Enginuity strategy for shift-knock

[Legend]

Enginuity.org is still down so I'll post here. I DL'd the 774 Enginuity, new defs, etc. and loaded my latest map. I see some tables under Knock that could be promising for the "shift-knock" phenomenon (about half of my spirited shifts into 2nd, 3rd, or 4th generate -2 to -8 FC (about 400ms after a shift and resuming full throttle)) There are now a lot more options for dealing with knock! In particular,

 

- Feedback Correction Retard Value - We can decrease the knock value from -2

- Feedback Correction Negative Advance Value - We can increase negative correction in larger increments too...

- Feedback Correction Negative Advance Delay - and we can increase negative correction at a quicker rate too!

 

Safety is concern #1 so my gut tells me to keep the retard value at -2.

 

I think targeting negative correction might be the way to go. I am thinking of increasing the advance value to 1.5 degree (vs. 1.0) and decreasing the delay to 100ms (vs 250). Since these shift-knock events are VERY transient (and may not even be knock anyhow), these changes seem reasonable. If I made these changes the shift-knock phenomenon would still exist, but the negative correction would get cleared up MUCH more rapidly. This would allow my runs through the gears to have something much closer to the optimum timing I have tuned for in single-gear runs.

 

Thoughts?

[Legend]

Alright I made some changes and did a few run through the gears. On this map I changed:

 

- retard to 1.8 (vs 2.0)

- value to 1.5 (vs 1.0)

- delay to 100ms (vs 250)

 

Here are 3 logs. My thoughts?

 

1) Damn I sure get big knock after a shift. Drives me nuts. If I shift slowly this doesn't happen, however.

 

2) Retard at 1.8 evident on logs... really don't want to go lower and might just set it back to 2 and more aggressively target the other parameters.

 

3) Value at 1.5 also evident on logs. I really like the idea of increasing negative feedback in bigger chunks. I think it might be reasonable to really ramp up this value... to a value equal or even greater than the retard. I am thinking 2.0-4.0 might be acceptable. After all, if there's no knock let's get the dang timing back to where it should be ASAP.

 

4) Delay at 100ms is not evident to me on these logs. I had expected the value jumps to happen much quicker but they don't look any quicker at all. Maybe I don't understand this parameter or maybe it's just not working?

 

FYI, 12.1 atm (Denver), IAT ~60-70 degrees, power mods are Titek UP, Perrin catted DP, SPT cbe, enginuity tune.

 

To view logs, dl and save, then rename .pdf to .csv

gears 1st-3rd run1.pdf

gears 1st-3rd run2.pdf

gears 1st-3rd run3.pdf

[Gator GT]

shift-knock? please explain.

[mickeyd2005]

If you look at run1, there is no knock in 2nd gear, then he shifts and gets knock. Everytime he shifts, he gets more knock but only immediately after the shift.

 

Dynamix said he believes it is heat soak as the TMIC is getting hotter after each runup. This could be true. However, if you look at run1, at time interval 9323, the knock clears. If this was heat soak, he should have had knock here as well.

 

I think it's leaning out due to the car dropping into CL between shifts. Then the large delta MAF / delta time causes a lean spot due to the inability of the ecu to compensate.

 

Also, the range 4000-4200 rpm is very susceptible to knock so any lean spots here would be bad.

[Legend]

My tune is very pretty for single-gear runs 2k-redline in any gear. With regular routine driving I very rarely knock (well, except for about 50% of the time I am starting from stop in first I get -2 knock around 1400-1800rpm).

 

Anyhow, if I floor it in first, shift around 5500-6000rpm into 2nd and resume WOT, I get knock about 2/3 the time about a half second after resuming WOT. This pulls timing, anywhere from 2 degrees to 10 degrees. The same thing happens going into 3rd, but less frequent. It also happens going into 4th.

 

It means timing is pulled several degrees at the very times when I want the most power. Not good.

 

I don't know if it is really knock or not, first of all, but either way it is very transient and pulls timing that persists as I sprint through the gears. Bottom line is itt slows me down. Since I haven't been able to get rid of the knock phenomenon, I am enabling the ECU to restore timing quicker (to get rid of feedback negative correction quicker). It means that even if I get knock and subsequent timing pull, the duration and/or severeity of the timing pull will be less, putting less of a damper on my spirited runs.

[LittleBlueGT]

I am going to follow this thread very closely.

