NefMoto

I have been trying for days to get my actual lambda to meet requested.  I have been bumping the KRKTE up and I have been bumping up KFKHFM on the higher load areas.  The car runs REALLY lean in boost.  Can someone take a peek and let me know what you think, and possibly point me in a direction to fix the problem?

CSV LOG, ECUX LOG, and latest revision of my tune attached.

Are these correct? both requested and actual seem miles off for both load and Boost..

1) Consider using the filter option.

You are using 2nd gear, so set the filter to look for 2nd gear

2) Consider using 3rd gear

Your log isn't very long (time wise)

3) I don't see that you are running lean (compared to your req afr). Do you have a wideband?

I don't have my laptop handy but when I do I will check out the file, in the mean time what did you change?
Did your lc1 reset?  If it did its going to be whacked and you have to reprogram.  What's the wideband reporting?

Maybe I don't understand my log.  I was watching my O2 sensor voltages and they seem to et around .937ish at the lowest point.  I   thought that was around 13.xx af at peak load.  This is pretty much the trend in all my recent logs.  No matter how much I bump KRKTE or KHLFM (SP) it doesn't give me what I thought I was looking for.

Changes I have made that affect afr.....I think it was just the screen shot you posted in the lamfa thread.  I pretty much copied it trying to figure out this lean condition.  I have also been messing with the table I mentioned in the first post.....adding on the higher load areas trying to.get more fuel, and pulling from the areas around 30-60% load to help with trims on the highway.

I didnt know the lc-1 could reset. I assumed since its still reading narrowband on putn outputs that its still set.  I do recalibrate it every time I pull the battery though.

Never EVER trust narrow band voltages. Get a wideband. Get a proper log (3rd).

What are your LTFTs?

It's a lc-1 emulating the two narrowband sensors.  I have a lm-1 that I can run too.  It's a bit nicer than the lc-1 setup since it isn't relying on system grounds to give me a actual reading.  Both of the lc-1 outputs are being used to satisfy the ecm.  I haven't used the lc-1 for logging yet.  Don't know how.  I just used it to eyeball afr. 

Where is the filter option?  How long of a pull should I log?

the filter option in ecuxplot....

youll want to log 3000 to redline, ideally

I cant get ECUXPLOT to work for me.  I spent a hour or so a while ago trying to get my logs to display on the graphs, and they just wouldnt.

AND......I am a dumb a$$.  I hooked up the LM-1 and AFR are as desired.  I can now start dialing back my KRKTE since my LTFT are retarded right now because I have adding fuel. 

Next is the load issue.  I have noticed that my load never really matches.  Its always requesting much more load than actual.  Even at cruise its rewuesting around 119, while actual at cruise is around 40-60.  How do you line this up? 

Did you read the FAQ? Did you configure your filter to match the gear you're logging? Why are you logging 2nd?


without knowing more about your MAF setup and/or the scaling you are using, or your KFMIRL, there's no way of saying. typically it doesn't matter much. you can run an underscaled maf just fine most of the time.

85mm Maf, same scale as the Nefmoto stg 3.  KFMIRL is in the bin I uploaded in the first post. 

Link to FAQ please.  I downloaded it from your link and didnt see a FAQ. 

I think the log I posted earlier was 2nd and 3rd gear. 

http://nefariousmotorsports.com/forum/index.php?topic=1714.msg16497#msg16497

Duh, I shoulda known its in the forum I never go in.    THANKS!

I left the LM-1 connected and watched the AFR on the highway today on my way to work.  THey stayed at 14.3-14.9 range.  I am going to call this good for now.  The voltage reading from my LM-1 are spot on for the AFR I was running, and this is what I expected from the LC-1.  I just dont understand how the ECU gets the right AFR if the o2 reorts something wacky.  Or maybe it isnt reporting wacky numbers, and I just dont understand what is going on. 

I also noticed that my LTFT didnt work.  They stayed all zero.  I am Ok with that, when the LTFT do actually kick in it usually splits my banks to where one bank is adding and the other bank is removing almoost the same amount.   I can actually feel it when the split goes over 10%, and as a result I have to keep my laptop in the passengers seat with VCDS open and reset the DTC's to clear out the LTFT.  This morning there was none of the crap, and it was actually quite nice having one less thing to worry about.

I tried for a third gear to WOT and scared the piss out of myself.  Right towards the top of the range, my 90 at the Throttle body blows off and I think it hit the hood pretty hard.  Loud Bang!!  followed by an immediate loss of power, rough running, I thought I blew the motor.  Anyhow here is the log.  I think I have now maxed out the MAF as scaled and maybe a few other problems the pros can see.  

I also mofified the BIN to KFLDRQ2.  I bumped up some of the values in it to help bring the boost on sooner.  I didnt want to mess with IRL since I dont fully understand how it all works together with IOP and any other maps linked to them.  I was noticing that the N75 was going to the 95% DC a bit later than I wanted.  I need it to go to a high DC to keep the VGT closed so I can spool the turbo. 

That isn't what Q2 does :(

What does Q2 do?

Q2 is the derivative portion of the PID. "Bumping" it would make things worse (if it actually was causing your DC to be too low, which it isn't)

Have you read TTQS' ME7 FR translation? Specifically, the LDR module?

