Tracy Crook, Real World Solutions 5500 NW 72 Way Bell, FL 32619 (386) 935-2973 Email : tcrook@rotaryaviation.com www.rotaryaviation.com EFI Design for Aircraft This article will detail the experiences and design philosophy behind the EFI/Ignition controller that I developed for the Mazda 13B rotary in my RV-4. I call it the EC2; EC for Engine Controller, and 2, because it is the second version I came up with. It wasn't a distrust of EFI that led me to discard the factory system on the Mazda 13B chosen to power my RV-4. Modern EFI systems have proven to be very reliable and I had originally planned to use it, but that was before I had actually seen one. When the truck delivered the used rotary engine I had ordered, I immediately had second thoughts. After acquiring some books on the subject and studying modern automotive fuel injection systems for a while, I saw that the design considerations for car use are vastly different than for aircraft. The Mazda rotary intake system was even more complex than most and the myriad of emission control systems alone was enough to discourage me. Since space is always a scarce commodity under the cowl, the tall manifold and bulky mass airflow sensor didn't help either. The final nail in the coffin was its weight of about 50 pounds. Still, there were a lot of advantages to fuel injection that I reluctantly gave up when I installed carburetors on the engine. The three Mikuni sidedraft carbs served me well for three years and 650 hours of flight but their shortcomings began to be more and more obvious. Even though they were CV (constant velocity) type carbs that compensate for altitude, their ability to control the mixture was limited to about 8000 feet. The desire for higher cruise altitudes and the lust for more horsepower started me thinking about EFI once again. I am a retired EE (electronics engineer) with 30 years of hardware and software design experience so the task of designing one seemed doable. PICTURE HERE (Engine4BW.jpg) This is an early photo of my RV-4 engine installation showing the three Mikuni carburetors. They mixed over 4000 gallons of fuel with air and flew over 100,000 miles before being replaced with an EFI system. Before deciding to roll my own, I evaluated several aftermarket EFI systems but rejected all of them for various reasons. Some of them require a laptop computer to setup or tune and it was (and still is) my opinion that the cockpit of an airplane in flight is no place to be fiddling with a PC. It might be argued that this could be done on the ground, but I have found that it is absolutely critical that the system be tuned under actual flight conditions if best fuel economy is to be achieved. Another missing feature was the lack of a manual mixture control on the aftermarket systems. When I mention this, many builders respond that they thought that lack of a mixture control was an advantage of EFI. I suppose it is in a car, but a little study of the subject will show this not be the case with aircraft. FADEC - SCHMADEC This might be a good place to comment on the term "FADEC" which seems to be all the rage now. Several companies are developing EFI systems for both homebuilt experimentals and certified aircraft that claim to be FADEC systems. The term is an acronym that stands for 'Full Authority Digital Engine Control', and proponents trumpet its great advantage of 'single lever power control'. The idea does have some appeal, but let's take a hard look at the subject. FADEC was originally developed for high-end applications like fighter jets and rocket engines where reducing pilot workload is crucial. The goal is to have the engine respond to pilot input in direct proportion to power control inputs and at the same time, maximizing engine efficiency. To do this requires that the engine controller have 'full authority' over all engine parameters. In the world of engines we are interested in, this includes manifold pressure. Manifold pressure is primarily controlled by the position of the throttle plates so if the controller is to have full authority, it must have control over the throttle. This means that a 'fly by wire' system is needed. The power lever that the pilot moves is just an input to the controller, which in turn positions the throttle plates with a servomechanism. Now here's the point. None of the so-called FADEC systems I have seen have this feature. The pilot's throttle control is linked directly to the throttle plate in all cases that I have seen. This may not seem significant at first glance, so let me illustrate with an example. It will also explain the answer to why I chose to have a manual mixture control on my controller. Let's say we are cruising along at 10,000 feet and our intention is to cruise at the most fuel efficient setting. Best engine efficiency will be achieved with the throttle plates wide open (to minimize pumping losses) and the mixture aggressively leaned. The rotary is especially well suited to this condition because there is no need to worry about burning pistons or exhaust valves. Now let's say we encounter a situation that requires maximum power, such as the need to climb over a mountain range or ridge of clouds. Or perhaps we have blundered into a box canyon and desperately need to climb out before we hit the wall. To get more power from the engine under these circumstances, the only option is to richen the mixture. The difference in power between best economy mixture and best power mixture is substantial. If we are flying one of these pseudo FADEC systems, what are we going to do? The throttle lever is shoved as far as it will go already and there is no mixture control (single lever power control, remember?!) The bottom line is that these systems have to be programmed to give either best power or best economy or a compromise