 

I used to get shift knock a little bit on WOT, but have since changed tip-in fairly drastically, even though I am using stock injectors. Tip-in is to compensate for extra air that enters the engine before it measures on the MAF. I think I have it set to 2 times stock value at low tip-in and 3.5 times at high tip-in. I have headers though, so that might change things a bit too.

 

Even after I changed tip-in I still get shift-knock now that I am on winter gas. it usually happens for me 200 ms after a shift. I have found that I can make it knock worse st about 1/2 throttle shifts in the 3200-4000 rpm range. it seems that the MAF not adequately keeping up for a brief period causes the engine to read too low a load, and this means more timing, then may be optimal for that particular load/rpm point. At least that is my theory. I have been backing off on the timing at slightly lower loads, by up to 3 or 4 degrees or more!!!! This seems to have helped, but not gotten rid of it.

 

The other thing I have found is I backed off a lot on my fairly aggressive AVCS settings. I lost about 200 rpms of spool, but again shift-knock seems better. I have also seemed to find some value in the OEM AVCS dip on the 3000 rpm range (or so) at low load. I replicated that in my latest strategy and again have found shift-knock harder to reproduce.

 

But. Alas, I still have some.

[LittleBlueGT]

My tune is very pretty for single-gear runs 2k-redline in any gear. With regular routine driving I very rarely knock (well, except for about 50% of the time I am starting from stop in first I get -2 knock around 1400-1800rpm).

 

Anyhow, if I floor it in first, shift around 5500-6000rpm into 2nd and resume WOT, I get knock about 2/3 the time about a half second after resuming WOT. This pulls timing, anywhere from 2 degrees to 10 degrees. The same thing happens going into 3rd, but less frequent. It also happens going into 4th.

 

It means timing is pulled several degrees at the very times when I want the most power. Not good.

 

I don't know if it is really knock or not, first of all, but either way it is very transient and pulls timing that persists as I sprint through the gears. Bottom line is itt slows me down. Since I haven't been able to get rid of the knock phenomenon, I am enabling the ECU to restore timing quicker (to get rid of feedback negative correction quicker). It means that even if I get knock and subsequent timing pull, the duration and/or severeity of the timing pull will be less, putting less of a damper on my spirited runs.

 

 

I should add this:

 

I talked at length with Jarrad and Tim at PDX about this and they both said, "engine mounts"

 

I now have group-N engine mounts in and very very rarely get shift-knock at WOT. I use to get it fairly often, but now only 5% of the time. If you have ever watched our engine twist while it is on the dyno you could see why it could possibly make noise.

[Legend]

LBGT: For my reference, what are your righter-most tip-in values? The stock map is at 3980 and right now I am around 4400. Motor mounts... I may end up exploring that option too eventually. Good info.

 

I had backed off timing in the 3600-4400rpm mid-load range up to 8 degrees to try to combat this phenomenon. In addition I added a little more fuel in this region too. Doing all that didn't seem to make much of a difference.

 

On Thursday I will add at least another 10% tip-in and I am going to change the advance value to 4 (rather than 1). My experience is that when my timing is pulled (somewhat drastically at times), it always gets cleaned up nicely... but it takes forever. I am going to have the ECU add a nice big chunk of 4 to negative KC, which will clear it up most the time in one fell swoop. Anybody think this is too aggressive?? At least I wouldn't have a whole run through the gears ruined... often when I get shift-knock in second it still persists all the way into 3rd.

[LittleBlueGT]

LBGT: For my reference, what are your righter-most tip-in values? The stock map is at 3980 and right now I am around 4400. Motor mounts... I may end up exploring that option too eventually. Good info.

 

I had backed off timing in the 3600-4400rpm mid-load range up to 8 degrees to try to combat this phenomenon. In addition I added a little more fuel in this region too. Doing all that didn't seem to make much of a difference.

 

On Thursday I will add at least another 10% tip-in and I am going to change the advance value to 4 (rather than 1). My experience is that when my timing is pulled (somewhat drastically at times), it always gets cleaned up nicely... but it takes forever. I am going to have the ECU add a nice big chunk of 4 to negative KC, which will clear it up most the time in one fell swoop. Anybody think this is too aggressive?? At least I wouldn't have a whole run through the gears ruined... often when I get shift-knock in second it still persists all the way into 3rd.

 

The motor mounts are only noticeable on idle to a trained listener/feeler. Still just fine for a DD.