S4 wiki:

"If you aren't using K03s, you may have to tweak the PID response. Note: this is NOT used to adjust requested boost. It is used to compensate for different waste-gate responses.
 KFLDRL - Map for linearization of boost pressure = f(TV). This is the post-PID waste-gate duty correction table.
 LDRQ0S - LDR PID Q0 in static operation (proportional term)
LDRQ1ST - LDR PID Q1 in static operation (integral term)
KFLDRQ2 - LDR PID Q2 (differential term) "

I looked through all the files I have for STG 3 cars for days and days trying to figure differences in the PID area.  This map was the part that had the biggest diference.  Maybe its all wrong, but I do know that the results of me changing this map has made the requested and actual boost line up better in the logs.  I am actually getting a really good amount of boost down low. 

I flashed the above file and the difference in boost response was night and day.  All i needed the N75 to do was get to a hihger duty cycle sooner in the RPM band than it was.  I didnt need to add more boost.  I am just tweaking it so I can block the boost from the VGT actuator so it can stay in its smaller position longer and spool faster.  Do you have any other suggestions?

If you increase Q2, you will get much more WGDC cut in response to the actual boost rapidly approaching requested.


It is unlikely you'll see anything but 95% wgdc (irrespective of ANY PID settings) if requested boost is much larger than actual, unless you programmed LDRXN wrong and are getting positive deviation limp (in which case you'll see 10% wgdc).

Maybe something else is wrong, But I do know that this map is the only map I changed that relates to the PID since the first post.  As seen in the last LOG it goes 95% DC around 2.3k which was perfect.  Before it would lazily progress to 95%, starting around 2.5k@~25-40%DC. By the time it hit 95% duty cycle I am assuming the VGT was open too far and it took a while to spool the turbo in its larger AR config.  When it progressively went to the higher DC the car didnt hit full boost until around 4.5k.  Boost would slowly ramp up from 2.5 to 3.5k...~10psi.  This turbo has spooled much faster than that with the VGT locked in position.

 Kfldrl is damn near maxed out, and I still couldnt get it to keep the N75 closed.  LDRXN is 207 from 2500 rpms up<------DAMNIT!  I changed this map and forgot. LDRXN was quite a bit lower in the low end.  @2500k it was 188, I changed it to 207 right before the log.  I will change Q2 back and see what happens.

I am kind of aiming for it ito be a backwards kind of preogression. 95% to start to keep the VGT closed, and then slow down the DC to allow the VGT to open up in a controlled manner so it doesnt kill spool.  I am trying to get it to stay closed until around 15psi, and let the larger AR take over from there   SLOWLY.  If it opens too fast or too soon the ECM gets pissed and intervenes with throttle cut, since boost falls on its face.

It is true that the WGDC stays high.  Looking at the log it stayed too high for too long and Boost kept climbing fast.  I was hoping to get the DC to drop and slow the boost once it was at set PSI.

This is getting really frustrating.  LTFT decided to intervene today causing one of the banks to start dumping fuel as seen in the log attached.  What other factors make the banks fuel separately? 

I have one O2 sensor and the wideband is feeding analog siganals to the ecu for both sensor inputs to ECM.  The voltages are not exactly the same between the banks, but not far off enough to wher I have seen from +/- 25% on one bank and almost the opposite on the other. (extreme example)  I can feel when the trims go past 10% in either direction.  The car has a slight hesitation.

I clear the LTFT every morning before I go to work.  The trims will go to around 5-6% and stay there for about the first 30 min of my 45 min highway drive.  Then all of a sudden the ECU starts adding more and more fuel to one bank, and pulling almost the opposite from the other bank.

I am going to recal my LC-1 in a few and see if that makes a difference.  To date I have recalibrated it many times over and it hasnt really made much of a difference.

Perhaps KFLDIMX is interfering (see tuning wiki).


You don't get to control that (and you don't want to). You just set req boost, and the PID should do the right thing in all conditions. If you still can't get the PID to behave, you can consider using KFLDIMX/KFLDRL to get open-loop control.

Finally, for proper tuning of the PID, ECUx won't cut it. You have to use ME7L to log the various PID related variables that aren't available in ECUx.

I just added 10 to the area inducated by the WIKI. 

I just went back to my basic LDRXN table, and made some more changes in the PID limiting table.  I didn't like that the PID was always at 95% becasue as explained to me earlier in this thread:
     "It is unlikely you'll see anything but 95% wgdc (irrespective of ANY PID settings) if requested boost is much larger than actual, unless you programmed LDRXN wrong and are getting positive deviation limp (in which case you'll see 10% wgdc)." - NYET

I want to actually control the actuator not just block off the boost from it.  I really need to figure out the ME7 logger apparently, if I want to get proper control of the PID.  

 

I don't understand how one follows the other :( If your deviation is large enough, the P portion will put the DC at 95% REGARDLESS of your I limit. Unless, like i said, your LDRXN is wrong, and you are getting positive deviation limp because you requested more boost that you can possibly make at a low RPM.


The only way to control the actuator is to progressively block boost from it, unless you are running an electric wg.

I sense you really have no idea what is going on or how the PID and/or N75 work.

jeez dude...  lmfao...

pid's aren't easy to grasp--- even for people with masters in engineering and highly trained applied mechanical designers...

Nvm.   Thanks.