Experimental RomRaider Definitions

[need bb]

I thought it would be great if we could share the experimental RomRaider definitions. These small "tweaks" can make a big difference in the drivability of the EJ255.

 

Current I am using the fuel consumption fix for those with larger fuel injectors:

 

http://www.romraider.com/forum/viewtopic.php?f=34&t=9530&sid=0d0dded6950fa9f872a42fb186613435

[utc_pyro]

Thank you for posting this, I was just about to order the parts to build an intercept the signal from the cluster to the MPG gauge, but this fixes it!

[need bb]

Glad to hear.

 

 

Does anyone have advice on the disabling AF3 correction, the deceleration latency, or the fuel pump duty cycle?

[Th3Franz]

Set AF3 correction/learning to 0. Fuel pump duty cycle should be set to at least 60% for a larger fuel pump. I like to run 60/80/100

[covertrussian]

Set AF3 correction/learning to 0. Fuel pump duty cycle should be set to at least 60% for a larger fuel pump. I like to run 60/80/100

Not that easy with Romraider, you have to find rom version and then specific locations to disable it.

 

Does anyone have advice on the disabling AF3 correction, the deceleration latency, or the fuel pump duty cycle?

 

What's your Rom version? Either way you should run A2WC522N so I'll give you the numbers for that. Update ECU Flash definition for A2WC522N and then zero out the below values.

 

AF3 Correction location: 2DF2C, 2DF30 (Hex value: 3F000000BF000000, Dec value: 0.5,-0.5)

AF3 Compensation Location: 2C91C, 2C920 (Hex value BF0000003F000000, Dec value: -0.5,0.5)

[need bb]

Great, I will give it a try.

 

What about front A/F sensor smoothing and decel fueling latency?

[need bb]

Not that easy with Romraider, you have to find rom version and then specific locations to disable it.

 

 

 

What's your Rom version? Either way you should run A2WC522N so I'll give you the numbers for that. Update ECU Flash definition for A2WC522N and then zero out the below values.

 

AF3 Correction location: 2DF2C, 2DF30 (Hex value: 3F000000BF000000, Dec value: 0.5,-0.5)

AF3 Compensation Location: 2C91C, 2C920 (Hex value BF0000003F000000, Dec value: -0.5,0.5)

 

I run A2WC522N. I tried loading the AF3 mod in ECUFLASH, but I wasn't doing something right as I couldn't get see any table. I put some of the value in the 32BIT Base and the rest in the A2WC521 file.

[covertrussian]

I just copied and pasted from these posts (modified them as needed for my post).

http://www.romraider.com/forum/viewtopic.php?f=34&t=8595

http://www.romraider.com/forum/viewtopic.php?f=25&t=8131

[need bb]

I just copied and pasted from these posts (modified them as needed for my post).

http://www.romraider.com/forum/viewtopic.php?f=34&t=8595

http://www.romraider.com/forum/viewtopic.php?f=25&t=8131

 

I am a little confused. I see only one address value, but your earlier post mentioned we need to use two values. Please see below:

 

<table name="AF 3 Correction Limits" address="2fae0" type="2D" level="1" scaling="A/FLearning#1Limits(%)">

<table name="X" type="Static X Axis" elements="2">

<data>High</data>

<data>Low</data>

</table>

</table>

[covertrussian]

There are two tables that need to be zeroed out, AF 3 Correction Limits and AF3 Compensation limits.

 

For A2WC522N use these values:

AF3 Correction Limits location: 2DF2C

AF3 Compensation Limits location: 2C91C

 

<table name="AF 3 Correction Limits" address="2DF2C" type="2D" level="1" scaling="A/FLearning#1Limits(%)">
<table name="X" type="Static X Axis" elements="2">
<data>High</data>
<data>Low</data>
</table>
</table> 

<table name="AF 3 Compensation Limits" address="2C91C" type="2D" level="1" scaling="A/FLearning#1Limits(%)">
<table name="X" type="Static X Axis" elements="2">
<data>High</data>
<data>Low</data>
</table>
</table> 

[RallyMachine]

What exactly does zeroing af correction 3 and these other mods do? Sounds intriguing...

