What's new
Van's Air Force

Don't miss anything! Register now for full access to the definitive RV support community.

Subaru to destruction

clarkefarm

Active Member
I have been persevering with Subaru engines in a 9A since starting with an NSI package about nine years ago. The latest iteration was an MPS 2.6 "stroker" modified EJ253 with a compression ratio of 10.4:1. It had about 5% more capacity than a stock EJ25 and was the subject of some discussion in the forums a couple of years ago when its performance in the now defunct MPS's Glastar was queried.
I finally got around to flying behind this engine which for me was Subaru #6 and was concentrating on fuel mapping to give ROP at 5,000 RPM or higher and LOP at lesser revolutions. I am still using the NSI 2.12:1 gearbox and prop (which has 22⁰ of reverse pitch available).
There has been significant concern about this engine since the first one to fly failed in the Glastar due to piston damage causing the oil to be blown out through the rocker cover breathers.
At about 7 hours total run time I experienced a similar failure in flight and the event was captured on a panel/cockpit camera. Footage is on Vimeo along with some other footage which gives a better look at the farm strip I fly off. Links are https://vimeo.com/82777093 and https://vimeo.com/82713897.
I am ashamed to say that I did not pick up the loss of oil pressure for a couple of minutes (the gauge is the small middle one to the left of the tacho) so expect things to look pretty bad on disassembly.
Other aspects which I should probably mention is that power is controlled by adjusting prop pitch with the L-R function of the coolie hat switch on the stick. You can see the very significant power drop when the engine goes >120⁰ LOP (peak is about 1450⁰) at 5,000 RPM and vice versa.
The landing was not well judged with 90 KIAS over the fence to a downhill strip but winding the prop back into beta always works well because the sprague clutch in the gearbox allows the prop to windmill without impedance and then this energy is used as an air brake as the pitch goes negative.
Earlier in the video you can see the aircraft's maximum performance which is about 163 KIAS at 2,000' although the density altitude on the day was closer to 3,500' which would suggest a KTAS of about 172.
I have thought about starting again with a proven engine/prop configuration but will probably go again with Subaru as I am still enjoying the challenge.
Regards from Australia.
Rupert Clarke
 
Rupert
How about one of these Subaru motors?

Subaru-V12-Boxer-F1-Motor-fotoshowImage-5be74114-481694.jpg
 
Tough news about your engine, but congratulations on a successful landing. Well done!

We all tell ourselves that we?ll pay close attention to the engine parameters, but there are many other things we need to do too when we are flying. This event is a good example of why it is important to have an engine monitor that provides some very alerting indication if an engine parameter goes out of limits. They are more expensive than simple gauges, but they add safety, and can pay for themselves if they allow a problem to be caught early when it may be cheaper to repair.

Good luck with the teardown and rebuilding.
 
Rupert, can you tell us how you were leaning to achieve LOP with the MoTec ECU? How many seconds would it take to go from say 12 to 1 AFR to 17? My concern would be a slow pull through stoich and with this high CR and possibly relatively low octane fuel, you could easily get detonation for a few seconds each time you go through a leaning event.

Were you running 100LL or auto fuel and what octane?

Ignition timing at WOT and above 4000 rpm?

Do you retard spark at high MAP?

Knock sensor fitted?

Spark plug heat range?

My preference on hard use engines is to use 2618 alloy forged pistons which will handle quite a bit of mild detonation without issues usually. The stock cast pistons are known to be quite weak with regards to even light detonation and the 4032 alloy, high silicon forgings also suffer from being relatively brittle compared to 2618 and don't do well in this environment.
 
Last edited:
Rupert congrats on surviving another Subaru event.
There are EGG H6's for sale over here but shipping to down under may not be worth it.
Good luck persuing the challenge - you can always hang a reliable boring Lycoming if it wears off. :)
 
Engine control

2j5gvnb.jpg
Thanks all for the replies.
Kevin, I have a flashing red LED bank which was actually the first thing that drew my attention because it is fairly large. The BMA Efis also throws up out-of-whack engine parameters but the flashing warning is pretty small and the very bright sunlight does not help. Still no excuses because there had been a momentary shudder earlier and I should have been more perceptive.
Ross, lots of interesting questions so many thanks. I have looked at the data log file and taking half second intervals the numbers on lean to rich go Lambda 1.19, .9 and .8. On rich to lean the numbers were .82, .995. 1.14 and 1.17.
I am running premium unleaded which has an Australian RON of 98 but this less than the US values (I think 98 Aus might be roughly equivalent to 94 US). I have set a fixed advance of 30⁰ for anything above 4,000 RPM and do not have any knock sensing (used to but wasn't happy with the complexity). I am using pretty ordinary plugs but will need to go to the hangar to check.
David, the idea of a 6 cylinder might present a bit of a challenge. The NSI package has a horizontal radiator behind the engine and internal cowl ducts to direct air into a plenum above it. (It works exceedingly well as without the thermostat the engine will barely reach 150⁰ at maximum power). Due to this you could already describe the plane as an RV-9A "Long Nose" and I have not been able to reduce the weight on the nose wheel below 310 lbs.
I am also pretty happy with the performance and still have much to do as I have induction losses of 45 millibars (1,500') and an inlet temperature of 10⁰C above OAT because I am drawing air from inside the cowl. I also have a huge 130 square inch cowl outlet which is not required for cooling so I think that I might end up giving the 9's 180 kts VNE a nudge before I am through. Before I get chastised for this the main aim is to be able to cruise at 155 kts burning about 7 gal/hr.
Rupert
 
2j5gvnb.jpg
Thanks all for the replies.

