AFR targets at loads higher than 2.50 ATR/ATP fix
 

As the temperatures got colder and people were pushing more and more air with their BT setups we started to notice that the cars were no longer following the target AFRs when loads exceeded 2.50. To properly diagnose this as the problem that is happening to your car, you should see a "step" change in AFRs AND injector duty cycle just as the car hits 2.50 load. See the log attached and highlighted for this and look at how the AFRs jump just as the load gets to 2.50. Make sure the car is leak free on the exhaust and intake and this is the actual reason for incorrect AFR readings

A few of the tuners on here made the same observation so we did some investigation and found that some of the fueling tables in ATP/ATR are actually mislabelled by COBB.

Generally when the AFR targets for normal operation are set, they are set in the No Knock, Base, and Unused (just to be safe) tables.

The knocking tables are generally left untouched at OEM values since they don't come into play until the car knocks - or so everyone thought.

This is actually not the case. When the car knocks the AFRs go 0.5 AFR richer from whatever table and target the car is following at that point.

So what do the knocking tables do??

They are actually tables that are invoked when the car goes above a certain load value. On most of the cars we have seen that load value is 2.50. This value is actually stored in a table that COBB hasn't revealed (yet).

The fix is actually quite simple. Set all the "knocking" tables to match the no knock tables. The "knocking" part of the table name is actually not true, these don't have anything to do with knocking.

The tables are all pretty much the same in all gens and the MS6, I have highlighted these below for a Gen1 Speed3.

http://www.mazdaspeedforum.org/forum/foru...1&d=1389660690

EDIT:

This problem seems to potentially have different solutions in different cars. If the above does not fix it, the following should work:

1. When the load goes above 2.5 and the AFRs come off target, calculate a new scalar for the MAF curve. To do this, divide the actual AFRs in the logs by the Requested AFRs in the fuel tables.
2. Multiply the MAF table by this scalar and multiply the load tables by the same amount.

If this still doesn't solve the issue, then there are two more areas to look at:

1. Instead of using an injector scalar for E85, just scale the MAF and set the injector scalar to 1.00
2. If you are using a phasing multiplier, set this to 1.00

Now I just need to get fast enough to hit 2.5+ loads.

Thanks Lex, I'm glad I wasn't going crazy in seeing afr's go lean above that specific load point.

Thanks @Lex; I've been having similar issues, I'll give this a shot.

Good job guys finding a fix and ACTUALLY SHARING it with everyone. That's what this place is all about.

Proud to be Lex tuned.

@Lex; once the load goes above 2.5 and transitions into the "knock" table, are we to only likely to see richer AFR's? Will it then remain within that table until the end of the run? Or does it transition back to no knock table at some point.

Reason for asking is because I do notice the richer AFR's to a point but also have seen them go the complete other way and settling towards redline. (Only occurs specific to the 2.5 load) Maybe unrelated

Once it drops below 2.50 it goes back to the normal tables so you might see some oscillation if you're around the 2.50 mark.

Also with the MS6 specifically the VVT VE table has a large effect on fueling as soon as you change the VVT advance table from stock.

thanks for the info Lex! I haven't looked into anyone going over that load yet but its good to know for the future.

Haha yeah I found that out the scary way, (trial "with" error). Thanks again for the info Lex.

are people at cobb still working on our platform?

lol good question

I have seen this on 4-5 gen 2 cars but never on a gen 1. I always set my knock/no knock tables identically along with the rest of my fueling tables and i still see the same thing happening. If you need some evidence/logs i can send over whatever.

There might be something else going on there. Are the AFRs going lean or rich for you? Also is this only on Gen2 cars?

Wow... so we dont have the option of setting a full table of knocking values (as we thought we did)? If the commanded fuel (knocking) table is only used by the ecu above load 2.5, 90% of this table is useless. Im assuming this is the case for OL/Part throttle as well?

Awesome find, thanks!

Have you guys tried scaling the load axis? One thing of note: It seems when scaling the load axis, the ignition AND fuel tables are affected.

Does anyone have a log of this we could look @?

+1 on log, would be very interesting to see. @Lex; did you find this/verify this with live tracing?

Cars I've seen it on actually go lean.

@rfinkle2; I will dig up a log shortly

I'm glad to see some progress being made on this issue. I see a little over 2.50 load pretty often and my AFRs go lean when it happens. It'll be nice when this is ironed out. Thanks again Lex.

Here are a couple logs to show the lean spot. I would like to mention too, that this never used happen and I have tuned quite a few cars that go over 2.5 load, so something has to have happened in an update along the way

@atvfreek; I'd be interested to know if making the knock/no-knock tables match fixes the lean issue for you. I haven't seen what you are seeing, only the rich condition. I wonder if that discrepancy is based on the different model/year versions of ATR.

