The unabridged version of the article appearing in the April 2005 issue of Kitplanes.
Subaru vs. Lycoming
Are Subaru engines really a viable alternative to the tried and true Lycoming? Racetech Inc. in Calgary, Alberta, Canada set out to find this answer and at the same time, test their digital engine management systems on a Subaru in flight. First, a suitable test aircraft for the engine had to be found. It was decided that a Van's RV6A would allow the best comparison as hundreds of O-360 powered ones are already flying. A quickbuild kit was purchased from Van's Aircraft in September 1999 and construction began. Construction proceeded steadily although it was found that there were numerous errors, omissions and mistakes in the plans and manual. Fortunately, we made our way through this and Van's always provided timely and efficient service. The latest pre-punched RV kits and CAD drawings are a vast improvement and should eliminate most of the concerns that we had with our earlier vintage kit. The pre-cover inspection was completed in September 2001, paint was applied in October 2002, assembly was complete in April 2003 when engine ground testing was started. Around 1600 hours were expended in completing the airframe and about 400 on the engine installation. On November 30, 2003, C-GVZX made its first flight.
Being a race car fabricator and engine builder professionally, I've never really liked certified aircraft engines. They do the job but seem crude and outdated by modern automotive standards. The cost of them is the big drawback from my perspective. A new Lycoming O-360 with propeller is in the $30,000 range. Our firewall forward installation came to less than $9000. Overhaul costs of the Subaru would be less than 1/10 that of the Lycoming also.
After much research on Subaru engines, it was decided that the North American EJ22 turbo was the best starting point to give the desired 180-200 hp reliably. These were readily available and bulletproof. The EJ25 available at the time was rejected due to its open deck cylinder design, making it less suitable for turbocharging. Although several aircraft Subaru modifiers claim to be getting up to 50 more hp than the stock naturally aspirated rating of 130 hp, this is highly unlikely in the real world at any reasonable rpm applicable to aircraft use. Forced induction was deemed necessary to achieve the desired performance. The factory rating for the turbo engine was 160hp at 5600 rpm. The EJ22 turbo was well engineered for the stresses of turbocharging with different pistons, wrist pins, oil cooler and piston oil cooling jets. It also featured milder camshaft timing than the Japanese domestic market EJ20 engines available at this time, offering the possibility of better fuel economy.
Inside the EJ22T
Of course I had to tweak the stock engine to my specs. Our engine had 26,000 miles on it when it was received. The EJ22T is a 2200cc four cylinder, horizontally opposed , liquid cooled engine with 4 valves per cylinder actuated by belt driven, single overhead camshafts, rocker arms and hydraulic adjusters. Cast iron cylinders are cast into each aluminum block half. The heads were match ported and flow benched. I designed some 9.45 to 1, 2618 T-61 alloy forged pistons to replace the 8 to 1 factory cast pistons to try to boost hp and thermal efficiency. These were custom made by JE in California and are the only non stock parts inside the engine. The forged crank is a real jewel, supported by 5 main bearings, only 13.5 inches long and 20 pounds. The short stroke design permits a stiff crank with .750 inch throw to main overlaps. The rods are forged with 9mm bolts and a 5.25 inch center to center dimension. Having worked on many different types of engines in the past, I was very impressed with the design and metallurgy of the EJ22. It's short, stiff and strong without some of the wimpy components used in some other brands. The long block weighed 188 pounds with no accessories or intake manifold. Total installed weight with all accessories, redrive, turbo, intercooler, exhaust, rads, and coolant is around 345 lbs.
The engine was reassembled using the original main and rod bearings as these were in mint condition. Valves and seats were ground with the seats being done slightly wider than stock for improved heat transfer. I fabricated a custom intake manifold from steel tubing to mount four Bosch 390cc/min. top feed injectors in place of the factory side feed injectors. The manifold mounted the factory 60mm throttle body and features internal velocity stacks and a lower profile to clear the cowling.
I sized a Garrett T3 turbocharger for best performance between 8-15,000 feet as we are based near the Rockies. I chose a Super 60 compressor, Stage III turbine wheel and a large .82 turbine housing with the integral wastegate. This combination suits the engine much better than the stock turbo at altitude and higher continuous power settings with a cruise compressor efficiency of 74%. The turbo is rigidly mounted to a solid structure bolted to the top of the engine block. The exhaust headers are slip jointed in two planes to avoid transmitting any thermal expansion stresses to the turbo or mount. They are constructed from .063 thick 321 stainless tubing. The turbine outlet pipe is 2.25 inch, .058 wall, 304 stainless. A Spearco intercooler core measuring 9 X 6 X 3.5 inches cools the intake charge. The factory EJ22T was not intercooled.
