John Steere's Supercharged Thunderbird Powered BD-4
CONTACT! magazine reported on John Steere’s BD-4 in issue No. 66. A few years ago, we ran into John at Oshkosh and asked him to give us a follow-up report. Here’s what he sent us:
Story by John Steere, EAA 301134
This is the configuration as flown to AirVenture 2003. Modifications since then are described in the article.
The plane is a BD-4, completed in May of 2000. I chose the BD-4 because it was easy to build, relatively inexpensive, carries four people, and is relatively fast. Never satisfied with the status quo, I incorporated significant modifications to accomplish specific objectives. To reduce the stall speed, the wings and flaps were each lengthened 22 inches per side, and the flap deflection was increased from 30 to 40 degrees. The wing was changed from the original fiberglass wing panels to aluminum in an effort to reduce the fuel leakage problem frequently experienced with the fiberglass panels.
Rudder trim was added, and both stabilator and rudder trim are driven by small, industrial DC gear motors controlled by a small electrical joystick located between the front seats. The doors were modified to hinge from the top rather than the front and were supported with gas springs. The objective was to make it possible to open the doors in flight for better photography. The modification also significantly improves cabin cooling while on the ground on hot days.
Photo by Pat Panzera
I chose an automotive engine because I could not rationalize the value of a conventional aircraft engine, and I enjoy a technical challenge.
The engine is a 90-degree, 3.8-liter V-6 from a 1990 Thunderbird Super Coupe. It is supercharged by an Eaton blower and intercooled. The compression ratio is 8.2:1 and the fuel is supplied by six fuel injectors. The engine is rated at 210 hp at 4000 rpm and delivers 315 foot-pounds of torque at just 2600 rpm.
Among other essentials is a Northwest Aero Products 1.7:1-ratio cog belted propeller speed reduction unit (PSRU). This engine conversion includes the standard automotive sensors, multiport fuel injection system, and computer ignition systems, less the standard emissions junk. An emissions air pump (not standard in the car) was added to serve as a vacuum source for the flight instruments.
The philosophy I followed in converting the engine for aircraft use was “don’t change anything unless absolutely necessary.” Lacking automotive engineering credentials, the safest tactic was to keep the conversion as simple as possible. There was no reasonable way to improve on the thousands of engineering hours that Ford invested in the development of this engine. However, some external adaptations were required to match the engine to the airplane.
A thorough study of available data on water-cooled aircraft engines was completed, and information from the most successful installations was used for guidance. The data indicated that the design targets for the active frontal area of the radiator should be 1 square inch per horsepower and that the radiator volume should be 2.52 cubic inches per horsepower.
A radiator shop in Indianapolis fabricated the custom design based on the design targets. Assuming the engine would produce the factory-rated horsepower, the radiator was to be 23.5 inches wide, 9 inches tall, and 2.5 inches thick. There are small tanks on each end with 1.25-inch hose connections, a small drain valve in a lower corner, and a flange on the upper surface to mount it to the airplane. The radiator is riveted directly to the fuselage, perpendicular to the airstream. The radiator is connected to the engine by aluminum tubes (with 1.25-inch-thick walls) that run from the firewall to the belly scoop under the fuselage. These are enclosed in a small tunnel for drag reduction and appearance.
The belly scoop, made from foam and fiberglass, slips over the radiator from below and is fastened to the fuselage with four bolts in the corners. Guidance from National Aeronautical Charting Office (NACA) publications dealing with inlet and outlet configurations used in warbirds was used for the belly-scoop design.
When the shop completed the radiator, I found it to be much thinner than I had requested. Instead of the specified 2.5 inches, it was only 1.2 inches thick. Based on his experience with auto racing, the shop owner believed it would be sufficient. So I mounted it, and flight-testing has proven that he was basically correct. However, there have been times, on hot days during long climb-outs, that the thicker radiator might have been beneficial.
Early Development Issues
One of the early issues with the engine was an intermittent misfire that occurred sporadically during full-power climb-outs. After eliminating every other possibility, I removed the fancy high-performance, spiral-core spark-plug wires that were advertised as being far superior to the OEM equipment and replaced them with OEM resistive wires. The misfire problem was fixed. I did not say solved, as the actual underlying cause is not understood. It may have been an impedance mismatch with the high-voltage coils. This is one of the areas where straying from the “don’t change anything unless absolutely necessary” rule created an unnecessary problem.