 

I was backing off timing on the line in the log that showed the pulled timing as well. I think the timing there is OK. (basing this off your 8 degree comment)

 

I started to pull timing from the line in the log just before the pulled timing, this line was usually at significantly lower load. I see that you log g/s and rpm (and then likely use Airboy's spreadsheet to compute load). Try logging ECU load and comparing that to calculated load. You will see that it changes very quickly, and is likely behind by a hundred ms or so. This is enough to have a couple of fires at 25 or 30 degrees timing when it should really be at 14 or so. Just a though, but it does seem to be working for me so far.

[LittleBlueGT]

I personally don't think what you are doing is too aggressive, as long as you are always watching and logging.

 

I don't know if the proposed changes will affect the ECU pulling DA slower or not.

 

Here are my tip-in values. I am still working on it, and I have headers, so who knows if they will work for you or not.

 

 

 

If you log feedback correction and display it on your dashboard you will notice that when the ECU pulls timing but you keep your foot in it, it takes a long time to go away. If you let off the gas, the pulled timing disappears at a much faster rate.

[Deer Killer]

Enginuity.org is still down so I'll post here. I DL'd the 774 Enginuity, new defs, etc. and loaded my latest map. I see some tables under Knock that could be promising for the "shift-knock" phenomenon (about half of my spirited shifts into 2nd, 3rd, or 4th generate -2 to -8 FC (about 400ms after a shift and resuming full throttle)) There are now a lot more options for dealing with knock! In particular,

 

- Feedback Correction Retard Value - We can decrease the knock value from -2

- Feedback Correction Negative Advance Value - We can increase negative correction in larger increments too...

- Feedback Correction Negative Advance Delay - and we can increase negative correction at a quicker rate too!

 

Safety is concern #1 so my gut tells me to keep the retard value at -2.

 

I think targeting negative correction might be the way to go. I am thinking of increasing the advance value to 1.5 degree (vs. 1.0) and decreasing the delay to 100ms (vs 250). Since these shift-knock events are VERY transient (and may not even be knock anyhow), these changes seem reasonable. If I made these changes the shift-knock phenomenon would still exist, but the negative correction would get cleared up MUCH more rapidly. This would allow my runs through the gears to have something much closer to the optimum timing I have tuned for in single-gear runs.

 

Thoughts?

 

This just seems like a bad idea. Keep in mind you are missing dozens if not more ignition events between polls with the datalogger. If we were to assume running higher cylinder pressures makes the knock detection less sensitive, you should be increasing these values if anything.

[LittleBlueGT]

This just seems like a bad idea. Keep in mind you are missing dozens if not more ignition events between polls with the datalogger. If we were to assume running higher cylinder pressures makes the knock detection less sensitive, you should be increasing these values if anything.

 

I am not disagreeing with your reasoning, but the ECU does not miss the knock events. If they happen it shows up in feedback correction. You may not know at exactly what ms it happened, but you know it happened within about a 200 ms window.

 

Many stock Subarus knock, right off the lot. IMO they run waaaaaaay too much timing in low load situations, and of course stay in CL far too long.

 

This LGT is tuned significantly better then that. I think he is OK doing it.

 

I also think that if he keeps trouble-shooting he will solve the problem w/o changing the knock control settings.

[bugblatterbeast]

my shift knock issues were fixed with adjustments to the primary ignition table tip-in and FIBET tables. FWIW, it took a bloody long time to dial in FIBET and tip-in because the logging isn't fast enough to catch what one is looking for.

[LittleBlueGT]

my shift knock issues were fixed with adjustments to the primary ignition table tip-in and FIBET tables. FWIW, it took a bloody long time to dial in FIBET and tip-in because the logging isn't fast enough to catch what one is looking for.

 

I am just starting to understand what FIBET does, but I can't seem to understand how to adjust it to get the correct tip-in.

 

It also means your target boost tables have to be setup 100%.

[mickeyd2005]

The new tables that OP are referring to are described here:

 

http://forums.openecu.org/viewtopic.php?t=3029

 

They are part of the new defs release.

 

OT, be aware that the latency definition has been changed in enginuity for 32 bit ecus. This was something the enginuity community discovered was wrong on the 32 bit ecus. The whole description is in a thread on enginuity.org. We definitely knew it was not consistent with 16 bit ecus because the 2007 32 bit cars now use topfeed injectors so we had something to compare. Most people adjust latency via AFR correction percentages so it shouldn't matter how it is tuned. Just be aware that the numbers are now displayed differently.