Fuel and Spark Control
Much discussion usually centers around redundancy of the fuel and ignition systems on automotive conversions. On ours, we have none. I can hear the gasps now. Here is my logic on the matter: There are two magnetos on aircraft engines because magnetos are not all that reliable and fail more than once in a while. Dual ignition is also provided to light off the mixture on these large bore engines. The Subaru has a centrally located spark plug and a small bore. Twin spark plugs are not needed here. Spark plug fouling simply does not happen these days on modern auto engines. I find it interesting that many people from the certified engine camp are horrified that someone would fly a single ignition, computer controlled engine yet these same people will fly VFR at night or over desolate terrain in their single engine, single propeller, single crankshaft, single carburetor, single oil pump equipped aircraft. (I won't fly night VFR). In other words, they accept the risk of all the single points of failure on the certified engine under conditions where they would almost certainly be dead if anything failed on that engine. Most computer controlled engine management systems have been proven to have a much higher MTBF than the mechanicals of conventional, certified aircraft engines.
Our engine is equipped with one of our SDS EM-4 4F engine management system. This is a digital, programmable unit with crank triggered, direct fire ignition and fuel injection control. Sensors include manifold pressure, crank position, coolant and air temperature and throttle position. No oxygen sensor is required so operation on 100LL is no problem. Racetech has sold thousands of these systems for auto use where they have accumulated over 11 million hours of operation. In addition, many units have been sold for aircraft use over the last 10 years. These have accumulated over 15,000 flight hours to date. One of our bench test computers has run continuously for over 45,000 hours now. Fuel and spark parameters are adjustable with the engine running via a hand-held programming box containing an LCD screen. The full engine map can be changed and gauge modes read sensor outputs.
Reduction Drive and Propeller
A Marcotte M-300 redrive with a 2.2 to 1 ratio is used. This unit bolts up directly to the Subaru. It uses rubber drive bushings to absorb and isolate torsional vibration from the propeller. The gears are of the internal helical spur type, putting the drive gear inside a ring shaped output gear. This has the advantages of increasing clearances with temperature, better retention of lubricant, more tooth contact, same direction prop rotation as the crankshaft and quieter operation over normal spur gears. Lubrication is by bath rather than engine oil. We are very pleased and have had no issues with the redrive.
The propeller is a composite IVO Magnum 3 blade, 76 inch diameter. It is the high pitch version electrically in-flight adjustable between 45 and 105 inches of pitch. The prop weighs only 27 pounds. Some people have described problems with earlier IVO designs fitted to conventional aircraft engines. Other armchair "experts" have ridiculed its design and performance. We find no merit to these concerns and have had no issues with our propeller. It works as advertised and is simple, inexpensive and lightweight.
I rigged the engine carefully in the airframe to duplicate the stock offset angle used with the Lycoming and fabbed up a sturdy mount from .049 wall 3/4 inch 4130 tubing. This was TIG welded on a jig. The engine is bolted to two lower bed sub mounts and the main mount is attached via custom urethane bushings to them plus two separate upper mounts. The mounts weighed 21 lbs. vs. the stock Lycoming mount at 14 lbs. 83 hours were spent designing and building the mounts.
Cooling Trials and Tribulations
I knew that cooling the turbo engine might be challenging. That certainly proved to be the case. The engine was heavily instrumented with multiple temperature and pressure probes running through an SDS developed engine monitor. We originally hoped that we could reduce cooling drag using a liquid cooled engine compared to the Lycoming. This proved not to be the case with our installation. The deep spar on the RV6A makes running coolant lines internally to a belly mounted radiator very difficult. As such, it was decided to mount the rads inside the cowling and use the existing air inlets. I would not do this again. After many changes, theories and problems we have made this layout work but it likely has higher drag and lower heat rejection than a well designed belly scoop might.
The end result is that we have 3 rads mounted in the cowling. One GM evaporator core is mounted behind the left inlet cheek, one Ford aluminum heater core with electric fan mounted ahead of the nose gear, fed by a 3 inch boundary layer scoop below the spinner and a custom barrel shaped heat exchanger mounted in the lower right cowling fed from the right cheek duct. We now use Evans NPG+ non-aqueous propylene glycol coolant which has a 375F boiling point and eliminates any problems with vapor bubbles. We found it necessary, through cowling pressure measurements, to increase the cowling air exit area by installing an exit ramp and ejector duct on each side of the cowling. We were getting insufficient pressure differential across our rads for proper heat dissipation. Some of these problems are due to the short duct length provided by mounting the rads under the cowling with less than ideal pressure recovery. Coolant flow is in series through 3/4 inch silicone heater hose. We don't run a thermostat after encountering countless problems using them in this engine. A Toyota heater core and valve is mounted in the cabin for colder days. Over 100 hours were spent measuring, thinking, changing, cursing and testing to get the cooling system to function without creating a huge drag penalty.
