KITPLANES Magazine, April 2001
Turbo Power, Part 2
The author continues converting a Subaru EJ-22 for aircraft use.
By Tom Wyatt III
Last month we began the intricate task of converting a turbocharged auto engine for aircraft use. Specifically, we were converting a 2.2-liter turbo Subaru EJ22 engine to power Bob Goodman’s Van’s RV-4 homebuilt. If you’ve read Part 1, you know that we’d nearly finished the engine mount.
After picking up the engine mount and related parts at the powder coating shop, Bob and I spent a few hours with sandpaper and flapper wheels restoring the tight slip-fit of the gear legs into their sockets on the engine mount. Bob then took the engine mount with the gear legs installed home to install the assembly on the front of the completed and painted fuselage.
A few days later, he arrived back at my shop with the fuselage tied down on the back of a flatbed wrecker. With all three wheels installed, the assembly rolled easily into the shop.
We temporarily installed the engine assembly (engine with propeller redrive [PSRU]). Yet to be fabricated was virtually all of the plumbing: oil cooler, intake manifold (with injectors and fuel rails), all engine compartment wiring, (including the computers), instrument panel wiring, heat shields, oil tank, air box, and the radiator inlet and outlet ducts. The engine compartment of the RV-4 is quite compact, and it was going to be a challenge to squeeze it all in.
Designing the Intake Manifold
This was going to be a big job; there was no doubt about it. We had several problems to solve and a limited space in which to solve them. The usual composite solution to the Subaru intake manifold problem is to retain the stubs of the original intake manifolds, which mount the injectors and fuel rails. We weren’t going to have that luxury. The RV-4 cowling is extremely tight across the cylinder heads, and there was no way to mount the injectors in the original location. They would have to be mounted upstream where the cowling humped up over the center of the engine.
As the port runners were in place (fabricated along with the engine mount), I started carving foam to mold the body of the intake manifold. I decided to break the manifold into three parts to facilitate construction. The manifold would be constructed in these individual parts and then glassed together. I have found the lost-foam method of construction to be an excellent way to fabricate the complex shapes. Most of the composite components on 311U have been made using this method.
I spent more than 100 hours designing and fabricating the composite manifold. It is a complicated piece. Attached to the intake manifold are the throttle body, port runners to the cylinder heads, injector mounts, fuel rail mounts, attachment points to the block, and inserts for sensors and pressure taps. The manifold had to fit between the cowling and the block, clear the alternator and block breather tube and distribute the air evenly. One of the biggest challenges was to build a structure that could stand up to alternating vacuum/pressure cycles, backfires and the extreme under-cowling temperatures after engine shutdown. Not an easy task for what is essentially a plastic manifold.
The Lubrication System
Initially, we had planned to use the stock Subaru oil pan. This idea fell by the wayside when we discovered that we were not going to be able to mount the turbocharger behind the engine as we had planned. Instead, we moved it to the bottom of the engine where the oil pan used to be.
The oil tank had to be mounted at the very bottom of the airplane to allow a gravity drain-back of the oil from the turbo. When the oil leaves the turbocharger, it looks like dirty whipped cream, so the drain must be a relatively large 58-inch (the supply line is 14 inch), and slope continuously down to the tank. It must also enter the tank above the level of the oil, or it will not drain properly.
Once again the solution was composite construction. Like the intake manifold, I used S-glass and epoxy and the lost-foam mold. I constructed the tank in two parts–the top and bottom halves.
The oil tank is shaped like something between a football and a kidney. It holds 4 quarts below the turbo oil drain, but because it is funnel shaped, it would probably work fine with as little as 2 quarts.
Plumbing the Oil
For safety and convenience the entire lubrication system is plumbed with steel-braided Aeroquip hoses. The suction hose from the oil tank to the pump is -10, and all the rest are -8, with the exception of the individual supply lines to lubricate the PSRU and turbo. These individual supply lines are -4.
The oil leaves the oil tank through a scavenge filter that protects the oil pump from large chunks that could cause sudden stoppage or serious pump damage. It contains a stainless steel screen that is easily removable for inspection–a nice feature.
The oil cooler we initially installed was an oil-to-water unit instead of the usual oil-to-air unit found on aircraft.
We had positioned the radiators behind the original nose inlets when we built the engine mount. Once the cowling was in place and the radiators were bolted in, it was time to fabricate the inlet ducts to capture the ram air at the inlets and smoothly conduct it to the face of the radiator. In our situation, we had to design and fabricate ducts that swelled from a 3.5×7-inch oval inlet to the squared 8×10 face of the radiator.
