anybody leaned out the closed loop target tables for gas mileage?

What table are you editing? I don't even see a CL target table...

I'll see if I can get a screenshot from AccessTuner later. I just installed Windows 7 on the desktop and so a bunch of my software isn't on there yet. AccessTuner is not wanting to install on Windows 7 right now.

CL needs to be "stoichio" (14.7-1) for the cats to work properly.

These leaner mixtures increase NOx in the converter feedgas, while decreasing HC, CO, and CO2:

http://www.legacygt.com/forums/attachment.php?attachmentid=81654&stc=1&d=1268329060

Additional oxygen will improve the oxidation process in the cat while hurting the reduction process due to a lack of CO. As far as reliability goes, the only thing the car has is a single aftermarket cat with metallic substrate, part of the ERZ downpipe. As long as the cat doesn't overheat (the EGT probe is not installed currently) I'm not worried about damage.

http://www.legacygt.com/forums/attachment.php?attachmentid=81655&stc=1&d=1268329246

Besides, it's a cheap aftermarket cat. It's mostly on the car so that there are no fumes under low-load driving.

I've successfully leaned out the low load highway cruising areas around 3000rpm. Right now AFR is hovering around 15.5:1 there. I may try to get it to 16-16.5:1 if the car will still run smoothly. So far I've found that the engine does not seem to tolerate a leaner target under the low load, low rpm area. These areas normally are pretty sensitive to mixture, but the boxer is more sensitive than I would have expect. In fact, the factory targets are richer than I expected, but I can see now from firsthand experience that there is a reason for that. The idle seems to stumble a lot when I lean it out in those areas, so I just put it back to stock.

In the transition areas (between .60 and 1.0 load) I've leaned it out slightly under cruising rpms. I initially went too lean in this area and the engine would stumble some. 1.00 and above have remained unchanged.

http://www.legacygt.com/forums/attachment.php?attachmentid=81684&stc=1&d=1268348970

I haven't smoothed every cell out yet but that's the current calibration after some trial-and-error.

Have you actually tried running with those changes to see if they are taken into effect? I asked about this a few weeks ago, and the conclusion that was come to is the O2 sensor would need to be remapped as the ECU will always (outside of idle, cold start, or open loop) try to go to 14.7:1 regardless.

If this actually works it could be VERY useful for getting better fule economy.

uh yeah i'm running it right now. On the highway I'm between about 15.2:1 and 15.8:1. But adjusting these tables requires a reflash and I didn't feel like plugging in those damn diagnostic connectors. I still don't know how you guys put up with doing driveability tuning without realtime capability. That's alone is worth the $600 for the AP, especially considering that the equivalent capability on my Rx-7 costed me twice that much.

I can't believe nobody else has tried this... you know that a lot of regular port injected engines can run as lean as 17:1 ? It's just a matter of the NOx emissions as I've said. But a lot of it depends on engine speed. Low load, medium rpm (40-80mph cruising) is very tolerant of lean mixtures, and higher idle speeds can handle leaner mixtures. Lots of other engines can idle at 16:1 no problem if they happen to require a higher idle speed.

Even my Rx-7 can idle at about 14:1 with the idle set to 900-950rpm. And with the porting (aggressive cam equivalent) it only pulls 13 inches of vacuum at that idle speed, and closer to 10 inches at 750 rpm. At 750rpm that motor requires about a 12.5:1 idle mixture to be stable, depending on timing advance.

I was surprised at how crappy the EJ255 ran when I leaned it out at idle. But the different target AFR's may be somehow affecting the throttle control logic as a side effect. The dashpot logic to prevent stalling, which nobody has uncovered yet, may be based on commanded AFR. I also noticed slower cranking times when I experimented with leaning out the upper left corner of the map. Let me see if I can dig up my logs. I've got my laptop here.

Alright, I made some graphs. Given how big the Evo/Subaru market is, I'm very surprised that there are no real log viewers available. Using Excel is very slow and clunky compared to MegaLog Viewer (Megasquirt), FC-Chart (Power FC), Hondata's log viewer, etc. There's no map tracer or slider bar.