Ross, lots of interesting questions so many thanks. I have looked at the data log file and taking half second intervals the numbers on lean to rich go Lambda 1.19, .9 and .8. On rich to lean the numbers were .82, .995. 1.14 and 1.17.
I am running premium unleaded which has an Australian RON of 98 but this less than the US values (I think 98 Aus might be roughly equivalent to 94 US). I have set a fixed advance of 30⁰ for anything above 4,000 RPM and do not have any knock sensing (used to but wasn't happy with the complexity). I am using pretty ordinary plugs but will need to go to the hangar to check.

Your fuel would seem to be close to our standard premium 91-92 AKI we commonly have available in North America.

I believe with your CR, you could easily encounter detonation during the leaning event to LOP with 30 degrees of total timing although this would be of very short duration judging by the Lambda figures and your data log sampling rates. Heat range 7 NGK plugs would be advisable in aviation use.

When you disassemble, it will be interesting to see the failure mode of the piston. Ring land breakage is almost 100% due to detonation whereas edge distortion/failure of the crown usually indicates pre-ignition or sometimes just general overheating of the crown. Does this engine have under-piston oil squirters installed?

I have seen countless photos of damaged Subaru EJ/EZ pistons now and believe most are due to too much timing and/or too lean mixtures. The factory is very conservative at WOT on both these parameters. I've seen timing on the atmo versions down to 22 degrees and AFRs as rich as 11.2. The jury is out on LOP operation of these engines, you are the first to try this that I have heard of so it was interesting to see your figures.

It seems like your coolant temps and EGTs are under good control.

Aural warnings are the best by far for out of limit parameters. Blinking lights or displays often don't get your attention quickly, especially with sunlight on the panel. Been there before myself and the airliner designers know this well too, hence the aural warning systems prevalent today.
 
Last edited:
WOT settings

Thanks Ross, interesting info.
I was actually tossing up whether to throw a second hand engine from a wrecker's yard into the plane until i was happy with the tuning. i probably destroyed a few engines already by relying on the notion that best power is around Lambda .89 (AFR 13) and setting up WOT 80-100% MAP to run at this. (Naturally I had backed off a bit with this last engine so some disappointment.)
I had thought that if I installed a stock engine (apart from the low price) I could put a car/wagon with a similar motor on a dyno to see what settings Subaru use. You might have saved me the trouble but I guess that I would be passing up some horsepower with an AFR of 11.2 and timing of 22⁰.
I will check to see if the MPS 2.6 engine had under piston oiling.
Regards
Rupert
 
I like my sheep (and I am not from NZ)

Hi Dj
The sheep pay the bills around here so have priority.
They are pretty easy to see and I have a fair idea of what they are going to do, although it gets a little more complicated when the lambs wander off then try to rejoin mum.
I can usually find a bit of the strip unoccupied but in the Spring their calling cards are fairly sloppy and the wheel pants take a bit of cleaning.
If you ever need to use your plane to remove a flock of anything from a landing area don't forget that you need to work on the edge of the flock, if you divide them they will just rejoin by the time you get back.
Regards
Rupert
 
Talking to a guy I'm currently in training with...he's heard several stories of RVs powered by Subarus that have experienced engine failures in flight.:( I'd ditch that thing and go with a Lycoming - an engine actually built for an airplane!
 
Talking to a guy I'm currently in training with...he's heard several stories of RVs powered by Subarus that have experienced engine failures in flight.:( I'd ditch that thing and go with a Lycoming - an engine actually built for an airplane!

I think Rupert is pretty aware of all this, making his position clear earlier in these posts... Let's just keep this discussion on topic here and learn something for those interested in Subaru power for aircraft.
 
Last edited by a moderator:
Amen

I think Rupert is pretty aware of all this, making his position clear earlier in these posts... Let's just keep this discussion on topic here and learn something for those interested in Subaru power for aircraft.

We all know Lycomings are the "tried and proven", but some of us enjoy the "experimental" part of experimental aviation. We can't help ourselves.

I've enjoyed these discussions. I'm sorely tempted to develop a Subaru EE20 diesel FWF, but haven't got the time right now.

Rupert, your RV9A looks great! Could you post some better pictures of your cowling, showing the larger cooling air inlets? I'm currently working on another project (not an RV) which may required modification to the original cowl to allow for more cooling air, and am looking for a sanitary, aesthetically pleasing way to do it.

Thanks
Kurt Goodfellow
RV9 WAM120 turbodiesel, flying since early '08, 480 hours.
 