[Deer Killer]

speaking of changing defs (sorry about the hijack) mickeyd have you noticed that the boost/wastegate tables are in different places with 522N? I'm running stock boost with the maps translated from 510N grr

[Gator GT]

 

my goodness....apology for the hijack, but what is a good source to learn about the parameters of the ECU and tuning? You guys seem to know a thing or 2 about this stuff (<-- sarcasm....more like 20 to 30 things about this stuff!)

 

thanks.

[mwiener2]

Ever consider it's not knock at all and the super sensitive knock sensor is picking up noise from the actual shift itself?

[Legend]

LBGT: Thanks a lot for your tip-in graph. My current tip-in increases from 1% (greater than stock values) on the left up to about 10% on the right. You are showing me I have some room to crank it up some. And I will.

 

Deer Killer: I really don't want to hamper the ECU from pulling timing when it should. To that end I think I will keep the stock retard of -2. BUT, for a large knock event (timing of DA -6 or -8), I think it is unreasonable for the next 15-30 seconds of driving to suffer as Feedback KC remains negative. I know from logging my WOT single-gear runs that I don't knock. I don't want the ECU piddling around adding 1.0 degree every second or two to big negatives when I know it'll run very happily at the full timing I've tuned it for right now. And...

 

mwiener2: (con't from above)... these knock events may not event be knock anyhow. It could be clutch chatter, the torque of the engine on the motor mounts, etc. And if so, all the more reason to restore timing to what it ought to be when I am WOT.

 

 

I'll work on this some more Thursday. Thanks for the comments, guys. The contrary ones definitely welcome too.

[Deer Killer]

What would be interesting - set your WGDC/boost targets to zero, reduce timing by a few degrees, set CL delays to 0, and then see if you can reproduce it.

 

Also another value of interest in the new defs:

 

CL Delay Maximum Engine speed (neutral) = 6000 RPM.. how about something like 2000 rpm?

[Legend]

I got out and logged this a.m. with a few changes:

 

1) Retard - set back to factory -2 as I couldn't see any good reason to change it and want my LGT logs to look like LGT logs.

 

2) Advance - set all the way up to 5 (vs. 1 for factory). Again, my rationale for doing so is to keep from timing being pulled for a ridiculously long amount of time due to a knock event.

 

3) Delay - set to 200ms (vs. factory 250ms). Same reason as above, however I am a little reluctant to shorten the delay too much as I do want the knock to clear.

 

4) Tip-in - Thanks to LBGT, I have some confidence really ramping up my tip-in. Previously I had max tip-in around 4400 (vs. 3980 for factory), but since LBGT showed me that his max is >10000, I am cranking it. My peak tip-in (>31.3 throttle) was set to 6200.

 

In addition, my logging parameters were different. The logs are attached.

 

My conclusions thus far:

 

1) OL/CL - I logged OL/CL status on each run. On these runs I do stay in OL, which is good and can now rule out a role of OL/CL transitions in this phenomenon. I won't need to log OL/CL status again.

 

2) AFR - I don't have wideband, but from the logs it doesn't look like I am leaning out... although the working premise is that I am. Since there doesn't look to be much value to logging AFR for this phenomenon, I won't need to log this again either. (I think).

 

3) Advance - Damn I love this!! See run3 where I get triple-knock going into 3rd gear. Instead of feedback KC remaining negative for a very long time, the first advance of 5 clears up the negative KC and full timing is restored AND the knock sensor is happy. Still gotta deal with the knock but I am pleased not to be penalized for the whole gears run.

 

4) Tip-in - Looks like the big increase may have helped.

2nd gear knock events before - run1 had 3, run2 had 1, run3 had 6(!) for a total of 10 knock events

2nd gear knock events now - run1 had 1, run2 had 0, run3 had 3 for a total of 4 knock events

 

3rd gear knock events before - run1 had 2, run2 had 5, run3 not applicable due to feedback KC already -8

3rd gear knock events now - run1 had 0, run2 had 1, run3 had 1 for a total of 2 knock events

 

So, in judging the role of tip-in it looks like I have made good progress. Previously in 2nd and 3rd gear I had 17 knock events in 3 runs and this time I only had 6! That's a good improvement! In addition, the severity of knocks looks to have improved, as the worse I had this time was a triple-knock where before I had a 5x knock and a 6x knock.