Thanks for sharing :) What do you guys suggest moving forward? If we edit our (knocking) fuel table to match (no-knock) fuel tables, and cobb corrects this... we are then essentially tuning out a fail safe. From reading @lex's post, it sounds to me like the (no-knock) table is not used until a calc load of 2.5 is met. So are we stuck with the ecu correcting 0.5AFR if there is knock? Has this issue been seen with the Ignition Tables (@ 2.5)? ATR tables go to a scalable calc load of 2.0, is our ecu being limited therefor searching for a cell to select when hitting a calc load of 2.5? This theory would be in line with @rfinkle2; suggestion of scaling load axis.

Based on timing I assume that car is on E85? If it is and you are using the fuel scalars this will cause the car to go lean in the midrange. Use MAF scaling instead to avoid the car going lean in the midrange. I highlighted this region below

http://www.mazdaspeedforum.org/forum/foru...1&d=1389727622

For the top end ripple, I suspect it is a hardware issue. The car is below 2.50 load at this point and the gen2s use feedback from the WB02. So check for exhaust leaks and I would also look at the intake for leaks, MAF sensor, etc. Some intakes can also cause this kind of effect in the top end.

http://www.mazdaspeedforum.org/forum/foru...1&d=1389727622

It is isolated to the load tables and not the scalars. I have a ton of logs from a couple different cars that it is as soon as it goes over 2.5 load, whether on 93 or e85. One of the cars does it if it just goes over the TRL-x gear table at all (not specifically 2.5 load, as his is 2.3 load)

Here is a link to a guy having a similar issue. When he raised his load tables out of reach (ots) his issue went away
http://www.mazdaspeedforum.org/forum/foru...7/#post2408721

If all of the fueling tables are set to the same target, the ecu will inject fuel when it detects knock without regard to the knocking table.

(assuming the tables are still working as they had in the past).

I haven't seen this issue and I fully load tune so load tables/targets are always in play. I've attached a log for reference. The car follows target AFRs aside from going a tad lean after spool which is fairly normal due to the ECU ending spool enrichment and then starting to lean out up top when his IDCs get above 110.

I have seen the issue when using the injector scalars on the same car and others above certain loads.

Right, this is where im lost. If this issue in ATR exists and we cannot tune for knock properly.. we're kind of screwed. If the ecu is bumping the AFR down by 0.5, that's more then sufficient to counter-act knock, but it is tuning around a larger issue. I realize this is a fix and not an "end all be all", just wanted to clarify before running my (knocking) tables the same as (no-knock) tables :sad: that there is in deed "something" to deal with knock.

Without completely derailing the thread, I have long set all of the fueling tables the same and allowed the ecu to pull timing and inject fuel when KR occurs. Some people use this strategy, other set the knocking table .5 point or so lower.

I wonder if this is what happened to Jeff's car?

@mzr0818;

The ECU will add fuel when KR occurs regardless of what you set the knocking fuel tables to.

If you don't see above 2.50 load and your car follows the AFRs you are commanding, don't need to change anything.

Thanks, this is what I was looking for :)
And thanks for sharing this discovery!

If mine was going lean at load I should see signs sunday when she comes apart. I had switched back over to 93 map for the winter about 3 weeks prior to blowing out. 93 would have offered less resistance to KR if it was going lean. I wish I was logging at the time but as I said I was in cruise control for 27 miles before it happened

Interesting... I have been having issues with this as well.

mine was always rich and only gen 2, ill attach logs in the morning.

I am going to get some more logs later today, but was curious if you or anyone has tried this fix to the fuel tables yet with logs to show it working. I don't always hit above 2.5 so haven't been able to verify success. I have been having the same "lean" issue on both 93 and e85 maps as well. Here's my thread started back in August when I first noticed it, along with the recent occurrence. http://www.mazdaspeedforum.org/forum/foru...an-wot-153472/

Your log was exactly 2.5 load as well. lol. I haven't seen them go that lean before, but same scenario for sure.

I just sent out a map with the boost toggle unchecked and want to see if that makes any difference.

What are your trl-x tables set to? Are they at 2.5 or above 2.5?

How heavily have you tried to correct that portion of the fuel curve freek?

Not heavily because if the log stays at 2.49 the afr is dead on perfect. The second it hits 2.5 on this specific car is goes from 11.8 to 12.2.

Now when it hits the same g/s yet at a higher rpm, that will give less calc load and the afr will still be spot on :bs2:

everything is set above 2.5. (3.0)

Has anyone talked to Cobb about who is providing atr support for the mazdaspeeds now?

@Evan@COBB; @Cobb Tuning;

it sucks that this stuff is mislabeled =/ at least we have smart tuners in the platform that not only figure it out by trial and error, but then share it with the community. I could see some tuners on other platforms figuring this out and them keeping it to them selves.

I always always always had issues until I set my all my load limits to 5

It was just a weird thing where even when I was under the limits of 3.0, I still saw issues. I feel like I had other issues that showed up too, other than afr, but it's been so damn long I can't remember for sure.