Subaru cowling

Hi Kurt
I do not have any really good images of the cowl inlets/outlet but these might give some idea of how the ducting works. The radiator is below a plenum chamber which has a fuel header tank and other radiator related plumbing above it.
I have included an image of the LHS inlet which shows an ss screen which is intended to reduce and smooth the airflow over the engine so that the ducted radiator air has some negative pressure.
Ross, I have sent a PM with details of the plugs and piston oiling.
Regards.
Rupert


9tk234.jpg

w0hnrk.jpg

2h6zsjt.jpg
 
Ross

You haven't been updating your test report on the new cooling system. It looks like you were getting very good results. Anything to report?
 
I think Rupert is pretty aware of all this, making his position clear earlier in these posts... Let's just keep this discussion on topic here and learn something for those interested in Subaru power for aircraft.

Easy there...not trying to stir anything up.:)

Engine looks nice with the cowling off, by the way!
 
Ross

You haven't been updating your test report on the new cooling system. It looks like you were getting very good results. Anything to report?

Record snowfall in December, and doing an intercom swap at the moment. I updated my site a few weeks back but don't post much here as it seems of little interest to the majority here on VAF.
 
Easy there...not trying to stir anything up.:)

Engine looks nice with the cowling off, by the way!

You're pretty new here it looks like so you might not be aware of how these alternative engine threads often degenerate into general attacks on alternative engine choices. As a moderator here, just giving everyone a heads up that those comments are not constructive.
 
Looks very nice

Thanks for sharing the picture with the cowl off. The reduction unit looks very nice and very well built. I really like the plenum setup. Seems like you did a very nice job with everything.
 
I updated my site a few weeks back but don't post much here as it seems of little interest to the majority here on VAF.

Ross, that is very true, but also remember that Lycomings are in the majority so it is natural that they would not be interested. Rest assured that there are those of us that are definitely interested! :)

Even a quick post with a direct link to the updates on your website would be appreciated.

Thanks,

-Dj
 
Good Show!

Subie fan or not, you have to appreciate the craftsmanship that went into that FWF build.

I applaud your experimental spirit, Rupert!
 
Rupert;
Thanks for posting, and the videos as well.
As a Mazda Rotarian, I pay close attention to Subies, an 'other' great alternative engine.
It looks to me that you have a well thought out arrangement. My engine required changes along the way to get into the air. I blew it up within hours of receiving the Airworthiness Certificate! (misguided 'aviation mods')
Now I've repaired the engine and had that first flight, but before that, I had to resize my turbocharger. I'm sure more adjustments will be required before I could consider the thing 'finished'.
My only comment: I'm a big fan of Low Compression engines. I know high compression is supposed to be more powerful and more efficient, but Low Compression = Low Stress + increased Reliability. Reliability is a quality all it's own in a single engine airplane.
Keep Posting! I'll read it!
 
First post says NSI gearbox and this is the 6th engine to go south...

Goodness gracious Ross, surely Rupert hasn't blown up 6 Subaru engines in his RV9A since 2004. Are you sure? That just seems totally beyond incredible to me. I'd heard about 4 of his catastophic failures and I thought that number was amazing. If what you say is true then I'm guessing that Rupert may not be getting much more than 50 hours per Subaru engine on average.

I did note when I viewed Rupert's video of his latest engine failure that the oil pressure guage started dropping immediately after he pushed the engine revs out to 6000 RPM. That must be pretty close to redline for a Subaru EJ253. Mind you the engine probably doesn't reach peak power until about 5600 RPM and I noticed that the tacho was sitting around that mark on take-off. You can hear on the video that the engine is simply screaming at those RPMs.
 
Goodness gracious Ross, surely Rupert hasn't blown up 6 Subaru engines in his RV9A since 2004. Are you sure? That just seems totally beyond incredible to me. I'd heard about 4 of his catastophic failures and I thought that number was amazing. If what you say is true then I'm guessing that Rupert may not be getting much more than 50 hours per Subaru engine on average.

I did note when I viewed Rupert's video of his latest engine failure that the oil pressure guage started dropping immediately after he pushed the engine revs out to 6000 RPM. That must be pretty close to redline for a Subaru EJ253. Mind you the engine probably doesn't reach peak power until about 5600 RPM and I noticed that the tacho was sitting around that mark on take-off. You can hear on the video that the engine is simply screaming at those RPMs.

6000 rpm, WOW! But that's how to make power.

The only time I saw over 5000 was taking off with the MT in manual and it over sped, normal TO rpm was 4600. That's probably the formula for a bit of longevity.

I flew behind 2 Subbies for 5 years, 2days. The first H4 ran for 90 hours, the H6 230.3. Some of that flying was fun, but one always wondered when and where the flight would end. It was interesting for a while but the older one gets the less interesting it is. Rupert obviously has not reached that point. :)
 
I've discussed this with Rupert by PM at some length. Being involved in this field for a long time, I've seen this happen many times. This is not the fault of the engine at all as we have literally hundreds and hundreds of Subaru engines flying worldwide successfully with collectively hundreds of thousands of flight hours on them, many going hundreds or thousands of hours before overhaul.

Clearly, 7 hours in not usual nor acceptable.

IMO, every one of these piston failures has a pretty clear cause and usually that is either too much timing and/or too lean mixtures.