For the first 100 hours or so, the plane was burning about 13.5 gallons per hour in cruise, and the exhaust was very sooty. The primary contributor is the OEM computer programming that adopts a wide-open-throttle (WOT) algorithm when certain operating parameters are met. In the WOT mode, the computer stops looking at the oxygen sensors to determine how to adjust the fuel injection duty cycle and switches to an open-loop mode, using a look-up table that injects a predetermined amount of fuel based on data from the mass airflow sensor (MAF). To protect the engine, this is designed to be a rich mixture. Unfortunately, the computer defaults to this mode for any power setting greater than that required for minimum cruise.
To overcome this problem, a passive voltage divider circuit was designed and placed between the mass airflow sensor and the computer. (See above diagram.) This circuit supplies an adjustable percentage of the signal from the MAF to the computer. This causes the computer to think the engine is taking in less air than it actually is, so the computer tells the injectors to deliver less fuel. The manually adjustable potentiometer is positioned next to the throttle, allowing the mixture to be adjusted in flight. The potentiometer is the only component in the circuit that has any significant potential for failure. Its likely failure mode is to electrically open rather than short out, so it’s positioned in the circuit where a failure will cause the mixture to go rich if it opens - a safe condition.
On this water-cooled engine, the EGT does not peak and then noticeably drop off as the mixture is leaned. The procedure that seems to work best is to slowly lean until the engine starts to run rough, then enrich the mixture until it runs smoothly.
Note the rotary dial beside the throttle vernier, used to change the mixture while the engine is in the open-loop mode. The fuel gauges have also been recalibrated.
Photo courtesy of John Steere
The original induction system was a simplified version of what was supplied in the car. Due to size constraints, the original ducting and resonator were eliminated and replaced with a flexible tube that carried induction air from a T-Bird air filter directly to the MAF sensor. The air filter was positioned in a plenum immediately behind a NACA flush inlet on the left side of the cowling.
This system worked but took up a lot of space, and I was concerned that the air filter might soften and collapse into its plenum if the plane was flown through a heavy rainstorm. The air filter location was also the only practical place to mount a larger oil cooler that was needed to extend the time between oil changes.
In late 2003, I decided to reconfigure the induction and oil cooler systems to get them off the worry list. I found a simple, well-constructed cone filter at AutoZone with an exit flange that fit perfectly into the inlet of the mass airflow sensor. (See below photo) The opposite end of the cone filter is supported by an aluminum bracket attached to the engine mount by an Adel clamp. This all fits neatly under the cowling and ends the concern about a wet filter collapsing and choking the induction system. I may be paying a small power penalty since the induction air drawn from under the cowling is warmer than air sourced through the original NACA flush inlet.
The new, more compact air cleaner and intake system.
Photo courtesy of John Steere
The original oil cooler was a small single-loop unit set in the NACA inlet plenum on the right side of the plane immediately in front of the intercooler. With the NACA flush vent on the left side now available, the 11-inch square plenum was filled with an automotive oil cooler. The effect on oil temperature was very gratifying. With both measurements taken on 70-degree days, the oil temperature dropped from 232 degrees to 210 degrees. With the new oil cooler now taking on a larger share of the overall cooling load, the water temperature dropped from 227 degrees to 220 degrees. Very satisfying results for an amateur.
The exhaust system has been the most troublesome aspect of the engine conversion. To save costs, the original headers were fabricated from mild steel tubing. Bad decision! Because the original header design brought them within 3/4-inch of the spark plugs, they had to be insulated with an exhaust wrap to protect the spark-plug boots. The insulation was very effective, but it did not last long. Within 10 hours it lost its flexibility, and if you inadvertently touched it while working on something nearby, it would fracture and break off. The more serious matter was the higher manifold temperature caused by the insulation. With any cooling air to the header effectively blocked by the insulation, the mild steel overheated, and finger-sized holes developed in both headers within 20 hours.
The original exhaust system, as described above, proved to be problematic.
Photo courtesy of John Steere
The new exhaust system appears to be more efficient, being closer to an “equal length” system than the previous arrangement. But the real difference here is with the spark-plug boot clearance.
Photo courtesy of John Steere
The completed new system includes headers, a heat shield, and a new single muffler, which replaced the twin mufflers.