Anyway here is a sample of some highway cruising at 60mph (cruise control) if I remember correctly. These are all the same dataset. I hate putting more than one parameter on a graph. I always make separate charts of each parameter versus time, with the y axis scaled appropriately for whatever I am looking at. This is quicker and easier with the aforementioned log viewers that you can find on an AEM EMS and other systems.

http://www.legacygt.com/forums/attachment.php?attachmentid=81692&stc=1&d=1268367921

Have you actually tried running with those changes to see if they are taken into effect? I asked about this a few weeks ago, and the conclusion that was come to is the O2 sensor would need to be remapped as the ECU will always (outside of idle, cold start, or open loop) try to go to 14.7:1 regardless.

Don't believe everything you read on the internet. Are these the same people who call the front O2 sensor a narrowband? Because I'm pretty sure the motor hasn't blown yet.

Have you actually tried running with those changes to see if they are taken into effect? I asked about this a few weeks ago, and the conclusion that was come to is the O2 sensor would need to be remapped as the ECU will always (outside of idle, cold start, or open loop) try to go to 14.7:1 regardless.

 

If this actually works it could be VERY useful for getting better fule economy.

That was me who led you down the wrong path. These tables only become available in the last round of definition updates for RomRaider. I wasn't aware that they could be used to lean out closed loop. I thought they worked like the open loop tables, which can only be used to enrich beyond stoich.

However if the table I'm looking at in RR right now is the same as the one that Boostin is editing - and I think it is - then it's possible to do this with RR as well as AccessTuner.

I still don't know how you guys put up with doing driveability tuning without realtime capability. That's alone is worth the $600 for the AP, especially considering that the equivalent capability on my Rx-7 costed me twice that much.

My workflow goes like this: flash, drive and log, come home and play with Excel, create new tune, repeat. Real time tuning would only remove 30 seconds from a cycle that takes an hour or three. I'm too chicken to reflash without thinking about it for an hour first.

My workflow goes like this: flash, drive and log, come home and play with Excel, create new tune, repeat. Real time tuning would only remove 30 seconds from a cycle that takes an hour or three. I'm too chicken to reflash without thinking about it for an hour first.

Fair enough. That's more necessary when you tune solo, and I've used a similar process in some situations. When tuning WOT in realtime, the driver hits it in 2nd or 3rd, I look at the laptop. Then I change the tables and he hits it again. Then I change the tables one last time and we decide where to get a burger... all without even pulling over The map tracing is very helpful in the Cobb software.

Still, the Cobb software is really not very good for a system that popular. There is no map watcher. In my Rx-7 Power FC software I can blank out the fuel map and populate each cell of the fuel map with the logged AFR. I can blank out the timing map and populate it with knock values. I can do the same with idle speed control valve duty cycle, TPS voltage, etc. I can see a realtime curve of TPS voltage, injector pulsewidth, and rpm. This is very helpful for tip-in tuning. Hondata etc all have similar capability.

Don't believe everything you read on the internet. Are these the same people who call the front O2 sensor a narrowband? Because I'm pretty sure the motor hasn't blown yet.

Heh, I dont think so... I was asking more about the open source side of things and the information I was getting over here matched what I had dug up over on the romrader forums. I just looked into it a bit more and as NSFW stated, it was a recent definition thing. Not that any one seems to have tryed it before you though. The original post I was refering to start about half way through my EGT sensor replacment thread.

How's drivability getting up to speed with leaner AFR, and in particular when it transistions between richer and leaner portions of the map?

How's drivability getting up to speed with leaner AFR, and in particular when it transistions between richer and leaner portions of the map?

Initially it would stumble a little bit because the .60 - .80 columns were too lean, but at these settings posted above there are no noticeable differences in the way the car drives. The next step for me is to further lean out the 2800-3600 rpm, .20-.50 load matrix, as this seems most tolerant of lean mixtures.

you need to add timing when you go to lean cruise if you want to get the maximum fuel economy. I haven't finished my experiments but the initial data seems to indicate an optimal between 5 and 10 deg extra advance at highway cruise between 15.5 and 16 afr.

you can also use AVCS to dilute the mix to decrease pumping losses a bit. the experiments I did a few years ago (with stoich mixtures) indicate two fuel economy peaks. One at 0 AVCS with more ignition timing and another at ~15 deg AVCS with timing backed out a few degrees. This was all done at 100kph cruise.