between the two. You can't have both and still have 'single lever power control.' I was unwilling to give up either power or economy so a manual mixture control seems like a small price to pay to get both. MAF or SPEED-DENSITY? After deciding to dispense with the idea of single lever control, it was time to move on to choosing the basic type of control mechanism for my EFI design. There are currently two types in use. Most automotive systems in use today use the Mass Airflow (MAF) scheme. If we know the absolute mass of air that the engine is breathing, we can calculate the appropriate amount of fuel to add in order to get an appropriate mixture. This approach is elegantly simple because we do not need to know anything about cubic inch displacement, how fast it is turning, or volumetric efficiency of the engine. Theoretically, a good mass airflow controller would work on virtually anything from a Briggs & Stratton lawnmower engine to a Pratt & Whitney R4360 radial engine and require very little tuning between engines. The downside to MAF is that the sensor required to measure airflow is rather large (vane type) and in the case of the hot wire type, fragile. The hot wire type measures airflow by the cooling effect the air has on a thin wire, which is being heated by passing an electric current through it. The other system in current use is called Speed - Density, which determines the amount of fuel to inject by measuring several parameters. The main two factors are RPM of the engine (speed) and manifold absolute pressure (density). If we also know the cubic inch displacement of the engine, we can then calculate the approximate amount of fuel to inject. The calculation is complicated by variations in volumetric efficiency. These variations are due to tuning effects of the intake and exhaust manifold, cam timing and overlap (or port timing in the case of the rotary), and air temperature. The temperature is easy enough to measure with a sensor but it would take a super computer the size of a car to compute all the other factors affecting volumetric efficiency in real time. Despite these apparent disadvantages of speed density, this system was chosen for my design due to the small size, low weight and low cost of the required sensors. These factors were especially important because I had decided that redundancy was a must for my controller. There are two of everything, including sensors. This problem of variable volumetric efficiency can be side stepped by using what is called a "look up table". The controller performs its calculations assuming a volumetric efficiency of 100%. For example, if a four-stroke engine of 1 liter displacement has turned two revolutions and the manifold pressure is 15" Hg (1/2 of atmospheric pressure), we would calculate that the engine had breathed in ½ liter of air (at atmospheric pressure). The actual amount would be affected by the factors already mentioned, so the controller would then refer to the look up table for information on how to compensate for these factors. In aircraft applications, the manifold pressure most accurately defines the engine's operating condition. Since the engine load is a propeller, if we know the manifold pressure, we also know approximately how fast it is turning. This is unlike a car where the engine could be at full throttle (high manifold pressure) but only turning at low RPM (lugging up a hill in high gear for example). For this reason, the lookup table in the EC2 is a list of numbers corresponding to different manifold pressures. We will refer to it from this point on as the MAP (Manifold Absolute Pressure) lookup table or "MAP table" for short. There is one entry in the table for approximately each ½ inch of manifold pressure. To illustrate, if the engine is running at 20" of manifold pressure the EFI controller calculates the theoretically correct amount of fuel to inject, but before opening the fuel injector it looks up the 40th number in the MAP table (hence the term 'lookup table') and adjusts the mixture according to the number it finds there. KISS Since simplicity is one of the factors that attracted me to the rotary engine, I decided this would have a high priority in my EFI design as well. Some of the things I decided to eliminate were the throttle position sensor, engine coolant temperature sensor, O2 sensor, EGR control and idle speed control. This reduced the number of wires in the system to a fraction of the number used in a typical automotive installation. In addition to making it simple to install, it also increases reliability. If it isn't there, it can't fail. The reasoning behind eliminating each system will provide additional insight into the design philosophy. The throttle position sensor is used mainly to sense rapid changes in throttle setting. This is important in a car where the throttle may frequently and quickly go from idle to wide open. There is no need to do this in a plane where changes in power are smooth and gradual and the engine spends 90% of the time at the same throttle setting. Hence, there is no need for this sensor. PICTURE HERE (EC2bd.jpg) The simplified aircraft EFI/Ignition controller is small enough to fit two complete controllers along with switching relays to select the active controller on a single board. Even the manifold pressure sensors (bottom) are separate. Let the pilot do some of the work Even though it is not good practice, a car engine is often asked to start in sub-freezing temperatures and immediately be driven down the road. To give good drivability under these conditions, the automotive engine controller must automatically compensate for wide variations in engine temperatures. This makes the coolant temperature sensor a necessity. In an aircraft, the engine should always be properly warmed up before takeoff so