I've said this so many times now, I am becoming hoarse: Fuji validates these engines for hundreds of hours at WOT and they don't fail so we must look at what we are doing differently from the OEM.

Subaru backed up the dyno testing with real world proof long ago for you skeptics out there: http://drive2.subaru.com/Spr09/Spr09_20years.htm
 
Last edited:
100,000 miles at 100 miles per hour

Subaru backed up the dyno testing with real world proof long ago for you skeptics out there: http://drive2.subaru.com/Spr09/Spr09_20years.htm

Ford did a similar thing back in 1963. They ran four 1964 model Mercury Comets for 40 days straight at Daytona. Had a big ad campaign at the time (the tagline is the title of this post). I happened to be living in Daytona (as a kid) at the time - and I remember my Dad taking me out to the track to see these cars in action. I don't know if they had any engine failures or not during the test though. ;)
 
Last edited:
Fuji validates these engines for hundreds of hours at WOT and they don't fail so we must look at what we are doing differently from the OEM.

Subaru backed up the dyno testing with real world proof long ago for you skeptics out there: http://drive2.subaru.com/Spr09/Spr09_20years.htm

While I agree that the basic engine should last, as I would expect any modern auto engine to, and I agree that the key is to look for the differences between the OEM's intensions/use and ours, keep in perspective that this test duration was only 480 hrs. They would need to run for 250,000 miles to get to a comparable 2,000 hr. TBO.

Incidentally a couple of guys did fly a C172 for almost 1600 hours non-stop back in the 50's (yes, non-stop). This on an airplane that already had "hundres of hours" on it before the stunt started. They said that when the stunt was complete the engine had gone "significantly beyond tbo", which on an O-300 is 1800 hours. They did terminate the stunt due to low engine power, though, apparently the airplane would hardly climb.

Tim
 
While I agree that the basic engine should last, as I would expect any modern auto engine to, and I agree that the key is to look for the differences between the OEM's intensions/use and ours, keep in perspective that this test duration was only 480 hrs. They would need to run for 250,000 miles to get to a comparable 2,000 hr. TBO.

Incidentally a couple of guys did fly a C172 for almost 1600 hours non-stop back in the 50's (yes, non-stop). This on an airplane that already had "hundres of hours" on it before the stunt started. They said that when the stunt was complete the engine had gone "significantly beyond tbo", which on an O-300 is 1800 hours. They did terminate the stunt due to low engine power, though, apparently the airplane would hardly climb.

Tim

I know of one EJ powered gyro in Australia used for flight training which had 3800 hours on it 3 years ago without being touched internally- not bad I think for an engine putting out around 60hp/L.

Lycomings are well known to usually make TBO and beyond and the GPUs would run 2000+ hours continuously. The idea with an auto engine is the intial costs and overhaul costs are way lower so we can accept a shorter TBO. Most EJ Subes are going 500-1000 hours between overhauls, depending on many factors from the feedback I receive. A $2K Sube requiring a $1000 overhaul every 500 hours will cost about $5K for 2000 hours of use (plus gearbox maintenance). You can see this would be a fraction of a new 320 Lycoming. If it went 1000 hours between overhauls, it would be much cheaper still. Now if we only get 7 hours... not such a rosy picture!

At the same time, we all know someone with a Lycoming who has not got anywhere near TBO without replacing a bunch of parts... expensive parts.
 
Last edited:
The idea with an auto engine is the intial costs and overhaul costs are way lower so we can accept a shorter TBO.

I would be entirely on board with it if this was the total package deal - even considering the additional time and effort for "tinkering" to get the package running right - the bigger problem to me is the unacceptably high risk of sudden stoppage in flight, with the resultant risk (both health and cost). If we could be relatively sure that the decline of the engine would be gradual and predictable, then it would be an easy decision. In reality, that decline is highly subjective to the particular install and that was ultimately what drove me to a standard Lyclone on my airplane.

If I had a very light LSA that I could stall in the 30-35 knot range, and thus be relatively certain of surviving an off-field engine stoppage without major damage, I would probably be inclined to look at it again - but for my 9A it's simply not in my comfort zone.
 
I would be entirely on board with it if this was the total package deal - even considering the additional time and effort for "tinkering" to get the package running right - the bigger problem to me is the unacceptably high risk of sudden stoppage in flight, with the resultant risk (both health and cost). If we could be relatively sure that the decline of the engine would be gradual and predictable, then it would be an easy decision. In reality, that decline is highly subjective to the particular install and that was ultimately what drove me to a standard Lyclone on my airplane.

If I had a very light LSA that I could stall in the 30-35 knot range, and thus be relatively certain of surviving an off-field engine stoppage without major damage, I would probably be inclined to look at it again - but for my 9A it's simply not in my comfort zone.


I know of at least 20 separate Subaru piston failure events and not one has stopped cold. They have continued to keep running on 1, 2 or 3 cylinders and allowed the aircraft to be forced landed with little or no damage in 95% of the cases. Breaking the ring lands or putting a hole in the crown from pre-ignition just gives a power loss and lots of oil out the breathers in most cases. I've broken many pistons in turbocharged race engines in years past and not one of those disintegrated either. In most properly tuned Subarus, the decline is slow and predictable just like any other engine.