My workflow goes like this: flash, drive and log, come home and play with Excel, create new tune, repeat. Real time tuning would only remove 30 seconds from a cycle that takes an hour or three. I'm too chicken to reflash without thinking about it for an hour first.

I use a different method to remove the effects of road variation from my logs.

Once a mixture is loaded using Romraider, I pull up the cobb RT controls and vary the ignition timing up and down a degree at a time every second or so while cruising. you get much better statistical certainty this way as a given road condition is tested over several different timing conditions (assuming the road conditions don't change as rapidly as the timing chages).

Subscribed. I wondered about the same thing when looking over the romraider editable maps a couple months ago when I first got my OBXT. I would think you'd want to also retrim your MAF table so the ecu isn't constantly taking out fuel to hit your AFR target. Not a big deal I guess once it trims itself, but if you ever reflash or reset the ecu, it will have to start over with the trims to get back to that AFR target.

Or I'm way off here, still a Newb with this subaru.

I run my 1.8 miata at 16.5-17.0:1 under cruise with a bunch of timing to lower EGT. I don't really know how much you save by going that lean however.

I would think you'd want to also retrim your MAF table so the ecu isn't constantly taking out fuel to hit your AFR target.

bit of a misconception here. you don't need to trim the MAF calibration if it was right to start with. once you move the CL target the car won't be undoing your changes by adding long term trim. that's the whole point of the CL targets. the car will be applying trims to achieve the desired AFR

bit of a misconception here. you don't need to trim the MAF calibration if it was right to start with. once you move the CL target the car won't be undoing your changes by adding long term trim. that's the whole point of the CL targets. the car will be applying trims to achieve the desired AFR

The car would be modifying the long term/learned trim, correct? The stuff you see using learning view? The zones where you modified the CL AFR target (I assume it'd be the left most g/s zone, 0-5.6g/s with my car, or the next one to the right) would be a large negative number as the ECU is trimming it's fuel map (based partly on MAF) to hit that leaner AFR target. Which I agree is fine and all to let the ecu do it's thing closed loop.

Or you can trim your MAF table appropriately so the ecu won't have to trim as much to get to your leaner AFR target. Right? Ultimately the same result would occur, the ecu using CL to hit your AFR target regardless, but scaling the MAF table as well seems cleaner and quicker and would result in less overall "A/F Learning" trim in your cruising zone.

Sorry if I'm dense and have this all wrong, I'm still learning.

sorry, I might have not been to clear in what I was trying to convey.

If the maf calibration is accurate to begin with, the long term trims will be small regardless of the fueling target called for (within reason). For example, if the mixture was set to lambda 1 and the maf was calibrated, when you ask for a new CL target, the car should hit that new value and keep the trims small.

What you don't want to be doing is trimming the maf table to try and achieve a steady state AFR that is different from the CL target. by all means, adjust the maf to bring down the LT trim value, just don't try and change the CL AFR using the maf. the change will be compensated out for with the trims until you run out of adaptive range, then you will get a CEL and the system will be essentially open loop.

I probably won't bother with adjusting timing and AVCS for this project. Although it may very well help, it would probably be time consuming and have a marginal benefit that would be difficult to quantify. As for the fuel trims, I didn't notice any difference in those initial tests. It was still about +4% under those conditions. I'll keep an eye on it though. A Perrin TMIC is going in soon so more tuning will be going down.

Even if the trims do change a little bit, it's not going to affect any higher load driving.

I probably won't bother with adjusting timing and AVCS for this project. Although it may very well help, it would probably be time consuming and have a marginal benefit that would be difficult to quantify.

You gotta do that on the dyno. I tried on the street and it was too hard. It only took about 30 mins on the dyno to get AVCS and timing dialed in for optimal fuel economy. The problem is my setup has change since I was last on the dyno and I havne't had time to go back and re-optimize. The best cruise economy (on E10) I managed was 6.8 to 7.0 L/100km at 105. At 80 I was getting around 6.25 L/100km.