We lost 2-3 quarts of coolant on one winter test flight due to shocking the thermostat closed by opening the heater valve and having a restrictor in the bypass hose. All initial test flying was done 2500 feet over the airport so by throttling back and nursing it down, we made it safely down under power all the way. Oil temp rose to 120C, power off, as the coolant spiked to 120C then started down as there was none left to cover the sensor. I knew we had lost a lot of coolant as the heater started blowing cold air at the same time. We flew for about 10 minutes without coolant. After a compression test and oil inspection, no damage was found, proving the strength and design of the Subaru design and the amazing properties of Mobil 1 synthetic oil. Test flying started to lose its entertainment value after this and I started to think maybe Lycomings weren't so bad afterall!
In addition to the water cooling issues, we had oil temperature issues to deal with. Oil temperatures were getting too high in the climb, especially at higher density altitudes. Due to the design of the Subaru engine which is basically a big block of aluminum with oil and water in close proximity of each other, when the oil got hot, the coolant followed and vice versa. We finally had to use the original turbo air inlet duct to blow air over the oil pan and add another NACA duct to feed the turbo. An Earls 10 X 10 X 2 inch oil cooler is fed by the right cheek opening via a Mocal oil thermostat.
Finally, we found that our intercooler setup as originally installed did virtually nothing to cool the charge air with a measured effectiveness of less than 10%. Again, pressure measurements using our sensitive gauges showed almost no differential across the intercooler core. This meant no airflow, hence no cooling. More inlet area was added until the original area was tripled. Then we found that internal cowling pressures rose as all this additional air could not escape. This in turn caused coolant and oil temperatures to start to rise again. More fun, more mods. The intercooler core was sealed to the firewall and exit tubes were run through the firewall to exit out three custom made ducts. We finally worked the core effectiveness up to 50-62% depending on density altitude and power setting. Again, many changes were made in this area to arrive at a workable solution.
As there is no published data for the EJ22T for aircraft use, we have set conservative limits initially for all pressures, temperatures and rpms until we get more hours on the engine.
For normal takeoff power we use about 4800 rpm and 38 inches, resulting in about 170hp. For climb, we use 4600-4800 rpm and 35 inches which is around 150-155 hp. Max cruise power is 4400-4600 rpm and 30 inches resulting in around 130 hp.
EGT is limited to 1400F which is well on the rich side of peak but designed to not stress the automotive type turbine wheel. Maximum oil temperature is 248F, max coolant temperature 230F and max intercooler discharge temperature is 131F.
Performance and Fuel Burn
Our RV6A weighs in at 1136 lbs. empty (with paint, interior, pants, oil, coolant, ELT etc.) which is about 30-50 lbs. heavier than the average O-360 powered one using the constant speed prop. Gross weight is set at 1750 to allow a reasonable useful load and not stress the airframe and landing gear too much. The C of G turned out ideal. The aircraft is impossible to load out of limits with legal baggage. Weight is the enemy of performance but our RV6A numbers are in the ballpark with published figures for heavy O-360 powered RV6As. Climb rate is highly dependent on weight and temperature. In the winter at 4000 feet MSL solo, light fuel, I've seen 2300 fpm sustained. At 1750 lbs. in the summer at 4000 feet MSL, we see around 1000-1200 fpm at 80-85 knots IAS. With the turbo, rate of climb does not diminish so fast with altitude. We still see over 600 fpm at 15,000 feet MSL in the summer at gross climbing at 90-100 knots IAS, leaned to 1350F and 12.1 gal./hr.
For speeds and fuel flow, we see the following at density altitudes of 13,000- 17,000 feet (GPS verified):
4400 rpm/ 25 inches/ EGT 1400F- 153-158 knots TAS- fuel flow 7.8 gal./hr.
4600 rpm/ 30 inches/ EGT 1400F- 170- 175 knots TAS- fuel flow 9.7 gal./hr.
Lycoming powered RV6As can match these speed figures at lower altitudes. At around 8000 feet we true about 10 knots less than the figures above. We attribute the lower speeds to the higher cooling drag of our cowling configuration over the very clean factory design and our lack of some of the gear leg intersection fairings on our RV during testing. Our SFC is on the order of .46 lbs./ hp/ hr. at cruise power settings which is slightly inferior to the smaller Lycomings at. .42- .46. The higher frictional losses of the Subaru turning 4200-4600 rpm largely contribute to this over the slow turning Lycoming. I'd dispute any claims that liquid cooled engines are going to better the "ancient" air cooled engine in fuel flow at the same speeds in an RV. With the higher operating temperatures of the Lycoming and higher delta T than a liquid cooled engine, basic thermodynamics tells us that we'll need more airflow to cool things. More airflow likely means more drag. Perhaps with a well designed belly radiator, we could achieve close to equal cooling drag to the Lycoming setup but this is unproven.