 

So, I am making some changes and heading out again soon...

run1.pdf

run2.pdf

run3.pdf

[LittleBlueGT]

1) Retard - set back to factory -2 as I couldn't see any good reason to change it and want my LGT logs to look like LGT logs. I might put mine to 1, but I will wait until I get a knock light.

 

2) Advance - set all the way up to 5 (vs. 1 for factory). Again, my rationale for doing so is to keep from timing being pulled for a ridiculously long amount of time due to a knock event. I put mine to 3, and noticed similar results.

 

3) Delay - set to 200ms (vs. factory 250ms). Same reason as above, however I am a little reluctant to shorten the delay too much as I do want the knock to clear.

 

4) Tip-in - Thanks to LBGT, I have some confidence really ramping up my tip-in. Previously I had max tip-in around 4400 (vs. 3980 for factory), but since LBGT showed me that his max is >10000, I am cranking it. My peak tip-in (>31.3 throttle) was set to 6200.

 

Tip-in is partially a function of injectors, but mainly a function of how well the engine moves air. Changing the way it moves air by headers, DP, intake etc, will necessitate a change in tip-in. I don't think tip-in is the main culprit though.

 

In addition, my logging parameters were different. The logs are attached.

 

My conclusions thus far:

 

1) OL/CL - I logged OL/CL status on each run. On these runs I do stay in OL, which is good and can now rule out a role of OL/CL transitions in this phenomenon. I won't need to log OL/CL status again.

 

How did you test this (10 vs 8)? I have noticed that the ECU registers a 10 when it is in decel mode. Like when it runs ultra lean. I am not sure if that is just another mode of OL. Just throwing this out for discussion. I was thinking of changing the throttle position settings vs rpm (I think it is requested torque table in Enginuity). That way above a certain rpm the engine will not go back to 0 throttle position and stay at the primary fuel table's AFR vs going ultra lean. Am I saying this right?

 

2) AFR - I don't have wideband, but from the logs it doesn't look like I am leaning out... although the working premise is that I am. Since there doesn't look to be much value to logging AFR for this phenomenon, I won't need to log this again either. (I think).

 

3) Advance - Damn I love this!! See run3 where I get triple-knock going into 3rd gear. Instead of feedback KC remaining negative for a very long time, the first advance of 5 clears up the negative KC and full timing is restored AND the knock sensor is happy. Still gotta deal with the knock but I am pleased not to be penalized for the whole gears run.

 

4) Tip-in - Looks like the big increase may have helped.

2nd gear knock events before - run1 had 3, run2 had 1, run3 had 6(!) for a total of 10 knock events

2nd gear knock events now - run1 had 1, run2 had 0, run3 had 3 for a total of 4 knock events

 

3rd gear knock events before - run1 had 2, run2 had 5, run3 not applicable due to feedback KC already -8

3rd gear knock events now - run1 had 0, run2 had 1, run3 had 1 for a total of 2 knock events

 

So, in judging the role of tip-in it looks like I have made good progress. Previously in 2nd and 3rd gear I had 17 knock events in 3 runs and this time I only had 6! That's a good improvement! In addition, the severity of knocks looks to have improved, as the worse I had this time was a triple-knock where before I had a 5x knock and a 6x knock.

 

So, I am making some changes and heading out again soon...

 

Again, I don't think tip-in is the real culprit here. I think it compounds the problem. The conclusion I am leaning towards is that the load is climbing very quickly, and the ECU is one step behind on the timing table. I sometimes manage to get a log with shift-knock where I see the overall timing is 31, then the next line it is 14. I think that delay is what is killing us.

 

I will try and run real high octane gas for a while, and report back.

 

FWIW I ordered a Turbo XS knocklight. I don't want to mess around with changing the feedback correction settings w/o being able to monitor knock. I ordered it from PDX, they have it on special right now for less the $100, vs $139. You might want to consider it.

[SeeeeeYa]

Again, I don't think tip-in is the real culprit here. I think it compounds the problem. The conclusion I am leaning towards is that the load is climbing very quickly, and the ECU is one step behind on the timing table. I sometimes manage to get a log with shift-knock where I see the overall timing is 31, then the next line it is 14. I think that delay is what is killing us.

 

I will try and run real high octane gas for a while, and report back.