Many requirements were needed for this duct to function at its best. First, it had to be perfectly air tight at the inlet end and at the radiator face. Second, it had to have a smooth inner surface that transitioned from the oval inlet to the square radiator face over a very short distance (4 inches near the spinner, 7 inches outboard). Third, it had to be light, rigid and airworthy. Once again I used the lost-foam mold for the complex shape, and S-glass and epoxy for the fabrication.
Electrical System and Engine Management Computers
One of the most complicated parts to design was the electrical system. With so many devices that could stop the engine powered by electricity, much consideration had to be put into the design. We must have done our job well because we modified only one thing after we started flying the airplane. Actually, that single modification took place before the airplane left the ground because we discovered it during taxi testing.
The computer-controlled engine management system was part of the original design concept. The original-equipment Subaru computer is quite reliable and would have worked well on an engine that was closer to stock than ours. In fact, for anyone seeking the least expensive way into the air, I highly recommend it.
Our problem was that we didn’t have the stock computer, and our engine/turbo setup was too far from stock to use the stock computer. Another problem was that we were not going to be able to use the stock injectors. The aftermarket offers several choices of engine management systems, all of which allow easy manipulation of the fuel delivery and spark curves using a laptop computer.
For safety we have installed two trigger wheels and two crank position sensors. In fact, we have two independent CPUs, each with its own set of sensors. The pilot can switch back and forth between them in flight, and either CPU can drive the injectors and coils, which are shared. There are only four spark plugs. My experience is that coils, injectors and spark plugs are reliable enough to depend on one each. The coils fire the sparkplugs directly. There is no distributor, nor is there any easy place to put one on the Subaru engine. This is one of those modern engines that was designed from the beginning to be computer controlled.
The other critical component that we considered important enough to make redundant was the electric fuel pump. There is no engine-driven fuel pump, so a pump failure could cause the engine to stop. This is a good reason to install a spare.
The pumps are not particularly expensive, so installing a backup was not a big deal, but the computers are expensive. There are also installation hassles when pairing components so that either one will work and not interfere with the other.
The battery and charging system also required careful design to provide maximum safety. This will raise some eyebrows, but we only have one battery. The alternator is our primary source of electrical power, and the battery is the backup.
We used an external linear voltage regulator (B&C Specialty Products) that provides many features and is vastly superior in terms of safety. The linear voltage regulator applies a steady voltage to the field windings producing a steady output. There is no on/off-duty-cycle noise, and the battery is redundant, not essential.
We have installed a battery switch in N311U that allows the battery to be completely removed from the system while the engine is running.
If the alternator should fail to provide adequate output, the engine would continue to run off of the battery until it was depleted. This would be a gradual failure that could go unnoticed until the engine quit. The external regulator we installed features a low-voltage warning light that alerts the pilot to a deteriorating battery situation.
As bad as a low-voltage situation could be, worse things can happen. Over-voltage can occur in just a few seconds, and the failure of electrically driven components could be catastrophic. In this situation, a pilot’s reaction to the rapidly changing situation would likely not be fast enough. The external regulator features crowbar over-voltage protection that trips the alternator circuit breaker before the voltage can run away. (Don’t wire your ignition system to the same circuit breaker!)
As you can see, there are many considerations for making an airworthy charging system for an airplane with a battery-driven ignition/fuel injection system. Many of the same considerations also apply to standard aircraft, but a failure of the charging system on those, while inconvenient, is unlikely to make the engine stop running.
Radiant Heat Control
Many efforts were made to keep the hot turbine section from becoming a problem. The turbine housing was coated with a thermal barrier coating and was then enclosed with a close-fitting stainless steel and ceramic heat shield. Over that is the ceramic-coated barrier heatshield that separates all of the hot pipes and turbine section from the rest of the engine. All of the turbine inlet and outlet plumbing is also ceramic coated. The air that is taken onboard at the front of the airplane is directed under the engine and over the hot parts after it goes through the exchangers. Any parts of the engine mount, wiring harness or plumbing that pass near a hot part are insulated with foil-faced ceramic insulation.
From the outset, one of our design goals was to use the stock cowling with minimum modification. We already knew that we had two problems making it fit. The first was the intake manifold protruding through the top cowling in five different places. The other modification was going to be necessary because of the reversed rotation of the propeller. To offset torque and P-factor, the propeller disk is tilted one way for clockwise rotation, the other for counter-clockwise rotation. This caused us to have a wedge-shape opening between the spinner and the front of the cowling.
