new tactic, I set an altitude of 3,000 feet and programmed an indicated airspeed climb of 90 knots. The Warrior immediately dumped her nose into a steep dive.


We tried using the "Level" button, an emergency feature designed to recover the airplane from an unusual attitude created by a disoriented pilot. Upstate NY farm country immediately filled the windscreen as the Warrior dove hungrily for the Earth again. Was it "opposite day"?


Tom had a smart idea to test the "programmed backwards" hypothesis. We programmed a shallow 300 foot/minute descent that resulted in an immediate and extreme upward pitching moment.


We'd seen enough. At least the "nope" button worked. We returned to the shop and reported our observations to Jake. Up went the door again, back inside went the Warrior. 

As Jake checked the autopilot programming, I had the good fortune to reconnect with Ed V, a Le Roy pilot whom I have not seen in many years. [In Troy McClure's voice] You may remember Ed from such blog posts as "Rescue from Pontiac", "Where There's Smoke...", and "Leitmotif". It was Ed who introduced me and Darrell to Tom H, our truly excellent instrument instructor and all-around terrific human being. I was genuinely pleased to see him again.

Jake emerged from the Warrior looking chagrined. "I programmed it wrong, that's all on me." He was clearly beating himself up over the error. But this is why we do test flights. I told him that I expected some future mistakes of my own as I learned to manage the system. With assurances that everything was set properly, up went the hangar door and out went the Warrior again.

Magic Carpet Ride

This time, there was no need to use the "nope" button. The autopilot functioned flawlessly. Activated 1,000 feet above the ground, take-off mode flew us away from the runway at a comfortable climb pitch. I programmed a top altitude, an indicated airspeed climb, and used heading mode to turn the airplane southbound. As the Warrior leveled herself at 3,000 feet, I supplied some nose down trim (as prompted) and reduced power to a cruise setting. I programmed the system to track the 314° radial to the Geneseo VOR. The interception of the radio beam was smooth and the tracking flawless. I reset the GNS 430W to fly the VOR A approach procedure into Le Roy using GPS guidance, including the procedure turn. Without any hands on the controls, the Warrior executed a flawless teardrop entry into the holding pattern and flew a perfect interception of the inbound course. The Level button successfully salvaged an intentional unusual attitude. 

It all worked. Having verified that the autopilot could track both VOR and GPS inputs, could climb and descend at indicated airspeeds and vertical speeds, and that the Level button actually lived up to its name (this time), we returned to Genesee County to report a successful test flight.

Flying home from Genesee County Airport. Look ma, no hands!

With the Warrior returned to service, Tom fired up Two Six Romeo for the flight back to Sodus. I collected the reams of paperwork making my modified aircraft legal to fly and launched into the evening sky. I started by hand flying out of habit, but then programmed my ship to fly me home. I folded my arms across my chest -- because that is how one properly rides a magic carpet -- and monitored the instruments. I realized how well the old vacuum gauge had been incorporated into my scan every time I caught myself glancing at the blank panel where it used to reside.

Now the real work begins. The next time I fly practice approaches is going to be interesting.


  1. Congrats on the upgrade! Managing systems really helps relieve the hand flying stress on those 'busy' IFR flights. You and I both have done a lot of hand flying, and I too am comfortable with those skills. I have watched a few videos on your new AP, and I have to say I like the altitude preselect and the LVL button, great features.

    It took me a while to learn the buttonology on the Stec 60-2 and I could always use more practice time to help ease the comfort factor of the plane flying the approach. It gets better each flight.

    I'm looking forward to your currency flights, and how you mix up the approaches. Have fun, and enjoy the new toys.

    1. Thanks!

      The thing is definitely slick! (I've already started calling it "HAL" - is that setting myself up for failure?) I think the trick will be working through how to manage it through approaches. It has an approach mode for procedures with lateral and vertical guidance. Non-precision approaches are flown in NAV mode, but the altitude preselect can be used to manage leveling at the MDA. I know this all in theory, but it will be interesting to see how it works in practice, what sequence needs to be followed to transition onto an approach after receiving vectors in HDG mode, etc. The manual is curiously thin on some of those procedures, but the videos on YouTube are really helpful. I'm comfortable enough with the basic "cruise" functions already that I'm looking forward to trying the more complex features.

      I will also have to use the GPS differently. In the past, I would happily activate approaches before being cleared for them and would simply ignore the GPS guidance (in favor of vectors or whatever) until I needed it. I will have to be very cognizant of what mode the AP is in before activating approaches in the future!