[covertrussian]

What exactly does zeroing af correction 3 and these other mods do? Sounds intriguing...

 

32 bit Subaru's, use the wideband front sensor, but they also use a dummy narrowband rear O2 sensor to adjust the fuel trims too.

 

Your rear O2 sensor is a dummy narrowband sensor, as in it only sees 14.7 AFR, if your below that your rich, if your above that your lean. Lots of cars use this kind of O2 sensors as the main front sensors, so it's not a bad design. O2 sensors have a filament in them that gives off a certain voltage based on the oxygen presence in the exhaust.

 

The cats burn the extra fuel/air but they should keep the AFR's fairly consistent, aka not swinging rich/lean. When you remove the cats, your rear O2 sensor starts swinging rich/lean, which can throw a CEL because ECU things the cat is not working properly. But it still throws valid/reasonable values thus AF3 correct is not a must.

 

When you remove the rear O2 sensor altogether and disable the CEL's, the ECU sees a voltage of 0, which it uses to look up against the O2 sensor scaling table which gives it a certain value. This value is then used in a multiplication equation against the fuel trims. The reason I said when you disable the CEL's, because with CEL's off ECU thinks everything is OK, instead of freaking out that the sensor is missing, and it goes about it's business like everything is in working order.

 

Now you have two tables (compensation limits and correction limits), which limit how much the rear O2 sensor can contribute to the overall fuel mixture, it's limited to 50% swing richer or leaner (in one of the posts it's mentioned as +/- .5 Lambda (+/- 7.35 AFR). That means the rear o2 sensor can make your 14.7 AFR be 7.35 AFR while cruising. When you zero out the two tables, you tell the ECU to swing 0% rich/lean thus it stops controlling the fuel trims and your gas mileage goes back up again.

 

TL;DR Removing Rear O2 sensor causes your car to be rich while cruising causing poor gas mileage.

[RallyMachine]

Very cool! I've got a couple questions... I don't have this currently set up for my tune but I have the rear o2 sensor removed and I'm not running overly rich according to my wideband. Do I still need it? Is this what causes the afr fluctuations at cruise which are supposedly to "heat the catalytic convetrr".

 

What about the decel latencys and fuel pump duty cycle, what do they do? Sorry it seems like I'm late I the party but just wanted to know a little more.

[covertrussian]

Well the ECU doesn't always count on rear o2 feedback, that's probably why your AFR's are still fairly decent. I would still zero it out though, it should help with the random richness, aka improve gas mileage.

 

Haven't looked into decel latency's or fuel pump duty cycles, still fairly new to Subaru ECU's & logic.

[RallyMachine]

Bump for any more info on the decel latencies and fuel pump duty cycle.

 

Also, just to confirm, I should copy that AF #3 code into my Program Files>ECUFlash>rommetadata>subaru>Legacy GT>A2WC522N.XML? Just in between the [rom] and [/rom] prompts?

[RallyMachine]

Bump for help on editing the definitions^^

[covertrussian]

Pretty much, I would backup the file before monkeying with it though.

[RallyMachine]

This is what I put into my A2WC522N.xml file:

 

Please let me know if any thing, especially the addresses look wrong. (This included the calculated MPG fix as well). I am able to view and edit these tables in ECUflash now, I am just worried that I will brick my ECU or mess something up with the coding. Also, as someone else said previously, are there 4 addresses I need to zero, or just these 2?

