I tried to make the retune as absolutely simple as possible so that I could do the same tune on 94171715 in the future...

I took a fresh copy from the Tephra V7 thread of 96531706.
I loaded in my injector settings and startup settings
Applied jitter fix "2"
Put the ECU into open loop through the periphery bits

I tried my previous SD table settings, but it was VERY lean with the change in baro.

Here are the settings I ended up with for the SD tables
"Stock" fuel map


Runs well above 2500 (same as it always has), which considering I went from the MAP VE table set 1:1 and the RPM VE set to 100 across the board to that in one revision, I think this is fairly easy to setup when the data is consistent. Now, with the car sitting idling, it gives me some interesting results.

First, let me just say I'm using 14.7*128/x as the evaluation equation in evoscan for the mapped AFR. MUT 32 (FFFF8AC3)

AFR Bounces around from 14:1 to 17:1.
Mapped AFR bounces from 13.8:1 to 17.1:1???? (small blue square wave at the bottom)
WTF?
IPW is the green square wave at the bottom with the mapped AFR...
Total sync FPW is at the top




Why is the mapped AFR value jumping around? The load is sitting around 30 with RPM around 950. Should be idling right around 14.7:1. Well, even if it jumped down to the next RPM, it would still not be at anything even close to 17:1.

 

Also, here is the results I get plotting the data. This is with fuel cut, decel, accel conditions removed.

If it is colored yellow, there is not enough samples, but I would still consider those values in building up the VE tables. They would just have a lighter weighting.

Green data is good data where there was a lot of samples AND a low standard deviation.

RED data is data with reasonable samples, but terrible standard deviation. You can see, what ever is happening in the ECU makes that area absolute crap as far as AFR consistency. Also, the AFR error jumps from being too rich to being VERY lean.



You can see, the green data provides good trending data. Even the yellow is pretty consistent and meaningful. The red on the other hand almost looks like it's a completely different tune.

 

There are actually three different algorithms that the ECU can use to determine AFRMAP. The two that most people know are 1) the fixed value of 14.7:1 for closed loop feedback and 2) the value looked up from the fuel map during normal loop conditions. The third algorithm uses the front or rear (can't remember of the top of my head) O2 sensor signal to generate the AFRMAP value. You must have somehow triggered the use of this algorithm. When I previously asked to see your AFRMAP value with your lean AFR issue, it was because I wanted to see if your ECU was using this algorithm.

 

Yeah, previously the AFR MAP wasn't doing that, as shown in the first data log I posted. But I have seen the AFR MAP value go to 16.7:1 before and thought I just messed something up in EVOScan or was getting bad data for some reason.

Pretty interesting that it's using O2 feedback though as FAA bit.4 is set to 0 so it should be full time open loop?

 

Even though it is feedback based, this algorithm falls out side the conditions that invoke the usual closed loop.

 

Do you have the location of the sub in 96530006? I must not have something related disassembled correctly because the AFRMAP Mut ram address only pulls up one sub routine (1433C).

 

I don't have my Evo laptop open, but all the AFRMAP variants are calced in the same subroutine.

 

Out of curiosity, would shutting off the rear o2 simulator and putting the rear O2 back in provide any kind of valuable test data if this is based on the rear O2?

 

Bringing this back up.

Any idea why this other correction seems to be kicking in?

My car has been driving like *** for months below 2000 RPM and I've just been trying to ignore it. I've retuned with Baro set to 86kpa and while it forced a complete retune it didn't change anything with regards to this issue.

It did make the car suffer the normal 96530706 issues that most have reported though (high idle on start up and throttle hang most notably) which I never had before locking baro to my atmospheric pressure. It did straighten out my car trying to cruise at 5 psi in closed loop though without changing all the open loop control tables like I was doing previously.

 

What is the front O2 sensor signal doing during these events?

 

03whitegsr, did you find any more info on this?

My car is doing something similar at idle, tho not as extreme.



Using 95531706 rom. Car is set to Open Loop. Idle is steady. Car is fully warm (drove around for a while). Load is steady. kPa is steady. IPW doesnt bounce around. Only the afr bounces from 14.0 to 14.9. LTFT lo is -2%

 

I haven't had a chance to look at it much because of school. I've just been driving around it.

It actually seems like it may be something related to the MAF filter switching that was causing the jitter noted by L2r99GS-t. When it's cold, it happens all the way up to 2500 RPM. It's pretty apparent fuel is more or less cutting in half. Once it's warm, the AFR still bounces around, but it doesn't missfire when it goes lean. I believe I'm using "fix #2" at this time.

Overall, car gets 24mpg and drives great once it's warmed up so I've just adapted.

 

Bringing this back up.

I believe this is related to an O2 feedback issue. I have disabled FAA bit.1 (rear O2), bit.4 (closed loop), and bit.14 (front O2) which are all related to closed loop control and I have removed the front and rear O2 sensors.

Having only bit.4 deactiveated does not eliminate all closed loop trims, I believe Aaron noted this with some of his tuning and the code looks to match his observations.

Doing this made the car run lean at low RPM (under 2000 RPM), however it doesn't have the "switch" like issue where it goes extremely lean in only certain circumstances, it is just always lean however it now responds very well to the VE tables. I retuned and everything transitions smoothly on AFRs now below 2000 RPM.

However, I am noticing some inconsistent AFRs. Particularly at idle and I believe this may be purge related. I believe disabling FAA bit.12 may be the final control bit to disable and may simplify a lot of the fuel calcs and make the ECU run much more simple and consistent, comparable to a standalone. Bit.12 appears to be Purge related, however disabling it causes the car to be virtually undrivable. I think however it may be tied to Bit.5 and they may need to work in compliment manner (if one is active, the other shouldn't be active).

My goal here is to eliminate a lot of the ECU function that isn't really needed on a more race oriented car just to eliminate any unknown fuel and timing trims. In the end, I'd like to have very simple front O2 feedback and then none of the emissions trims active at all. I think eliminating a lot of the unknown feedback trims is step 1.

 

After installing my FIC 1450cc blue max injectors my car is doing something similar. At cruise conditions it surges pretty badly. AFR swings from 17 to 11 over and over again. I goose the gas and it will settle for a little bit, then surge again.

At idle it also does this albeit not quite as extreme.

Fuel trims are within spec.

This condition did not happen until I switched injectors.

 

Any Updates ? ? ?