this sensor was considered unnecessary. Some method is still needed to enrich the mixture when doing a cold start so I opted for a cold start switch on the control panel. This along with the manual mixture control allows for starting the engine in most any weather. It does require the pilot to do a little more than just turn the key but is still easier than the priming required on most aircraft engines. More than you wanted to know about Oxygen sensors The Oxygen sensor is used in automotive EFI to keep the engine operating at stoichemetric mixture, which is important for the proper functioning of catalytic converters. Since we don't need catalytic converters, this was another part to get the axe. Another benefit of loosing the O2 sensor is being able to burn avgas, which is a leaded fuel. Lead impairs the functioning of O2 sensors and even with careful flight planning of fuel stops with Mogas, you will end up having to use 100LL avgas from time to time if you do any cross country flying. Leaded Avgas will be gone in the not too distant future but for now we have to deal with it. It is frequently assumed that an EFI using an O2 sensor for mixture control will always be running at optimum efficiency. This not correct, especially in aircraft applications. In fact, the operating point that these sensors were designed to operate at is almost never the point you want in an airplane. During takeoff you want maximum engine power in order to minimize takeoff distance and runway requirements. You also want to accelerate as quickly as possible and achieve that first 1000 feet of altitude in order to minimize that vulnerable period shortly after takeoff, even if it means burning an extra pound of fuel to do it. Maximum power requires a much richer mixture than the O2 sensor was designed for. Even automotive EFI controllers ignore the O2 sensor when you put your foot to the floor to pass that semi truck. Once established in cruise flight, minimizing fuel burn becomes the top priority. This requires running on the lean side of peak EGT. Proper leaning technique on standard aircraft engines is a subject of much debate. Some claim that a typical Lycoming engine must be run rich of peak (ROP) in order to avoid burning exhaust valves and other ills, while others claim that running 50 degrees lean of peak (LOP) does no harm. Since my standard aircraft engine experience was limited to the 40.2 hours spent getting my pilots license, I don't know which is true. I do know that on the rotary engine I fly now, there is no limitation at all on lean running. There are no valves to burn and the iron rotor is in no danger of being burned or seizing in the aluminum rotor housing. Getting back to our discussion of oxygen sensors, I have found that best economy is obtained at a fuel/air mixture well below the operating point of these sensors, which again points to the conclusion that it would be a poor choice for the primary mixture control mechanism in aircraft applications. Having said all this, there is a useful function for O2 sensors when used with an inexpensive mixture monitor. I used one of these to do the initial tuning of my EFI. This made it much easier than trying to do it with EGT readings. More EPA Stuff Keeping the catalytic converter happy and carefully controlling peak combustion temperatures (for emissions reasons) are among the reasons for the extensive control systems related to exhaust gas recirculation (EGR) and idle speed control. Both functions can be dispensed with in aircraft use. How Many Wires? After one look at the complex systems and wiring harness in most cars the thought of installing an EFI system is intimidating to many builders. However, after eliminating all the unneeded systems and simplifying the rest, the job becomes quite manageable. There are only a total of 30 wires from the controller that I eventually designed, and this includes the wiring to the small control panel. Between the controller and the engine bay, there are a total of 12 wires needed for a typical 13B rotary installation. All of the switching for the backup controller is done internally by relays to keep the external wiring simple. The manifold absolute pressure sensors are also internal to the controller so the only connections to these are two manifold pressure vacuum lines connected to the intake manifold. Automatic Backup? A frequently asked question is "Will the EC2 switch automatically to the backup controller in the event of the primary controller failing?" I looked into doing this but decided against it for a couple of reasons. First, the additional logic required to do it introduced possible failure modes of its own. Second, the additional hardware to do it properly was more complex than the controllers themselves. Have you ever noticed that more space shuttle missions are scrubbed due to failures of the failure detection system and sensors than to actual system failures? I decided that the pilot should be capable of deciding if the controller might have a problem and hitting the backup controller select switch if necessary. I have never had a controller failure, but I do practice my engine emergency procedures regularly. You want to be able to do this without thinking when the time comes. Use what you have As stated before, the stock engine controller was not considered appropriate for aircraft use but many of the mechanical parts of the system were perfectly adequate. The stock fuel injectors, fuel rails, pressure regulator, and some parts of the intake manifold and throttle body were used as-is or in modified form. PICTURE HERE (Engine stand.jpg) Pictured here is the engine on the test stand I used during development of the EFI controller. The intake system is in mostly stock form except that much of the emission control systems (about 20 