NefMoto

Hi

Another newbie question again :)

Just wondering how KFLDRL 'linearises' the boost.  I know I can cap N75 DC% with KFLDIMX however since using the 032hn KFLDRL on my k04 ECU running k03s turbo the boost was 5psi less.  IMX values are still up in the 95% through spool dropping to ~85% midrange so should have plenty of duty to keep things boosting.

A couple pics of the maps in question are attached

Top - KFLDIMX
Middle - KFLDRAPP
Bottom - KFLDRL

Im not too sure how/what KFLDRAPP does but the 032HN one was different to K04 so I changed that over too.

Does KFLDRL use the bottom line for full throttle since thats 95% throttle plate angle?  When i did a quick log the duty cycles were ~65% and actual engine load not meeting specified.

Anyone any pointers?

1) it does not linearize boost, it linearizes the wastegate response to the output of the PID
2) KFLDRL has NOTHING TO DO WITH THROTTLE PLATE ANGLE
3) KFLDRAPP does nothing unless you enable it via CWDMAPP

4) KFLDIMX does not cap N75 DC. It is the base for the upper and lower I limits.

Bit more reading for me it appears  Shame everything is in german on the FR

Doesnt help in a poor learner just by reading things either.  I have to put into practice the parts and then top it up with the theory after.

The result of the PID is the input to KFLDRL. The output of KFLDRL goes to the N75. That part of the FR is not in german, it is in the block diagram.

You'll also want to brush up on how PIDs work and why the output of PID often has to be linearized before feeding it into the control system...

Google translate is your friend. It has really come a long way. I feel it translates better than even a few months ago.

So far the best explanation of PID tuning I have found to date: http://youtu.be/SefKQb9y_B4

I reallllllly would like someone to help me rewrite the PID section of the s4wiki.

working on it.

That video is pretty helpful in understanding how the PID works to get the output wanted.

Looking at the LDRPID section I think I can see the P,I,D sections of it.

P - ldptv
I - lditv_w
D - ldptv

These will stabilize the boost to the output map of KFLDRL, which will then give the duty cycle to the wastegate?

Does this then mean that the base N75 input maps to the PID controller are: (P) - KFLDRQ0, (I) - KFLDRQ1, (D) -  KFLDRQ2


Sorry if im talking a load of rubbish, but ill try get my head round it all at some point

yes

you are getting there.

now read the attachment.

 I am going to refine it a bit more and then write my interpretation of the sentences that still aren't making total sense. I finished at like 5 am and went to bed so I still need to reread it still. I have to go to work soon though so after.
The text is a complete and direct translation of the LDRPID section from the FR through Google translate. I manually corrected what Google translate didn't as far as accent marks above letters and moving words up lines in the translate input box so that the translator didn't get so confused on where the end of the sentences are. Some of the German words have capitals in the middle of sentences which screws shit up in google translate, also periods(.) in odd spots help not either..other translations I've found seem to have simply C&P from the FR into a translator and left rough, and it makes a big difference in the translation.

During spool :
Q0DY - P shoots up and is your main influence
Q1DY - is building I up for later

Actual approaching target:
Q2 - aggresively offsets DC to correct under/overshoot

Steady state:
IMX - contains I (there are maps that offset IMX conditionally as well and adaptation)
Q0ST - small corrections
Q1ST - small corrections

P only corrects present error so it drives spool and does small corrections in steady state. I is built by the accumulation of past error, so it is the main value during steady state. D is predictive of future error by using the lde delta and is only active in dynamic state (-lde).

I will attempt to edit the translation time permitting as well. Thanks userpike!

Think ive had a bit read now.

So KFLDRL is the duty cycle map for the WG DC%.   The 'I' part of the PID will be the upper limit for the DC% which is map KFLDIMX.  Raising that will allow greater headroom for the WGDC% to meet those set in KFLDRL.

What does the % axis mean in KFLDRL. From the FR it looks to be the output of the PID controller (LDTVR_W). 

LDTVR_W - calculated output of the P,I,D functions?

Both the input and the output of KFLDRL are duty.

The input is the DC from the PID, the output is the linearized DC.

unity gain would be


such that input == output everywhere.

Look in your ME7L config (you should already have logged these before touching your PID)

Right guys, as promised id update you all once ME7logger fixed. :)

Done a few logs today.  First off, my actuator is weak.  Ive flatlined DC% at 95% to establish it and to see what boost its giving me.

The only other thing I will mention, is in higher gears, when accelerating 2000-2500rpm to do a WOT run, i get a soft limp mode.   Thankfully managed to get a limp mode run and a good run in 3rd gear where problem doesnt always happen to show you lot.

Going to have to get a 14lb actuator to put that on to let me get a bit more boost top end.

Ive attached the map file the run was used with so if you want a look at that, its there.

Cheers

Chris

You're missing all of the aliases. Your ME7 config needs work.

Also, what is the code that accompanies limp?