I'm gonna try to steer the thread back on course here. Your preference is your preference, I have no beef with what makes you feel safe but again, let's not bring the Lycoming into this discussion as Lycomings don't have much to do with this thread which is about lessons learned on the Subaru. We don't see Sube guys posting in the Traditional Engine Sections when a jug blows up, cam makes metal or a valve sticks...
 
Last edited:
I flew behind 2 Subbies for 5 years, 2days. The first H4 ran for 90 hours, the H6 230.3. Some of that flying was fun, but one always wondered when and where the flight would end. It was interesting for a while but the older one gets the less interesting it is. Rupert obviously has not reached that point. :)

Hi there David. Here's a report from Rupert Clarke in December 2007 describing how he had a Subaru engine completely sieze without warning in his RV9A 7NM out to sea and had to dead stick it all the way back to an island from 8500 ft. Then he had to find a hole in the clouds and dodge mountains to get back to the strip. Pretty exciting stuff. Too exciting for me. I suspect Rupert is not doing any more overwater crossings these days. :D

"I have 110 hrs on an NSI Subaru 9A in Australia. Had a bit of fun about a month ago when things went wrong about 7 NM off the coast of Flinders Island in Bass Strait (famous for ship wrecks).

The NSI motor is a 2.5 block with 2.2 SOHC heads and modified cams. A large horizontal radiator (fed by ducts on the inside of the lower cowl), header tank and oil/water heat exchanger are positioned behind the engine. Unfortunately mine was one of a couple of engines with mismatched cams. Cams designed for a friction rocker arm were supplied with engines with roller arms and there is a 3? difference in the keyway position (as well as some rollers not being centered over their lobes). The result was a 50 psi difference between cranking compression of the 1 & 3 and 2 & 4 cylinder banks. The advice from NSI was that this would not matter when the engine was running but I think that it may have led to my engine failure which resulted from the engine making metal and also possibly a MAP sensor failure.

I had climbed to 8,500' for the trip home from the island and had just reduced power by coarsening pitch to slow the engine to 4,500 rpm and set the AP when I had a few "misses" which rapidly became worse. Switched to the backup ECU and fuel system with no result so turned back to the island and pulled the throttle. The engine quit but because the NSI gearbox has a sprague clutch the prop kept spinning so unless you looked at the tacho it felt perfectly normal.

The plane glides well - say 1.8 NM per 1000' at 90 KIAS so it was easy to make land but I had to dive thru a smallish hole in the cloud about a mile abeam the strip and was a bit limited by the surrounding mountains from making a wide approach so had way too much speed over the threshold - about 110 KIAS. The strip (thankfully) was 4,000' but downhill and I was beginning to wonder 2/3 rds of the way along but managed to pull up OK at the end."
 
Hi there David. Here's a report from Rupert Clarke in December 2007 describing how he had a Subaru engine completely sieze without warning in his RV9A 7NM out to sea and had to dead stick it all the way back to an island from 8500 ft. Then he had to find a hole in the clouds and dodge mountains to get back to the strip. Pretty exciting stuff. Too exciting for me. I suspect Rupert is not doing any more overwater crossings these days. :D

"I have 110 hrs on an NSI Subaru 9A in Australia. Had a bit of fun about a month ago when things went wrong about 7 NM off the coast of Flinders Island in Bass Strait (famous for ship wrecks).

The NSI motor is a 2.5 block with 2.2 SOHC heads and modified cams. A large horizontal radiator (fed by ducts on the inside of the lower cowl), header tank and oil/water heat exchanger are positioned behind the engine. Unfortunately mine was one of a couple of engines with mismatched cams. Cams designed for a friction rocker arm were supplied with engines with roller arms and there is a 3º difference in the keyway position (as well as some rollers not being centered over their lobes). The result was a 50 psi difference between cranking compression of the 1 & 3 and 2 & 4 cylinder banks. The advice from NSI was that this would not matter when the engine was running but I think that it may have led to my engine failure which resulted from the engine making metal and also possibly a MAP sensor failure.

I had climbed to 8,500' for the trip home from the island and had just reduced power by coarsening pitch to slow the engine to 4,500 rpm and set the AP when I had a few "misses" which rapidly became worse. Switched to the backup ECU and fuel system with no result so turned back to the island and pulled the throttle. The engine quit but because the NSI gearbox has a sprague clutch the prop kept spinning so unless you looked at the tacho it felt perfectly normal.

The plane glides well - say 1.8 NM per 1000' at 90 KIAS so it was easy to make land but I had to dive thru a smallish hole in the cloud about a mile abeam the strip and was a bit limited by the surrounding mountains from making a wide approach so had way too much speed over the threshold - about 110 KIAS. The strip (thankfully) was 4,000' but downhill and I was beginning to wonder 2/3 rds of the way along but managed to pull up OK at the end."

