Very rich Subaru EJ253 NA fuel values

I have just watched Webinar 065 (Planning AFR values for NA engines) and I wonder if I have made a mistake with the high end values for a 2009 Subaru Forester NA SOHC motor which is driving a variable pitch aircraft propellor so may run at WOT for extended periods.

The attached file shows the values extracted from a similar engine on a dyno (the vehicle had some collision damage so the values may not be totally reliable) which appeared to show lambda .74 for 100 kPa at 5,500 and 6,000 rpm.

This seems very rich compared to the 350Z and I wonder if the EJ25 series of motors is a special case for tuning at higher power levels.

I am using a Motec M130 for engine control so have access to knock diagnosis although the engine is pretty noisy, mainly in cylinder 4.

It's totally believable that the delivered fuel could be 0.74 Lambda, particularly if there were compensations due to engine temperature. They mention going to 226 deg F ECT toward the end!

Thanks David.

Earlier reply seems to have disappeared but I do appreciate the advice.

It would be really helpful to have an experienced tuner's guide to the sort of peak lambda I should be aiming for before compensations are applied at the 5,500 and 6,000 rpm maximum load area of the Fuel Mixture Aim table.

I am also confused about how to use the Inlet Air Temperature Fuel Mixture Aim table. Unlike the Haltech ECU used in Webinar 074, the Motec M130 appears to be looking for a stand alone lambda value for temperature, without reference to load, rather than the Haltech's percentage compensation.

What is altered to the engine mechanically from its original form when it was in a forrester?

The long motor is stock 2009/2010 Forester S including the AVLS two position inlet cams. The inlet manifold is off an earlier EJ25 with a cable operated throttle but uses the 2009 MAF/IAT sensor in its original housing. The injectors are non-Forester from a 2004 WRX. The fuel pump is non-standard as is the exhaust system. Coils are LS1s with coil per plug.

Pressure recovery from the intake is quite good with instances of positive MAP over ambient. However MAP falls quickly as the aircraft climbs (about 5 kPa a minute) and there is also the potential for negative pressure in the exhaust system as the tailpipes are short and protrude into a slipstream which can approach a continuous 300 kph.

The application would probably have benefitted from using MAF as the efficiency parameter and I was somewhat disappointed to discover that this is not possible with the M130. I may still swap the Motec out for a Haltech Elite 1000 but will see how much progress I can make over the next few weeks.

I attach an image which includes the 2.12:1 gearbox and also shows the plenum above the horizontal radiator.

I've actually been involved with the tuning of a Subaru EZ30 engine that was converted for aircraft use and my tuning technique was quite similar to how I would deal with the same engine in an automotive application. Peak power in this instance ended up being achieved around 0.88-0.90 lambda and under cruise conditions we could achieve great fuel economy by tuning to lambda 1.00. I ended up choosing to tune a little richer in the end due to the reduction drive having the engine constantly operating around 4000-4300 rpm under cruise. Given that full power climb in an aircraft can last several minutes, it may also be wise to consider this for your full power operation and also target slightly richer, however you shouldn't need to be anywhere near 0.74 and tuning that rich will sacrifice power. The fact that you're dealing with an engine that is N/A and offers a relatively low specific power level means that you're not going to be quite so worried about controlling combustion temp with the additional fuel.

A properly calibrated MAF has the benefit of directly measuring mass airflow, and hence accounting for air temp and barometric pressure somewhat automatically, however properly configured, I think you'll find the M1 to do a great job. You have the option of adding a baro sensor to do compensation for altitude if desired, and the closed loop fuel control is excellent to take care of any small errors you may end up with.

In the M1 ECU the air temp lambda target can be used to override the main target table. The ECU looks at all the lambda tables and compensations and will use the richest target of any of the tables.

Thanks Andre

I agree that Lambda 0.74 seems crazy rich but the US EJ25 tuners have settled on AFRs of 11.5 for continuous full power and I think that they (SDS) are talking about non-turbo too. The aircraft engine fuel management wisdom has also defined an area to avoid in maximum power operation based on EGTs (between 200°F rich and 60°F lean of peak EGT) but of course they are dealing with fixed timing. My EGTs look reasonable (low 1300s°F) at full power with 5,500 rpm being a bit higher than 5,000 rpm.

