CL AFR Oscillations

[LittleBlueGT]

In the most recent defs from RR, we can log CL AFR target. The ECU seems to have two modes, one is where it actually targets what is in the table, the other one is where it does an oscillation, probably for better emissions.

 

I assume this table can be found in the ECU, and manipulated. Hopefully someone will do this for us!

 

I imagine this oscillation combined with another small problem can cause the problem. (for some a bad O2, for me Os in the DP, etc......)

 

Here is a graph from a log from yesterday showing the target oscillating, and obviously causing the AFR to oscillate.

 

http://i248.photobucket.com/albums/gg194/littlebluegt/oscilatingAFR.gif

[John M]

Back almost 20 years to my first DSM, AFRs did exactly this in steady-state closed loop. I assume all cars do the same thing. What problem can it cause?

[Boostin1657615274]

This agrees with the literature I have, an emissions training manual from Toyota:

 

http://legacygt.com/forums/attachment.php?attachmentid=87832&stc=1&d=1279690834

 

http://legacygt.com/forums/attachment.php?attachmentid=87833&stc=1&d=1279690834

[LittleBlueGT]

I have actually read all that stuff, but I didn't want to re-iterate it.

 

At any rate, for those of us w/o cats, might be nice to have this table defined.

[Legend]

I have actually read all that stuff, but I didn't want to re-iterate it.

 

At any rate, for those of us w/o cats, might be nice to have this table defined.

 

What would you want it to do? Steady stoich? Reduce amplitude of oscillations? Get rid of the rich spikes?

[LittleBlueGT]

What would you want it to do? Steady stoich? Reduce amplitude of oscillations? Get rid of the rich spikes?

 

I just want it to target the CL values, no ECU induced oscillations for cat efficiency, or at least have access to that table to do whatever I want with it.

 

The problem I have is that those 14.2-14.9 oscillations are noticeable when driving steady state. Because my front O2 is in the DP, the AFR corrections start to swing as well. When the front O2 was in the header, it was not noticeable, now it is.

[bugblatterbeast]

The swing you are seeing is really small compared to what we had on Al's car. His swing was form ~12 to 18 with steady throttle.

 

Other than what I PMed you, there is another table that allows you to drop the swing. I wasn't sure if it would work but I just got back from a test drive and I'm pretty sure the tweak worked. I'll mail you separately later if the data looks good.

 

 

BTW, a tweaked version of my emissions tune passed today. IDLE: 111ppm HC 0.05%CO, 2500RPM: 39ppm HC 0.05%CO

[Legend]

BTW, a tweaked version of my emissions tune passed today. IDLE: 111ppm HC 0.05%CO, 2500RPM: 39ppm HC 0.05%CO

 

Curious about your emission tune, BBB. I've worked on one too and haven't tested it yet, although I'm pretty sure I'll have to in August when my renewal card comes in the mail. I've seen a lot of cars fail here and the primary reason for failure is high NOx. Sorry to sidetrack, LBGT.

[bugblatterbeast]

are you guys put through a dyno test or just a two speed idle? NOx shouldn't be a problem at idle as there really isn't enough loading to generate much NOx.

 

I can see people failing if they lean out the mixture then advance the timing to make up the lost torque. what you need to do is lean out the mixture to pass CO and HC, then retard the timing so NOx doesn't get too bad. your HC emissions will also improve with a later ignition point as the quenched area decreases if the piston is closer to TDC when combustion happens. just don't lean out the mixture to the point that you get misfires or HC will soar.

 

don't worry too much if the car doesn't idle well at first. it takes a some time at idle for the car to figure out the correct throttle position, this is especially true when you retard the idle timing. also, watch out for the fuel enrichment that happens when the coolant temp gets above 200f. if your emissions test takes a while, that can cause you to fail HC.

.

[Legend]

^^ It's a 4-minute dyno test. If you fail you get a printout to take to a shop to help you pass. Here is a little work that I've done and a failure chart in this thread.

[bugblatterbeast]

from the looks of it, you just need to retard the timing a little and/or adjust the avcs to leave some exhaust gas in the cylinders at the end of each cycle.

[ClimberDHexMods]

I just want it to target the CL values, no ECU induced oscillations for cat efficiency, or at least have access to that table to do whatever I want with it.

 

The problem I have is that those 14.2-14.9 oscillations are noticeable when driving steady state. Because my front O2 is in the DP, the AFR corrections start to swing as well. When the front O2 was in the header, it was not noticeable, now it is.

 

Why do you suppose it became noticeable now that O2 is in the DP?

[LittleBlueGT]

Why do you suppose it became noticeable now that O2 is in the DP?

 

Slower reaction time, and the AFR gains or (PID controls) are set-up for manifold placement of the front O2. I imagine the JDM ROMs (TS) have different gains. But, unfortunately, those tables are not yet defined.

[Guest]

The swing you are seeing is really small compared to what we had on Al's car. His swing was form ~12 to 18 with steady throttle.

Other than what I PMed you, there is another table that allows you to drop the swing. I wasn't sure if it would work but I just got back from a test drive and I'm pretty sure the tweak worked. I'll mail you separately later if the data looks good.

 

 

BTW, a tweaked version of my emissions tune passed today. IDLE: 111ppm HC 0.05%CO, 2500RPM: 39ppm HC 0.05%CO

 

Just found this thread, very interesting.

 

BBB or LBGT, can you share your findings for the community sake?

[NSFW]

I'd like to see that table as well. I have spent some time hunting for the logic/tables that govern this behavior but I came up empty-handed.

 

I'd really like to suppress it a low loads and/or until the engine warms up, as I suspect it's causing my car to surge a bit when it's cold.

[Guest]

http://www.romraider.com/forum/viewtopic.php?f=15&t=7725

[bugblatterbeast]

it has been more than a year since I looked at this problem so my memory isn't the best. from what I can remember there are two things to look out for.

 

the first is to make sure the injector scaling is correct. if the scale factor is too low (and a correction made to the maf table to get the OL afrs right), you end up with a situation where there is too much loop gain in the feedback system. the magnitude of the CL oscillations will get quite large as the ecu constantly overshoots the targets by adding or subtracting too much pulse width. with this in mind, it is possible to push the scaling a little bit on the high side (with a matched correction in the maf tables) to somewhat decrease the AFR variations. just don't go too far or the CL corrections end up being very sluggish.

 

the next thing that can be done is an adjustment to the CL targets and warm up values. at cruise I lean burn. you can force the AFR target about half a point lean of stoich before the ecu gets pissed off. while lean burning, the variation in afr decreases. I don't know why, but the car acts as if it is OL even though it is reporting CL status. if I had to guess, I'd say the variation drops because the rear O2 sensor ends up getting pegged lean and it's contribution to the overal control is essentially turned off.

 

for warm up stumbles and surges, just increase the timers that hold the car in warm up enrichment + bump up the base warm up enrichment.

 

again, I did these experiments more than a year ago so my memory is a bit fuzzy

[NSFW]

If you can log AF Correction #3, you can watch it drive the oscillation. I've tried to find the code/tables that govern AF Correction #3, but had no luck. Generally once you find out where a value is stored in memory, IDA will list all of the place in the code that use that memory location, but for some reason that list is empty for the AF Correction #3 memory location. Probably the section of code that governs it just hasn't been disassembled yet, but finding that code is like searching for a needle in a haystack.

 

However, I'll look into how the CL target comp table works, maybe that will lead me to it.