One of the reasons why I tell people to keep the internals OEM... Seen lots of cam, spring, valve and rocker trouble caused by using poorly designed and poorly implemented aftermarket valvetrain parts- all the factory validation goes out the window. You can't blame Fuji for stuff like this any more than you can blame Lycoming for ECI barrel separations.

There is no evidence that aftermarket cams are of any useful benefit on the 4 valve EJ engines in the typical 4000-5000 rpm operational range- just money and reliability down the drain IMO.
 
Good to see this thread. I didn't know about the Subaru speed record test. IMO the Subaru is a great choice for an alternative engine. The engine parameters are getting to be well known (thanks to the testing to failure).

I have 635 hours on mine, and so far no changes to compression, oil pressure, or oil analysis results. I'm very happy with it and hope to get another 1000 hours or so out of it. Then I'll buy a new one from Subaru. I have the original Subaru ECU, so I feel confident that the engine will not suffer from detonation. I might be leaving some power on the table, but I don't need any more power.

-Andy
 
So what is working for other people?

The best combinations for 160-200 hp seem to be EJ22Ts (factory turbocharged) and EJ257 short blocks (STi) mated to EJ25 SOHC heads, either atmo or turbo. The stock EJ20 closed deck turbo engines and stock EJ257 engines with DOHC heads are also good if a bit heavier.

The only internal changes which may be advised would be long skirt, 2618 alloy forged pistons fitted at about .0035 clearance and possibly top quality stainless or Inconel valves offered by vendors such as Cosworth, Supertech or Ferrea. It appears that aftermarket bearings such as ACL, Cosworth and Mahle last at least as well as the OEM ones.

There is no need to change the cranks, cams, rods, oil pumps etc. which are very well designed and proven.

The STi also suffered from some separated oil pickup tubes for a few years resulting in many destroyed engines.

The most important thing of all is ECU tuning and this is why most piston failures occur. Subaru messed up themselves in 2007-2008 with some bad mapping and rather foolish closed loop code resulting in multiple piston failures on the STi models via detonation, even though knock sensors were installed.

These two major problems resulted in a Subaru USA Stop Sell Order until the causes were determined and remedied- very embarrassing for Subaru and showing that even a company with such massive resources and validation can get things very wrong sometimes.

In aircraft use, we'll suffer the same failures if detonation is present so mapping must be very conservative with regards to AFRs and ignition timing. One Crossflow EJ lasted only about 18 seconds according to the data log files I reviewed when the unfortunate user applied takeoff power. The mapping provided by the manufacturer was completely wrong.

At full power, we don't want to see AFRs any leaner than 11.5 or so and for most atmo EJ/ EZ engines with stock compression ratios, no more than 26 degrees total timing on 91 octane fuel.
 
Last edited:
Ross is right

I have not yet dismantled the engine but am pretty sure, looking at the plugs, that my ignition advance of 30⁰ has caused the engine to run high cylinder temperatures/pressures. I expect to find a melted piston. Mea Culpa.

I have purchased a low mileage EJ257 engine and as it is a later model it contains (I think) Subaru's I-Active Valve Lift System which takes it from 165 to 173 hp in the car. I will definitely not be pushing the tuning until I have run a vehicle with a similar engine on a dyno to see what values Subaru use at 5,500 - 6,000 RPM maximum sustained power.

For the record I do not think that 6,000 RPM is impossibly high. I understand that EJ25s can run to 7,500 RPM before valve float although I limit my 9A engines to 6,300 and would usually not exceed 6,000 (which is very smooth). The gearbox is an NSI/MPS A40 and has a 2.12 to 1 ratio so 6,000 RPM translates to 2,830 at the prop. The one engine I did put on a stand and dyno was a rebuilt NSI EJ25 with EJ22 heads. Compression was 9.31 to 1. It made 163 hp at 6,000 RPM and 154 at 5,500. At the normal cruise setting of 4,800 RPM (2,264 at the prop) this engine was making about 138 hp.

Most of my real engine failures (I have carried out a few precautionary rebuilds/replacements) have occurred soon after a full power climb and usually involve rod (big end) bearing failure which suggests excessive pressure/detonation. Perhaps the biggest handicap has been my perception that an engine makes best power around an AFR of 13, when at high power settings this is right in the worst part of the "red box" or "red fin".

The information now available on operating ROP/LOP is an eye-opener on how to avoid high pressures/temperatures so I look forward to adopting these principles.

As for over water flights, they are on hold until I get the tuning sorted but then no big deal. The 9A is a joy to glide and land power off.

Rupert
 
Good to see this thread. I didn't know about the Subaru speed record test. IMO the Subaru is a great choice for an alternative engine. The engine parameters are getting to be well known (thanks to the testing to failure).

I have 635 hours on mine, and so far no changes to compression, oil pressure, or oil analysis results. I'm very happy with it and hope to get another 1000 hours or so out of it. Then I'll buy a new one from Subaru. I have the original Subaru ECU, so I feel confident that the engine will not suffer from detonation. I might be leaving some power on the table, but I don't need any more power.

-Andy

Nice to see this report, building up trouble free time. :)
 
I have not yet dismantled the engine but am pretty sure, looking at the plugs, that my ignition advance of 30⁰ has caused the engine to run high cylinder temperatures/pressures. I expect to find a melted piston. Mea Culpa.

