A New Name on the Alternative Engine Block

Maxwell Propulsion takes over the old NSI line of Subaru conversions and changes everything but the engine.


Among the concerns of builders who select alternative powerplants for their homebuilts are a few that have absolutely nothing to do with the engines themselves. The top two are probably: Has the company done a good job of engineering any components unique to the installation and is its business model sufficiently sound to ensure the company’s survival? Lets face it: The engine manufacturer will likely have an installed base sufficiently large to support long-term survival.

But what about those components unique to the installation, all those cool firewall-forward bits, special systems, one-off castings? What happens should the manufacturer go out of business or, almost worse, stay in business but abandon an earlier design with no economically feasible way for the builder to get current and complete or maintain his airplane?

Builders using the NSI system of Subaru-engine conversions found out when that company failed in 2005 after months of reported difficulties by builders who could not get complete packages as promised. In the normal course of events, this would mark the end; any NSI installations yet to run and waiting for final parts would be doomed, as would any flying examples that needed repair or replacement parts. Warranty? Good luck.

Enter the Maxwells

Dr. John and Gwen Maxwell didn’t intend to get into the alternative engine business but, as the landlords of the old NSI facility, they found themselves with the assets of the company in place of the rent. Rather than let the concern die and the existing customers twist in the wind, the Maxwells started the long task of learning everything they needed to know about the industry and the product the previous company had been selling. It was, they now say, an eye-opening affair.

After an interim crew, Gwen Maxwell has taken over as chief operating officer, with the staff filled out by Craig Woolman, Colin Gillespie, and Dominic Acia, who operated a popular Subaru-tuning business until joining the company. Its important to understand that while the company is expected to be profitable in the long run, the Maxwells have other investments that allow them to develop the components of the firewall-forward kits without the pressing need to get parts out the door just the keep the lights on. They can avoid cutting engineering corners or using the deposit on firewall-forward kit #20 to pay for engine #15.

Core Values

Maxwell starts with new Subaru EJ25 engine components purchased locally. Components is the key term here, because this is not a totally off-the-shelf setup. While the old NSI conversion had many special parts and machining/assembly processes, the Maxwell conversion is, by design, much simpler. It starts with the engine block slated for the WRX STi automobile, a deep-breathing, four-cylinder opposed-piston design capable of more than the 300 horsepower it makes in the sports car. (Subaru sports car fanatics are known to turn up the boost and ask for a lot from this engine family.) The aluminum block features a stronger casting with internal risers inside the water jackets; they significantly improve the strength of the block assembly. These cases are one piece from the crankshaft split line to the junction of the cylinder heads-in all, a typically modern, weight-efficient, production-smart design.

However, the STi uses double-overhead cam, four-valve-per-cylinder heads, which are both heavier and overkill for an aircraft conversion expected to operate normally from 4200 to 4800 rpm, well under the nominal redline for the engine. Four-valve heads can have more airflow at high rpm than two-valvers, resulting in more power at the upper end of the rev band, and a double-overhead-cam layout affords more accurate control of the valves at high revs. But the DOHC layout of the STi engine is unnecessary in this application, so instead the Maxwell conversion employs the lighter four-valve, single-cam heads. Belts drive the cams from the firewall end of the engine.

In the bores are forged pistons with the turbos modest 8:1 compression ratio; the normally aspirated versions of the EJ25 typically use 10:1 pistons. This reduced compression ratio serves two purposes: providing additional detonation margin in an engine expected to work much harder than it does in a car, and making a later turbo model a simpler upgrade. Unlike the previous NSI iterations, the Maxwell-tuned Subaru is almost completely stock parts-no special cams or valves. (NSI was in the habit of mating EJ20 heads to the EJ25 bottom end, with many specialized parts.) As a result, continued maintenance of the engines should be possible with or without Maxwell Propulsion still in the picture.

Most of the induction system is stock-save for the specially designed and locally produced throttle body, which is much simpler than the stocker because it carries no emissions-control gear. The exhaust is aircraft-specific, as it would have to be to avoid the heavy, power-sapping catalytic converters from the car.

Systems Savvy

Its one thing to bolt a car engine onto the front of an aircraft, but another to make it happy there, and Maxwell has continued to develop the Subarus support systems. To ensure proper cooling, a large radiator stretches across the firewall of the firms GlaStar Sportsman test mule; cool air is fed from the standard inlets along twin channels outboard of the engine. The Subie is so much narrower than an angle-valve Lycoming that there’s ample room for these ducts. This is much like the old NSI setup, and Maxwell is considering options as development of the two turbo models (195 hp and 240 hp) progress.

Spark and fuel are delivered electronically. Maxwell uses twin injection and ignition computers (engine control units) paralleled. Only one operates at a time, but failure of one simply means reversion to the other fully featured channel; because they have equal capabilities, you really cant call one primary and one secondary. A third box, the Engine Management System controller, manages both ECUs and the electric propeller. The system remains singular in places. There is one spark plug and fuel injector per cylinder, bone stock Subaru items that have a proven life in cars.