Head to Head with the O-360
It's sometimes difficult to believe what you read when it comes to aircraft performance so we decided to test our Subaru powered RV6A directly against a Lycoming powered one. My friend Les Davenport has a very clean 6A fitted with an O-360 and Hartzell constant speed prop, so we arranged a day in November 2004 to have a fly off between the two aircraft. With two people and cameras on board, we headed for uncontrolled airspace. Our aircraft, VZX weighed in at around 1658 lbs. at takeoff. Les' ZRV weighed around 1590 lbs. due in part to different fuel loads.
We leveled out at around 6000 feet to take some photos and then both initiated a climb at 85-90 knots. Initially Les pulled away as I had to reset my prop pitch to get the desired 4600 rpm. Once this was set, we evened out in climb rate with Les about 300 feet above us and a 1/4 mile in front. After another minute, we slowly started to catch up. We arrived at 9000 feet within a few seconds of each other. This contest was pretty much a draw.
We then went for a level speed check side by side. I set 33-34 inches and 4600 rpm, Les set WOT and played with prop pitch a bit. Les pulled away at about 1-2 knots once we stopped accelerating and our GPS readings were within 1-2 knots. I normally use 30 inches for max cruise so I would hand this contest to Les and the Lycoming with a probable 5-8 knot advantage.
We then started a cruise climb at about the same speed to 12,000 feet. Once speed was stabilized, we started to draw up on Les and out climb him slowly. At our normal climb power of 35 inches and 4600 rpm, we arrived at 12,000 feet maybe 10-15 seconds ahead of Les. The turbo Subaru was slightly superior in climb rate above 10,000 feet.
Setting my normal max cruise of 30 inches and 4600 rpm, we slowly accelerated. Les selected his WOT and best prop pitch. Holding altitude as closely as possible we were pretty much neck and neck. GPS readings were called out back and forth and were within 1-2 knots. Fuel flow was virtually identical. This contest was essentially a draw. Pulling power back to our medium cruise of 27 inches and 4600 rpm, we saw Les start pulling away, our GPS reading about 4-6 knots less.
All in all, the test demonstrated that there is little to choose from between the two powerplants. The Lycoming is faster below 10,000 feet, the two engines very equal above this altitude. With equal weights, the engines would provide very similar climb performance, the Subaru perhaps having an edge above 8000 feet. We did not run the engine outside our previously established limits to win this contest as this would prove little. No doubt with higher manifold pressures, the Subaru would trounce the Lycoming in all contests but at the expense of reasonable engine life perhaps. Engine life on the Subaru in aircraft is not well established yet but we hope to have some data on that aspect in a few more years.
The Subaru is superior to the Lycoming in a few areas: initial cost, lower overhaul costs, no carb heat and noise levels. While I can't justify the price tag of a new Lycoming and still don't really like them, I can't dispute the fact that they do the job well for most people flying their RVs below oxygen altitudes. The amount of engineering, fabrication and testing/ modifying involved to make our installation viable is beyond the scope and capabilities of most builders. Firewall forward packages, offering Subaru power such as Eggenfellner's, save much of the engineering, fabrication and development but are also quite costly simply because of all the work already done for you here. Performance of his supercharged EJ25 engines would appear to be similar to our turbocharged EJ22 or the Lycoming.
Is Van's correct to say that the Lycoming is the best choice to power their RVs? This depends on your perspective. If you fly mostly below 12,000 feet, can afford the price tag and want to just bolt it up and fly a known quantity, the Lycoming is hard to beat. Let's face it, 90% of builders fall into this category. If you want something different at a somewhat lower price, a Subaru firewall forward package might suit you. Eggenfellner's are quite well proven now and are about the same amount work to install as the Lycoming. If you cringe at $20-30K for an engine and prop, dislike certified air cooled engines, are really good with engines and love R&D, I won't dissuade you from doing your own auto engine conversion. Some of us just want something different but be aware that this is a LOT of work.If I had to do it all again, would I chose the Subaru? We'll, knowing what I know now, I figure I could do it better next time. It's hard to beat the proper sound of the Subaru mixed with the altitude performance of the turbo. Yes, it's a keeper for me and I not only learned a lot about aircraft building aircraft but I gained a priceless education in the areas of radiators, cooling, ducting, airflow etc. Most importantly, I have something truly different from the other thousand or two RVs flying around. That being said, this is not the type of project/ ordeal that most builders would be capable of doing successfully.
As a correction to the article in Kiplanes, Van's recommends a normal gross weight of 1650 lbs. for RV6As, not 1600 as stated.
We are continuing to develop our EJ22T installation for better performance and lower fuel flows. Watch our Flight Testing section on our RV6A Page for current updates.