Photo courtesy of John Steere
New exhaust headers in the same general configuration (mistake number 2) were fabricated from 321 stainless. I inadvertently ordered the bends with a larger radius (mistake number 3) than were used with the mild steel header, resulting in the tube passing within a half inch of the plugs. The stainless tubing held up to the heat. But even with the insulating wrap, a spark-plug boot would occasionally break down, allowing an arc to occur between the spark plug and the header. This happened once while my wife was on board, leading to an emotional reaction to the sudden engine roughness and a significantly reduced confidence in the airplane. These events were not life threatening since the airplane could still climb very well on five cylinders.
After several “five-cylinder” events, I decided to redesign the entire exhaust system. The new design had to provide as much clearance from the spark plugs as possible, eliminate the need for insulation wrap, and make accommodation for the under-engine muffler I wanted to add.
The original design brought the header straight back toward the firewall before dropping down to exit just ahead of the firewall. The new design drops immediately from each exhaust port, with the three tubes joining approximately 8 inches below the ports. This significantly improves the clearance from the spark plugs and provides a more direct route toward the bottom of the cowling. Below the “Y” connection, the tubing diameter changes from 1-3/4 inches to 2 inches, and the oxygen sensor mounts just below the “Y.” The 2-inch tube continues down to a ball joint, then terminates with a V-band clamp that enables removal of the exhaust pipes or muffler without removing the headers.
The exhaust manifolds are enclosed in a thin (0.025-inch) stainless steel heat shield to prevent excessive radiant heating of adjacent components. The heat shields include 1-inch air duct couplings where outside air is introduced to provide in-flight manifold cooling. The spark-plug boots are also enclosed in protective sleeves because they are still too close to those glowing header pipes. This header arrangement has worked well since it was installed eliminating those annoying “five-cylinder” events.
This system was flown for about 38 hours with straight pipes plus a crossover tube attached to the V-band clamps. It performed very well, but the noise made communication in the cockpit very difficult, even with an intercom and good noise-canceling headsets. The neighbors (I live in a fly-in community) also said that the noise made the trees rattle during takeoff.
For both of the previously stated reasons, a muffler was in order. Due to the tight cowling and cramped quarters under the engine, a suitable off-the-shelf muffler that would physically fit could not be found. So I designed a muffler that would fit the available space and hopefully provide enough internal baffling to reduce the noise without causing too much back-pressure.
The old system was completed by routing the exhaust out the bottom of the cowl and along the belly to a pair of small and ineffective mufflers.
Photo by Pat Panzera
CAD courtesy of John Steere
CAD courtesy of John Steere
Since I am not a muffler engineer, some elementary research was required. An Internet search revealed that there are five basic principles available for use in passive mufflers - reflection, absorption, restriction, expansion, resonance, and combinations of these. Space constraints limited options for this application primarily to the restriction realm.
The muffler consists of two U-shaped tubes, two baffle plates, two end caps, and an outer skin. The tubes are 16-gauge 321 stainless, and the baffles, ends, and wrap are 0.025-inch-thick 304 stainless, all welded together with 308 stainless filler rod. The end caps were shaped using wooden form blocks for the perimeter and simple turned-steel mandrels to form the holes for the tubing. The tubes are preformed, 90-degree bends welded end to end in the center of the muffler.
The exhaust gasses enter from both ends of the flanged exhaust pipe, exit the pipe through a series of 1/4-inch holes along the forward surface, then travel downward (because they are blocked by a bent-up tab on the upper baffle). They then pass through the lower baffle, between the two pipes, through the upper baffle, through perforations into the tailpipe, and out the ends of the tailpipes to the atmosphere. Each set of perforations has about twice the open cross-sectional area of the tubes.
The entire muffler is enclosed in a stainless steel heat shield, including the exhaust pipes, until they exit the lower cowling. Cooling air is introduced between the heat shield and muffler by way of a 2-inch air duct. The cooling air is exhausted between the tailpipes and their heat shields. This cooling air is not used for cabin heat as there is a welded joint running the length of the muffler skin, and there is another, safe source for cabin heat.
The effect of the muffler was immediately apparent. The noise in the cockpit was significantly reduced to the point that intercom communication was possible without turning on the noise-canceling in the earphones. That’s not to say that I wouldn’t like it quieter, but the noise has dropped from oppressive to operationally reasonable in the cockpit.
The effect on power has yet to be fully analyzed, but it does not seem to be significant. Supercharger boost pressure during takeoff has increased from 9 psi to 11 psi, and at cruise rpm (2400), the pressure increased from 2 psi to 3 psi. With all else remaining the same, boost pressure is related to back-pressure in the exhaust system. The 11 psi full-throttle boost is still less than the 12 to 14 psi normally seen in the 1990 T-Bird Super Coupe from which this engine was removed.