 

FWIW I ordered a Turbo XS knocklight. I don't want to mess around with changing the feedback correction settings w/o being able to monitor knock. I ordered it from PDX, they have it on special right now for less the $100, vs $139. You might want to consider it.

 

Since I'm from the more sophisticated side of the tracks, having a 5EAT , I have no such issues as is being debated here, thankfully.

 

However, I have often considered this knocklight but about the time I decide to jump because of positive comments someone will burst that baloon with how functionally useless it is . The detractors say that set sensitive enough to catch knock it blinks uncontrollably because of noise, and when it is adjusted for a realistic knock threshold it doesn't catch knock when you need it to.

 

Add another reason I monitor this thread. One now will be to follow LBGT's progress with the Knocklight. The truth will come from him, finally.

 

Secondly, this is but another in an accumulating series of threads that IMO should be in a dedicated TUNING FORUM .

[Legend]

How did you test this (10 vs 8)? I have noticed that the ECU registers a 10 when it is in decel mode. Like when it runs ultra lean. I am not sure if that is just another mode of OL.

 

I logged OL/CL status and came to the conclusion that I stay in OL on runs through the gears by observing that status stayed at 10 (open loop) on every line of each of the 3 logs while racing through the gears.

 

I was thinking of changing the throttle position settings vs rpm (I think it is requested torque table in Enginuity). That way above a certain rpm the engine will not go back to 0 throttle position and stay at the primary fuel table's AFR vs going ultra lean. Am I saying this right?

 

I haven't looked into that. Seems like it might be worth playing with.

 

I really haven't had too much success as I ramped up tip-in to nearly the values you are using, LBGT. My top tip-in value is 9200 and the phenomenon still exists. I didn't expect tip-in to eliminate the problem, but I did expect it to greatly reduce the frequency and severity of knock events. I even went so far as to richen fuel to the failsafe values in 3600-4400 in loads less than 2.5 and dropped the timing a few degrees in 3600-4400 in loads less than 2.1 or so. These changes, coupled with the tip-in, still allow the phenomenon to exist. Bottom line is that I (once again) am coming to the conclusion that this can't be tuned out.

 

I am still very happy, however, with the advance at 5. That really mitigates the consequences of the problem, at least!

 

I hope your knocklight can shed some light on this so I'll be curious to know what you find.

 

Tuning forum on LGT.com?? Sounds great - sign me up.

[LittleBlueGT]

How did you test this (10 vs 8)? I have noticed that the ECU registers a 10 when it is in decel mode. Like when it runs ultra lean. I am not sure if that is just another mode of OL.

 

I logged OL/CL status and came to the conclusion that I stay in OL on runs through the gears by observing that status stayed at 10 (open loop) on every line of each of the 3 logs while racing through the gears.

 

I was thinking of changing the throttle position settings vs rpm (I think it is requested torque table in Enginuity). That way above a certain rpm the engine will not go back to 0 throttle position and stay at the primary fuel table's AFR vs going ultra lean. Am I saying this right?

 

I haven't looked into that. Seems like it might be worth playing with.

 

I really haven't had too much success as I ramped up tip-in to nearly the values you are using, LBGT. My top tip-in value is 9200 and the phenomenon still exists. I didn't expect tip-in to eliminate the problem, but I did expect it to greatly reduce the frequency and severity of knock events. I even went so far as to richen fuel to the failsafe values in 3600-4400 in loads less than 2.5 and dropped the timing a few degrees in 3600-4400 in loads less than 2.1 or so. These changes, coupled with the tip-in, still allow the phenomenon to exist. Bottom line is that I (once again) am coming to the conclusion that this can't be tuned out.

 

I am still very happy, however, with the advance at 5. That really mitigates the consequences of the problem, at least!

 

I hope your knocklight can shed some light on this so I'll be curious to know what you find.

 

Tuning forum on LGT.com?? Sounds great - sign me up.

 

 

I noticed the solid 10s as well.

 

Log decel, 0 throttle, you know where the stock O2 will show 20 AFR. It will read 10 as well. I think that will be enough for you to know what I am talking about.

 

You are coming to the same conclusion as me, we can tune everything else around it, hoping to help it, but these are not the root cause. Maybe we should change to a speed density set-up. If we could have the MAF after the TMIC I think the problem would be gone as the MAf would be more accurate on transition and the calculated load would also be correct.