 

<rom>

<romid>

<xmlid>A2WC522N</xmlid>

<internalidaddress>2000</internalidaddress>

<internalidstring>A2WC522N</internalidstring>

<ecuid>2F12785606</ecuid>

<year>05</year>

<market>USDM</market>

<make>Subaru</make>

<model>Legacy</model>

<submodel>GT</submodel>

<transmission>MT</transmission>

<memmodel>SH7058</memmodel>

<flashmethod>sti05</flashmethod>

<checksummodule>subarudbw</checksummodule>

</romid>

<table name="AF 3 Correction Limits" address="2DF2C" type="2D" level="1" scaling="A/FLearning#1Limits(%)">

<table name="X" type="Static X Axis" elements="2">

<data>High</data>

<data>Low</data>

</table>

</table>

<table name="AF 3 Compensation Limits" address="2C91C" type="2D" level="1" scaling="A/FLearning#1Limits(%)">

<table name="X" type="Static X Axis" elements="2">

<data>High</data>

<data>Low</data>

</table>

</table>

<table name="Info Panel Fuel Consumption Correction" category="Fuel" address="C1428" type="2D" level="1" scaling="rawecuvalue">

<description>This is the correction applied to the fuel efficiency calculation to account for injector size change. Adjust this value by the same percentage as used for injector scaling. NewCorrection = CurrentCorrection * (NewInjectorScale / OldInjectorScale)</description>

<table name="Fuel Efficiency Correction" type="Static Y Axis" elements="1">

<data>Value</data>

</table>

</table>

<include>A2WC521N</include>

</rom>

[covertrussian]

Here is what I'm using, hope this helps. I pasted it right below 32BITBASE so that it would be ontop.

 

<include>32BITBASE</include>

<scaling name="AF3 Lambda" units="Lambda Additive" toexpr="x" frexpr="x" format="%.4f"  min="0" max="255" inc="1" storagetype="float" endian="big"/>
<table name="AF 3 Correction Limits - 2DF2C" category="Experimental"  address="2DF2C" type="2D" level="1" scaling="AF3 Lambda">
<table name="X" type="Static X Axis" elements="2">
<data>High</data>
<data>Low</data>
</table>
</table>
<table name="AF 3 Learning Limits - 2C91C" category="Experimental" address="2C91C" type="2D"  level="1" scaling="AF3 Lambda">
<table name="X" type="Static X Axis" elements="2">
<data>Low</data>
<data>High</data>
</table>
</table>

[RallyMachine]

Ok thank you... So your af learning 3 is now zero? Any issues with your af1 corrections like you mentioned in your RR thread?

 

Also any reason behind using the af3 lambda scaling vs the af1 learning scaling? Not sure that this value means?

[covertrussian]

Ok thank you... So your af learning 3 is now zero? Any issues with your af1 corrections like you mentioned in your RR thread?

 

Also any reason behind using the af3 lambda scaling vs the af1 learning scaling? Not sure that this value means?

 

Yeah my AF3 stays at 0 constantly, even though I'm still running stock downpipe. No issues with AF1 at all, I actually think my gas mileage slightly improved after I disabled rear O2 interference haha.

 

The reason the rear O2 is narrow band sensor is simply due to cost, Wideband sensors are much more precise and more expensive. Most cars don't even use a wideband front O2 sensor due to cost. Though usage of widebands is becoming very common in factory cars due to better emission/gas mileage controls.

[RallyMachine]

Ok cool. Yeah I know about the rear narrowband, I was just wonder about the scaling you used in the coding. One RR thread has the af#1learning limits as the scaling but the other like you just posted uses af#3 lambda. Is that just different scaling parameters because of wideband vs narrowband? Would the lambda scaling be more accurate?

 

I'm guessing it doesn't really matter since they're going to be zeroed anyway, just curious before I make the jump and flash everything. Looking forward to the results!

[covertrussian]

Oh your talking about the scaling definition, from what I can tell it just defines on how it will display the values. I've used AF1 scaling and it worked fine too.

 

It really doesn't matter, like you said because it will be zeroed out anyway.

[Th3Franz]

Look for NSFW's 522N definition thread on the romraider site. He has the definition for the fuel pump duty table and you can also pull the overrun tables for 522N from that same sub-forum.

[RallyMachine]

Look for NSFW's 522N definition thread on the romraider site. He has the definition for the fuel pump duty table and you can also pull the overrun tables for 522N from that same sub-forum.

 

Ok thanks I will check those out. What are the overrun tables for?