I am not comfortable using the exhaust system as a cabin heat source unless the heat muff can be mounted over an area free of welded joints. The new, compact exhaust system doesn’t offer that choice. Water-cooled engines also provide the option of using a small automotive heater core as a safe heat source, but they require additional plumbing, add some weight, and take up space for the heater core and a circulation fan.
With the engine compartment and both sides of the firewall rather crowded, the heater core was not a good option. Fortunately, there is another source for safe, warm air from a supercharged engine - the intercooler. Cooling air for the intercooler enters through a flush NACA inlet on the right side of the airplane and passes through the intercooler into an exhaust duct that directs the warmed air out the bottom of the cowling, just ahead of the firewall.
A 2-inch air-duct flange through the firewall, coupled into the intercooler exhaust duct, captures part of the warmed air and delivers it to the cockpit through a butterfly valve. The intercooler only provides warm air when the supercharger is actually pressurizing the induction air, so there is no heat while on the ground. Since this system is pressurized by outside air through the NACA inlet on the side of the cowling, there is no risk of carbon monoxide poisoning.
Fuel Gauge Calibration
I learned a very important lesson a couple years ago while on a return flight from western Michigan to my home base at Twelve Oaks Airport in Martinsville, Indiana (II87). Shortly after passing Indianapolis Metropolitan Airport, the engine quit. A quick scan of the fuel gauges indicated about six gallons remaining in the left tank (the tank from which the engine was drawing fuel) and about 10 gallons in the right tank. I immediately switched tanks, turned on the second fuel pump, and said a quick but urgent prayer. I then remembered that Metro Airport had just passed under my right wing, so I made a quick right turn and found myself perfectly set up on downwind for Runway 15.
I continued to look for the source of the problem on downwind, believing it to be fuel related because of the way the engine died - a smooth rpm reduction over a period of about three seconds. But I had very carefully calibrated the fuel gauges while building the airplane and had marked the gauges at several points with the actual fuel level, and the gauge still showed six gallons. As I turned onto a short final, with the prop windmilling, the engine restarted on its own. I landed and taxied to the ramp, with everything performing normally. A visual check of the left wing tank found it to be completely dry.
After thinking about this for a few minutes, it dawned on me that the procedure used for calibrating the fuel gauges was faulty. The gauges were calibrated while the plane was still in the basement, with the plane on about 12 volts of battery power. In flight, with the alternator charging, the electrical bus is carrying about 14 volts. That two-volt difference caused the gauge to read six gallons higher on the low end of the scale. Needless to say, the fuel gauges were recalibrated with the engine running.
Photo by Pat Panzera
The BD-4 uses a castering nose wheel with a spring-loaded friction pad around the spindle to prevent nose-wheel shimmy. Early in flight-testing, the spring pressure was reduced because the brakes were wearing rapidly as I fought the nose wheel during mile-long ground excursions. Unfortunately, the reduced friction allowed the nose wheel to shimmy aggressively.
After fighting the friction adjustment battle and other unsuccessful mechanisms for three years - and destroying two nose-wheel pants - I decided to add a hydraulic damper. Not wanting to pay the ridiculous cost of an aircraft damper, a search for alternatives was in order. An Internet search led to a self-centering damper that, in normal life, is used to control hydrostatic transmissions, typically linked to a control lever or pedal to provide a controlled rate of actuation. This little unit cost less than $50 and has eliminated the shimmy problem.
Now a new wheelpant is in order.
With as neat and tidy an engine installation as this one, would you expect anything less in the interior? Clean, simple lines - elegant in form and function - grace the cockpit of this plane.
Photo by Pat Panzera
In the first report on this plane, I wrote that “I could not rationalize the cost of a new or used 180-200 hp Lycoming engine [that the airplane was originally designed to use] against my perception of the value of the engine.”
To put things in perspective, I saw a recent advertisement in an aviation periodical for a 180-hp fuel-injected aircraft engine with FADC controls for $29,900. The Thunderbird installation, including the PSRU, alternator, air pump, supercharger, fuel injection, computer control system, and all the sensors and other incidentals that were needed to complete the system, cost approximately $7,000. I think that is a pretty good value for a new engine with 21st century controls rated at 210 hp.
All of that is still true, and based on the 185 hours of experience with this engine, I would choose an automotive engine again. Hmm…that Corvette LS1 engine looks interesting!