I have purchased a low mileage EJ257 engine and as it is a later model it contains (I think) Subaru's I-Active Valve Lift System which takes it from 165 to 173 hp in the car. I will definitely not be pushing the tuning until I have run a vehicle with a similar engine on a dyno to see what values Subaru use at 5,500 - 6,000 RPM maximum sustained power.

For the record I do not think that 6,000 RPM is impossibly high. I understand that EJ25s can run to 7,500 RPM before valve float although I limit my 9A engines to 6,300 and would usually not exceed 6,000 (which is very smooth). The gearbox is an NSI/MPS A40 and has a 2.12 to 1 ratio so 6,000 RPM translates to 2,830 at the prop. The one engine I did put on a stand and dyno was a rebuilt NSI EJ25 with EJ22 heads. Compression was 9.31 to 1. It made 163 hp at 6,000 RPM and 154 at 5,500. At the normal cruise setting of 4,800 RPM (2,264 at the prop) this engine was making about 138 hp.

Most of my real engine failures (I have carried out a few precautionary rebuilds/replacements) have occurred soon after a full power climb and usually involve rod (big end) bearing failure which suggests excessive pressure/detonation. Perhaps the biggest handicap has been my perception that an engine makes best power around an AFR of 13, when at high power settings this is right in the worst part of the "red box" or "red fin".

The information now available on operating ROP/LOP is an eye-opener on how to avoid high pressures/temperatures so I look forward to adopting these principles.

As for over water flights, they are on hold until I get the tuning sorted but then no big deal. The 9A is a joy to glide and land power off.

Rupert

Most Sube users seem to be limiting max continuous rpm to about 5500 and cruise 4000-5000 rpm. A turbo allows you to chop the rpm downwards about 1000 rpm due to the higher torque. I'm a big advocate of lower rpm with a little boost. The frictional losses are lower and performance is maintained at altitude along with lower and more pleasing noise levels. The Inkster and Crother's turbo EJ257 RV7s are blistering fast and sound great.

Best power is usually around 12.5 AFR. Extra piston cooling can be had in the mid 11s with a small drop in power. At 75% power and below, we should be able to safely run mid to high 12s. LOP maybe but someone must be the guinea pig.
 
Last edited:
VAF folks,

I decided to go ahead and post this email here as well as send it on to Ross, sort of true confession time I guess.... Please don't use this to bash those who continue to experiment with auto engine conversions. It takes a lot of time and effort and persistance to get those engine packages working properly and my hat is off to those who do so successfully.

Hi Ross,

I have been feeling like I owe you an email for awhile to let you know what is going on with my project. The short version of the story is that I blew the heck out of my engine with severe detonation and I am now in the process of converting over to a Lycoming clone of some kind.

The end of my Subaru came on very early climb out from my home airport in Sedona. I was returning to a project I was working in Colorado and had made a few adjustments to the turbo that morning. I had been fighting boost control issues all along and had to be very careful to advance the throttle slowly as the power would come on very quickly and the engine could easily get way ahead of the propeller pitch change. So I was advancing the throttle very slowly. As I got airborne I felt it was time to get moving and went ahead and pushed it all the way forward as I had done successfully in the past, only this time when I did, the power surge was very strong and I felt a very strong and hard vibration in the airframe, letting me know that something very bad was happening.

There was not a lot of runway left and I was up a ways, so I needed to gain enough altitude to make the big turn to get back, or put it in the cactus in very steep terrain. I had automatically pulled the throttle back when I felt that shudder and it came back again when I pushed the throttle back in to gain the altitude. I think I went through a few more cycles of throttle up, feel the shudder and pull it back. Without actually thinking about it I turned out to the the right a ways before bringing it all the way around to the left to line back up with the runway and I got it down about two thirds of the way down with out any issues. It was nice to have the AOA during that turn:)

So what happened was full and uncontrolled boost and I cannot really blame anyone but myself for it as I found a single loose nut on the turbo that probably caused this particular problem. It was a little 3/16" nut that clamps down against a spring to lock the two halves of the turbo together. With that nut loose, the two halves rotated a tiny bit, and jammed the boost control rod from any further movement once the vanes got into the full boost position. I had accomplished a very tricky task of placing a new boost control spring in the diaphragm controller, and I managed to get the tiny little circlip back on the rod, without taking the cowl off and I was kind of proud of that accomplishment, but it was a big mistake. I did not really take time to perform the work properly and that little short cut could easily have killed me. The nut was out of sight and difficult to get to with the cowl on and I simply did not reach in there and feel if it was tight...

I was making that change to the boost control spring because, on the flight down from Colorado, it was stuck in position and I could not adjust boost with the bleeder valve. On the flight home I had about 30" MAP and cruised along at 150 knots, about the same performance I had without the turbo. I had hoped while putting the stronger spring in place I would be able to free up the vanes and that is what led me to loosen that clamp so I could rotate the halves a little.