How do you mean by aliases and the ME7 config?  That with ME7logger?

The code where the limp mode happens if the file I attached to the post previously.  Limp happens when it hits peak boost (23psi roughly on gauge) so I guess I could try pulling the duty/ load back as im making good boost from around 175 engine load through LDRXN  (peak in map is 199) if those 2 are related.  Let off throttle twice and its back to full boost and will run perfectly, once its cleared that 'spike' stage in the map.

Still all pretty new to this at moment. 

Read this before you do anything else!

http://www.audizine.com/forum/showthread.php/479416-****-Complete-Guide-on-How-to-Take-and-Graph-Logs-****

You lost me....

Im logging all of those variables in my files (all be it a few not there which I dont need like EGT model before cat and fuel trims).

Ive no issues making various graphs from the data logged.

The way you have your ME7logger config set up seems to royally screw with xplot. I think the aliases have to be what they are originally in a fresh config file copy. You can add to the config files alias column, just don't take anything away I believe.

They show fine for me.

All I done was remove the data where there was 'nothing happening' such as between pressing start logging and the time when the runs started.

Heres the unedited ones direct from ME logger....but you will probably have to go through them and pick out the actual runs at 100% throttle.


172516 - is the good run with no limp
172539 - is the run with limp

(http://i93.photobucket.com/albums/l42/carsey2006/good.jpg)

Shows up ok for me with the files I uploaded.  Same with fuel, load, boost etc.

Me7 logger config file.

those newer UPed logs look fine. I wonder whats up with the other ones.. ??? ??? weird.

God knows....work for me.  All I did was remove the header info and the rows for the variables so I could work out what did what in ECUplot.  And remove redundant data which wasnt part of the WOT run.

Always post unedited CSV's. They wouldn't filter for me either, and I am too busy/lazy to use excel :(

Strange.

Hope you guys can work out where the runs were.  most were done from 2000rpm in 3rd and 4th gear to redline.

Hopefully you can spot whats causing the limp mode.  I think its boost overshoot as only happens silly high load areas such as 4th gear 2000rpm,  soon as it passes 21psi...it splutters and then soft limp.

Was thinking of pulling max load back a bit to try say it shouldnt make any more load than what its currently good for.

You lot will know better than me though :P

95 percent duty cycle FTW!

I don't know what to say about that.

As I said,  was verifying the actuator was weak.  hence i maxxed the maps out to make sure.  Safe to say I need a uprated one.

Still goes pretty well though hahaha.

14psi actuator needed - then can start getting boost where I want it.  Strangely - EGTs aint that high either.  Setup must be doing a good part to make sure its flowing nice.

It looks like your code is going to be P0299, or underboost limp.

No boost related fault codes at all.

You can't go limp w/o throwing codes

Don't do that. ECUxPlot depends on the headers for all kinds of things


ECUxPlot does this for you. Only post unedited logs.

I think the reason for limp must happen a few times before it throws the code. not 100% sure though


carsey, whats the altitude in your region?


maybe the turbo can't compensate enough for being in the mountains?  :-\

Had a look there and maf signal too high

Must be the cause then

135m or 445ft I think

bad MAF, time to it change out.

if you have time, log a run with the MAF unplugged.

Just curious, have you experienced lean spikes coming on throttle? I ask because I had this same MAF code but I also had a bad B1S1 02 sensor fail one of the diagnostic tests VCDS uses to determine if the sensor is faulty. The ECU threw a code for it one time in the past. After repeated DTC resets for various miner things it never came back. I did the checks anyway and ended up changing it out. So I dunno which was the cause for my lean spikes.

Brand new maf the other week before is done the crank seal.  Bosch sensor too

It's not the MAF, you're likely exceeding the values of KFLDMX.

That's the map I'm raising next.  Got the file ready to flash on at some point.  Hopefully it stops the soft limp mode.  Worth touching kfmldmn too?

this will trigger a signal too high code for the MAF?

yes

there are several MAF sanity checks.... check the FR..

keyword please? I put kfldmx in the search box in my pdf reader and nothing, kfldimx and got a short description of the map and the schematic showing the to's and fros.


sanity checks = diagnostic testing procedures?

DMX is the threshold for max airflow before MAF diagnosis is triggered. It is in kg/h.

There is also DMN, which is the threshold for minimum airflow before diagnosis is triggered, but generally speaking this isn't touched (unless you're using a HUGE housing with shit resolution).

KFMLDMX, not KFLDMX :)

Quoted for truth :o

ahha! ok!


FR says it should be set about 20% above max possible air mass over the throttle plate if I read right. I know to convert to kg/h for the mapping but I don't think he's close as the log is showing prettymuch stock numbers in g/s for 1.8t MAF data. He should be seeing way higher numbers in the first place but the boost isn't there I guess.