pounds worth, not needed for aircraft use) have been removed. Even after this, the stock intake manifold is a complex three-part assembly that stands up quite high. Several (but fairly easy) modifications are required to fit it under the cowl. The PSRU shown here is the prototype of the one currently flying on my RV-4. Its development is a long story for another day. PICTURE HERE (Fuel rail.jpg) PICTURE HERE (lower manifold.jpg) The secondary fuel rail and injector assembly was modified and moved from the stock location on the intermediate intake manifold to the lower manifold (shown on right) to lower the profile. The injectors now squirt into the holes formerly occupied by the secondary intake port valves. These valves normally only open at mid to high throttle in automotive use. Since the engine is always running at mid to high throttle, these valves (and actuators) can be eliminated. PICTURE HERE (upper manifold.jpg) As shown here, the intermediate manifold has been cut off right after the 90 degree bend over the top of the engine. Thinwall aluminum tubes were then bonded into the manifold and connect to the home made dynamic chamber (bottom). The top of the chamber is not installed in this photo in order to show internal details. The throttle body is mounted to a semicircular plate that covers this chamber. Total intake runner length should be in the ballpark of 16 - 18" for best performance. Some of the terms used in this article may be unfamiliar to those without rotary engine experience. I publish a book (Aviator's Guide to Mazda rotary Conversion) for those looking for a crash course on the subject. PICTURE HERE (Thrbody.jpg) This is the stock Mazda 13B throttle body before and after cutting away all the unnecessary metal and bracketry. Only 2 of the three barrels are used. Over 4 pounds of metal was cut away. PICTURE HERE (Engine-L.jpg) Here is the mounted manifold and throttle body view from the left side of the engine. The throttle body is fed from a ram-air scoop (not shown) on top of cowl. EC2 EFI/Ignition controller is not visible in these photos because it is mounted on the cockpit side of the firewall to keep it away from high temperatures. There are very few under-cowl wires in this EFI system. Note the use of aluminum foil and thin stainless steel to shield paint and hoses near the exhaust headers. PICTURE HERE (top front.jpg) Front view of EFI manifold installation. Note the bracket on the left side of dynamic chamber that is clamped to the oil filler tube. This eliminates stress on the long intake runners. Some builders react with horror to the use of hard lines used to feed oil pressure to the PSRU and use of automotive hoses. Contrary to popular opinion, hard lines are perfectly acceptable as long as there is no relative motion between the two ends and they are supported at intermediate points to avoid long unsupported sections that could vibrate. See any turbine or fuel injected aircraft engine for examples. My rule on automotive hoses is never to re-use a piece of hose. If it is removed, it is replaced. This installation now has about 1100 hours on it. Test Results The increased power was obvious from the moment I advanced the throttle on that first flight test. I had estimated that horsepower would increase from 160 to 180. Twenty horsepower does not sound like a lot but the difference can be readily felt in the seat of your pants. I personally do not put much store in dyno testing for our purposes. A dyno is an invaluable tool when doing serious engine development but that's not what we are doing. If you just want a plus or minus 5% figure, use a software program like those available on www.bgsoflex.com (it's free) to estimate power. I used this to estimate my original carbureted engine HP at around 160. I am satisfied that this was close because my performance was very close to other RV-4s using the O - 320 160 HP Lycoming. My new estimate with the EFI is about 180 HP and this seems justified because I can now keep up with the O - 360 180 HP Lycoming powered guys. The increased HP is mainly due to increased volumetric efficiency as a result of intake tuning. I am still using the same prop, so part of the increase comes from being able to turn the engine faster. Max engine RPM went from 6250 to 6500 which puts my propeller tip speed at the very highest you want to use. I need a slightly higher pitch prop now. The improvement in efficiency was small but measurable at low altitude where I got a reduction of about 3% fuel burn at any given airspeed. There is no free lunch of course and if I use the increased power I also get an increased fuel burn. Maximum power fuel burn went from 15.5 to 18 GPH. Climb rate went from 2000 fpm to 2300 and top speed at sea level increased from 204 to 214 mph. While top speed and vertical climb figures are good for bragging rights, they have very little practical use in everyday flying. Burning 2.5 GPH to go 10 MPH faster is a waste of gas. The real payoff was in efficiency at higher altitudes. This alone made the effort worthwhile. At 14,500 feet I got a 20% improvement in fuel burn. That works out to 160 extra miles per load of fuel at economy cruise! Two factors were involved here. First, of course, was the fact that I can now optimize the mixture but a second factor was that the EFI controller was programmed to boost ignition advance at low manifold pressures. Later experiments with better exhaust headers and mufflers resulted in further improvements to BSFC. A full discussion rotary engine exhaust systems could take several pages so I'll hold that for another time. If you find yourself looking for EFI controllers, reduction drives or just looking for more information on aviation use of rotary engines, check my website at WWW.rotaryaviation.com. Happy flying Tracy Crook