I have had bearing problems with four EJ25 engines, probably too much timing, but going forward I might still stay fairly rich, say .82 but then enrich with a trim activated by 30 seconds or so of continuous high load, however I cannot see the ability to do this in my M130 package.

I do have an ambient MAP sensor and compensation for altitude would be good but I am yet to locate the channel which enables this. I am also a bit perplexed by the Fuel Mixture Aim channels for Coolant Temperature and Inlet Air Temperature which require lambda entries but without an engine speed axis.

I agree that closed loop control is a feature of the M1 and also like the knock monitoring.

Thanks again for a great resource.

Just a bit of a follow up and expression of surprise about how significant a change in ambient pressure is for this engine.

I became aware of how the Motec M1 "Fuel" (read VE) values were jumping around depending on whether the aircraft was on the ground or moving along 750 ' higher.

At 2,850' higher with an ambient pressure of 88 kPa compared to 95 on the ground, the VE for MAP 75 kPa, RPM 4,500 jumped from 96.7 to 104.9 - about 7.55%. (I am yet to look at other values or go higher.)

This would be a good explanation of how I have trashed 5 engines in 190 hours using a Motec M4 which was not running closed loop and probably had values around .92 lambda as a starting point.

Maybe the effect is not so pronounced on turbo motors ?? but this would be a bit of an incentive to complete the Ambient Pressure axis in case the lambda sensor has a bad day.

Hi Rupert, the MoTeC M1 ECUs include an Average Load and Race Time function which can be used to modify aspects such as ignition and fuel delivery based on how hard the engine is being used. You can check these out in a webinar here: https://www.hpacademy.com/previous-webinars/044-average-load-and-race-time-functions-motec-m1/

If you have a baro sensor connected to the ECU then you can set the VE table up with a 4th axis for barometric pressure to correct any error that may be present. You don't need to get too in depth with this and a single zone at 95-100 kPa and then another at perhaps 70-75 kPa will normally suffice since the ECU will interpolate between. The closed loop fuel control can then pick up any small errors that may remain.

As far as the fuel mixture aim tables, there is the main target table which you use for the actual tuning, and then a number of compensation tables that can adjust the final lambda target. The coolant and air temp tables are simple 2D tables by default but if you press the 'A' key you can enable additional axis options to give you more control. The ECU will constantly look at all the tables and use the richest value from them at any particular time.

Thanks Andre.

It is a relief that I do not need to go into too much detail with the ambient pressure fourth axis and thanks for the Average Load and Race Time Webinar ref.

I have played around a bit already with the ancillary fuel aim maps and managed to frighten myself a bit with the ECU unexpectedly picking up on the EGT one. I set it up on the basis of the existing aircraft probes which run around 1280 to 1340°F but the Motec sensor was indicating 1420 and so triggered its fuel aim map. I should have realised what was going on but Exhaust Temp is not included in the default logging.

I attach an image of the log file which I think indicates the fluctuations due to the above. I have now disabled all the ancillary fuel aim maps until I have the main table worked out.

With what you're doing, you really should have the level 1 logging in the M1 (especially if you don't have another MoTeC logger) You need to be able to log the parameters you're tuning! Alternately, have your laptop connected when testing, and at least you can then go back and look at the last few minutes of running (and see all the channels, but at a fairly slow logging rate).

Thanks David

I think that I have level 1 logging or something similar. At a guess I would have about 130 channels to choose from in i2 but not Exhaust Temperature and to date I have not worked out how to include it.

However I can include it in the "Time" recording easily enough and have been doing that. Changes in EGT are relatively slow so I am probably OK with this.

Rupert

My mistake -- yes, I meant you needed to add Level 2 logging, which will allow you to choose additional channels (up to 200 including the ones already logged as part of diagnostics) and record them at up to 200 hz. This is almost a requirement if you aren't logging the CAN bus channels, or are trying to setup a specific subsystem (gear shift, knock control, multiple lambda aims, or engine speed limits, etc). You need the data to learn what's going on.