So in the end, literally I guess, my poor Subaru experienced 55" MAP on auto fuel and that did a fine job of taking out cylinder number two. Draining the oil showed lots of shiny aluminum pieces had been distributed throughout the engine. The turbo only turned with some effort. Oil got pumped into places it did not belong.

My wife was there to see me off and that did not help my situation much. That was the second time she saw me come back and make a landing trailing heavy smoke. I had promised her that if I had any more serious problems with the Subaru that I would give up on it, and I am in the process of keeping that promise.

I have not shared this on the Sube News forum, I guess because I hate to discourage those folks any more than they already have been.. I think I will post a version of this out on VAF to sort of get the news out.

Well, that is my update. I have all the Sube parts in the corner of the hangar and have no idea what I will end up doing with them. It is hard as **** to justify spending another $45K on my airplane to make this conversion but that is what I am doing. I seriously considered many other options. (Including just buying an RV4 for less than the cost of this conversion). If the Aerocharger folks could have made me feel more confident in solving that boost control problem I might have given it another try. I thought about using a manual cable to move the control rod but they said I would have to mess with it continually. I also considered mounting an external wastegate to the exhaust system and they talked me out of that idea also. As I started pulling parts and pieces from the engine it just felt right to keep going and strip everything right on back to the firewall.

I sure appreciate all the help you gave me along the way. I put way too much faith in that Aerocharger turbo and unfortunately, the fit was so tight that a Garrett would not go into the same location, making that option rather difficult.

If it is any consolation for you, I do plan to continue to use your system to control the Lycoming replacement engine via Robert Paisley's EFII setup. I have even been considering whether to go with two of them for some redundancy, but I am not sure if adding the wiring and switch to control which unit fires the injectors is worth the added complexity of it all.

I am hoping that after installing four different Subaru engines in this airframe that converting to the Lycoming will be a little easier for me:).

Randall Crothers
 
Randy,

Sedona does not offer many options, glad you made it back ok.

You sure gave this challenge a good go, no one has pushed the envelope like you have or hung in there as long trying to get it to work, except Ross.

It is disappointing that so many guys were left with nothing to show for their time and money trying to get one of these engines to work reliably. Locally, my friend Ron Burnett has also tossed in the towel, not sure exactly what went wrong but his Egg H4 is history. He made a safe landing at his airport with an internally damaged engine. I was so relieved it was not off field. I have discontinued trying to help guys get these things to work, I just do not have the answers.

At some point we are faced with quitting flying or moving on to something that is not a continuos test flight. The notion of going auto conversion and saving money is a long standing myth.

NO one has tried to get a Subaru to work and not spent as much or more than going with a Lycoming. Period!

Hope the "reversion" goes well with you. I went Barrett with the -7A and am doing likewise with -8. They are good people. Good luck.
 
Randy,

Sedona does not offer many options, glad you made it back ok.

You sure gave this challenge a good go, no one has pushed the envelope like you have or hung in there as long trying to get it to work, except Ross.

It is disappointing that so many guys were left with nothing to show for their time and money trying to get one of these engines to work reliably. Locally, my friend Ron Burnett has also tossed in the towel, not sure exactly what went wrong but his Egg H4 is history. He made a safe landing at his airport with an internally damaged engine. I was so relieved it was not off field. I have discontinued trying to help guys get these things to work, I just do not have the answers.

At some point we are faced with quitting flying or moving on to something that is not a continuos test flight. The notion of going auto conversion and saving money is a long standing myth.

NO one has tried to get a Subaru to work and not spent as much or more than going with a Lycoming. Period!

Hope the "reversion" goes well with you. I went Barrett with the -7A and am doing likewise with -8. They are good people. Good luck.

I think everyone who failed still learned plenty just like anything else in life, depends on how you look at it. We learn as much from failures as we do from successes in life, the failures just aren't as happy.

Certainly many users had lofty and possibly naive visions based on marketing hype. Others have piled up many trouble free hours like Andy in this thread. These engines simply don't tolerate "abuse" the same way a Lycoming engine will so they are unforgiving in that regard and literally many dozens of people have experienced that the hard way.

Subarus have been flying successfully for literally hundreds (thousands) of people worldwide for many years, accumulating hundreds of thousands of flight hours, notably in the gyro world. These certainly saved a lot of money over a Lycoming just as I have- thousands earned in investments by not laying down an additional $20K 10 years ago.

What we can learn from Randy's experience and any other is that even the smallest detail, part or mistake can have serious consequences. We need to be aware of each one, address it and be as careful as humanly possible.

In this case, the overboost situation could have been negated by having a MAP cut entered in the ECU. I have mine set at 38 inches on Mogas, in case of a wastegate failure. I also run very conservative timing and AFRs- vitally important on these engines when operating on Mogas. 55 inches on Mogas would have blown up a Lycoming just as surely as this EJ or Scott Emery's Wankel. You can't blame this on the engine itself.

If anything is important in the auto conversion world, it is probably best to exactly copy what has worked for long time flyers of these engines. A successful recipe should turn out the same each time just as it has for Lycoming engines. A good example of this is the over 600 RAF gyros fitted with EJ engines. All done the same way and work very reliably for most of the users for nearly 20 years now.
 
Last edited:
Back
Top