To clear the runners of the intake manifold, we cut away the offending portions of the top cowling. We then installed the cowling with the intake runners protruding through. Half-inch-thick pieces of foam were then taped to the high points of the intake runners to establish adequate clearance. I then set about shaping 0.25-inch Clark Foam filler strips to bridge the gap between the top of the cowling and the chipmunk cheeks. I made these strips almost the full length of the cowling so they would fair in a smooth-looking aerodynamic way.
The 0.25-inch foam readily formed to the necessary compound curvature and was bonded into place using 5-minute epoxy. After that, I followed with three layers of S-glass and epoxy, running the glass out about 1 inch beyond the foam onto the existing cowling (gel-coat removed). After curing, we then removed the top cowling and cut away everything that was under the filler strip. Then the inside of the cowling was glassed in with two layers of S-glass.
Making It Run
By this time, the engine had been on and off the airplane dozens of times. We had been using it as a dummy or mockup to allow us to fabricate parts. Now it was time to think about starting it.
We pulled the engine and PSRU one last time. We had all the recently fabricated metal parts powder-coated, and we painted the composite ones like the intake manifold, oil tank, and radiator inlet and exit ducts.
While the engine was off, we prepared the engine and the firewall for permanent installation. The firewall served as a vertical mounting surface for many components, both on the engine side and on the cockpit side. With the engine temporarily out of the way, we installed a foil-faced ceramic insulating blanket to add fire protection and to stop noise and heat.
With the engine bolted back in place on the fuselage, we got serious about finalizing each system. We started by completing and leak checking the fuel system. The wings were not installed and the fuel tanks live in the wings, so we rigged a small 2-gallon plastic fuel tank alongside the fuselage at the wingroot position.
All of the high-pressure lines are -6 Aeroquip. The low pressure (return) hoses are of high quality material, but they do not have steel-braid outer sheaths. A single -6 Aeroquip line passes through the firewall (using a bulkhead fitting), then through the fuel filter and to the fuel rails in series.
Once the fuel system was able to supply fuel to the injectors, the lubrication system was installed, and the electrical system was hooked up and tested. On the first attempt, it fired right up–on one cylinder. Pop, pop, thump, thump–not good. We delved into every possible problem; CPU (A) and CPU (B) acted exactly the same. The firing order was correct. The plugs were OK. The fuel pressure was indeed 45 psi. Finally, I pulled the intake manifold up a few inches and had Bob crank the engine over. Only one of the four injectors was putting out a stream of fuel. Ah-ha!
After a quick call to the injector supplier, I had a possible answer. A couple of years had passed since we had purchased the injectors, so it was likely that the protective oil inside the injectors had gelled and was freezing up the solenoid operated valves. I was, however, a bit dubious about the solution suggested.
They recommended that I tap the sides of the injectors with a wrench to free up the valves. I cringed as I smartly smacked each injector in turn. It worked! The next time Bob spun the engine, four injectors squirted. When I reinstalled the manifold, the engine started right up–this time without a pop or a thump. Sans prop, it revved like a Ferrari!
We ran the engine only a few seconds to make sure that there were no oil leaks. It has been said that this is the male equivalent of giving birth. It was alive! We spent another day finalizing the plumbing of the cooling system, and we ran the engine for a more extended period.
The Cold Air Box
One of the final pieces to be fabricated was the odd-looking structure that conducts air from the stock air inlet on the cowling to the compressor inlet. There were many requirements to satisfy and a very limited and defined space in which to satisfy them. We needed a way to get the cold high-pressure ram air from the inlet on the bottom of the cowling to the turbo’s compressor. The solution? A lost-foam mold, of course.
To the Airport
With the engine running and all systems complete, it was time for the airplane to leave my shop and go to the EAA Chapter 690 hangars at the Gwinnett County Airport where the wings would be installed and final preparations would be made for flying. In Part 3 of this series, I will detail the test flying and many modifications necessary to turn N311U into a functional flier.
Tom Wyatt III has been turbocharging engines for 25 years, but this is his first airplane engine. He is a glider flight instructor, glider towpilot, and he is currently building a Europa. He works as a turbocharging consultant and software application specialist for computer-controlled management systems. (Tom Wyatt died as a result of a traffic accident early in 2006. Ed.)