Maxwell has done something clever here. When you are ready to install your kit, send the company the desired harness lengths. Youll receive professional looking, strain-relieved fully tested harnesses in return. Given that many amateur-builder miscues involve wiring, this is a great service.

Between Prop and Engine

Where NSI arguably had the most trouble was with the propeller speed reduction unit (PSRU). Maxwell has completely abandoned the old design and started fresh. The 2.13:1 reduction is achieved through a simple, helical-cut spur gear, the drive shaft at crankshaft level, and the prop-drive (driven) shaft just above it. Yet, the devil remains in the details. Maxwell tested several iterations to check on damaging harmonics between the prop and the engine (and airframe/engine mount). The final iteration starts with a full-size flywheel/starter ring. Eight elastomeric bushings are pressed into an aluminum carrier, which are then bolted to the flywheel. The carrier uses a central pilot bearing extending into the crankshaft bore for alignment. The drive gear rides on the splined end of this carrier shaft where it slides into the overhung PSRU body.

Prop loads are borne by a massive shaft carried fore and aft by roller- and ball-element bearings. And the entire PSRU is bolted to the bell housing of the engine by a thick machined-aluminum plate and anodized spacers. (The starter mounts to the engine plate as well, this time an off-the-shelf item, not a modified part.) Maxwell had examples of the old NSI gearbox on hand to hold and photograph next to the current parts, and the differences are astonishing. If material size and overall beef are among the ingredients for a good PSRU, the Maxwell is off to a good start. Rated TBO is 1500 hours with a total weight of 72 pounds.

Flying It

Rated for 165 hp at 5500 rpm, the Subaru causes the factorys Sportsman to perform as youd expect. Startup is turn-the-key simple, and the engine settles into a pleasant idle. Runup is fairly routine except you have to do a quick confirmation of the prop setting. The current system, with an electric pitch-change mechanism in a proprietary hub mated to Whirlwind composite blades, is not automatic. It acts like a fixed-pitch prop that you can adjust in flight.

Initial takeoff performance is reasonably good with near-full-fuel and two aboard (estimated at 2050 pounds all up, against a max gross of 2350), with the engine note turning strident at the takeoff setting of 5200 rpm. With obstacles cleared, you can pull back to 5000 rpm, which quiets some of the din and nets a 700-fpm rate of climb. (For perspective, my own Sportsman, with 215 hp, would be doing twice the climb rate at 100 knots instead of 80 indicated.) Cooling performance is very good, easily maintaining 180 F coolant temp and 215 oil temp at 4900 to 5000 rpm. Fuel flow in the climb started out at 14 gph and was 10.7 gph at the top of the climb to nearly more than 6000 feet (density altitude of 8000 feet). All of this was on a day with moderate (below 70) ambient temperatures.

We performed cruise checks at altitude. The process of setting up for cruise is simple, though not fully automatic. As you level off, let the airspeed creep up, toggle a bit more pitch into the prop to keep close to the desired engine rpm. Naturally, the airplane accelerates a bit more, so you must repeat the process until, finally, equilibrium sets in. At 4800 engine rpm, the Maxwell Sportsman shows 116 knots indicated, 130 knots true on 9.9 gph. A two-way GPS run verified the speed calculated on the Dynon EFIS. Pull back to a more modest 4250 engine rpm, and you’ll see a true airspeed of 122 knots on 7.4 gph.

The Inevitable Comparison

How does that compare with a Lycoming? Given that a 180-hp Sportsman is capable of 137 KTAS on 10 gph (give or take), the Maxwell Subaru is in the ballpark. Current ECU mapping keeps the Subie from going aggressively lean at higher power settings, which gives the Lycoming a slight edge here. If the Subaru sacrifices a few efficiency points, it at least gives back ease of operation-aside from the need to set the prop pitch before takeoff (and once more back in the pattern before landing).

Costs round out the equation, of course, and here the Maxwell is in line with traditional powerplants. A package price of $29,540 gets you the engine, propeller, systems and electronics. Add around $2300 for airframe-specific engine mounts-currently available for the GlaStar and two-seat, side-by-side RVs. Firewall-forward, your costs are going to be close to a new Experimental-class IO-320 Lycoming, constant-speed prop, governor and accessories. In theory, reduced overhaul and ongoing costs tip the bargain in favor of the Subaru.

Perhaps the big story here is simply the revival of this firewall-forward program as an example of a good idea failing to die off, and illustrating how to take on the inarguably intense development process with a level head and good business practices. After all, a great idea, poorly executed, never really overcomes that handicap.

For more information, call 360/474-8118 or visit www.maxwellpropulsion.com.

Previous articleDecember 2008
Next articleDown to Earth
Marc Cook
Marc Cook is a veteran special-interest journalist who started as a staffer at AOPA Pilot in the late 1980s. Marc has built two airplanes, an Aero Designs Pulsar XP and a Glasair Aviation Sportsman, and now owns a 180-hp, steam-gauge-adjacent GlaStar based in western Oregon. Marc has 5000 hours spread over 200-plus types and four decades of flying.


Please enter your comment!
Please enter your name here

This site uses Akismet to reduce spam. Learn how your comment data is processed.