I say take the MAF back anyway to rule it out, it's under warranty. Just because it is new doesn't mean it isn't faulty.

add:
I went ahead and found DMX in his file and it seems as if the limits are not being exceeded according to the most current log. Even where throttle plate angle blips under 100% where the 50% column would be used in the map.

looked in the 539 log, nothing is showing up for anything until I turn off the filter. Maybe I am misunderstanding but the limp mode log is of a WOT run?


an attempt anyway happens it looks like, then maybe boost hits the map limit, then limp kicks in, you stomp your foot on the accelerator pedal 3 times and power comes back on and you finish it out through 6600rpm?

Since it seems they have been lost in the mix, here's a link to 2 fully translated versions (using translation tools, not human, but it is the entire Funktionsrahmen). I use these all the time, and between these translated versions and the English Alfa version, I usually am able to get everything pretty clearly.  Check it out:

http://nefariousmotorsports.com/forum/index.php?topic=555.msg5138#msg5138

or could be the accel pedal potentiometer(s)?

Thats what happens userpike...

High gear (4th or 5th)  2000rpm,  gets to point where it hits peak boost, overshoots to about 23psi, gauge needle flutters, drops into 5psi soft limp mode.  release throttle, press it again, (do that a couple times) and once you get full boost back out of limp mode itll pull strongly to the redline as if it hadnt dropped into limp mode.

The MAF signal looks really rough right before boost cuts out. Perhaps there are spikes (not logged, probably between samples) causing the MAF code?

IM not sure.  Your the guys with the knowledge haha.

Think the thing to try is the KFMLDMX table and see how it pans out from there.   Id go out and test it now, but the roads are damp and having road legal track tyres on it makes for very fun driving (not) and would probably just light the tyres up.

Given this:


... I'm pretty sure that isn't the right approach, aside from just proving that the MAF signal is causing limp. A noisy MAF is a bad sign... could be MAF itself, harness, flaky ground, or even an intake leak.

Aside from that what else could be causing it?   Ill try raiding the samples on me7logger up to 20/sec.

Looks like ECU is cutting the load as soon as boost goes above the max load specified in the map.  You can see if you look boost pressure PSI and max specified load,  soon as it hits the specified load, the ECU pulls it back.  Ive put a red circle round the area I mean.

What normally happens when the ECU hits its specified load through LDRXN, but the duty cycle maps are still asking for more?  (if that makes sense?)

That last one isn't limp, its throttle cut. Need to see req boost vs actual boost.

Also, whenever you post a pic, please post the log you made it with!

In any case, there are a million things that can go wrong if you don't have proper WG control.

No point in dicking around with bad hardware... get your WG fixed, and start from scratch then.

A good rule of thumb is to never bother tuning bad hardware; you'll just have to start all over when you fix it.

you sure throttle cut?  I was flat to the floor until it dropped into limp mode where I had to fully let off and then accelerate again to clear the limp.

I know what you saying about the WG, just would be good to try get my head around why its wanting to boost more causing it to cut power.

Only thing I can think of is because of the higher load making more boost in the higher gears,  hits its max spec load and WG DC isnt saying 'right, we;ve hit the target, pull some duty back so it holds it nicely'.  Instead its just wanting more and more boost.

Only reason I can think of.....

Going back to one of the first logs you posted (which is definitely throttle cut), I can only assume, since you did not post the log (or desired boost vs req boost)


The LDR PID (and hence WGDC) is NOT driven by spec load actual load error, but rather lde, which is boost error.

Ahhh got you.  I was under impression I dropped onto 5% duty cycle when it hits limp,   That would cause a throttle cut too.  Mis-understood your post.

What would happen with a stronger wastegate, tuned for a good boost profile in 4th gear,   but a higher load run was done in 6th gear for example where overshoot is more likely to happen.  How would one work around that?  Tune the boost profile using a higher gear with more load demand?

Are you aware you still have the EGT sensor enabled!

Yep,  Not crossing my 825c threshold for BTS fuelling according to logs though.

That's what the PID is for.


That is modeled EGT, not EGT sensor readings.

tabgbts_w

Thats the reading from my EGT sensor isnt it?   I have a physical sensor mounted in my downpipe.

http://s4wiki.com/wiki/Exhaust_gas_temperature


tats_w          , {EGTSensor}                       , 0x3824F0,  2,  0x0000, {▒C}      , 0, 0,    0.0195312,     50, {Abgastemperatur aus Sensor}

vs

tabgbts_w    , {EGTForPartProtection}      , 0x384C0C,  2,  0x0000, {▒C}      , 0, 0,    0.0195312,     50, {Abgastemperatur f▒r Bauteileschutz}

Got both of them in my log.  None go above the threshold.

Not sure why he brought it up in this context... but it is common practice to disable the EGT sensors so when they fail, the ECU doesn't dump fuel.

IMO this has nothing to do with the problem you are seeing.

In any case, I generally keep the EGT sensors enabled.

 I brought it up as he is running K04 software on a K03s turbo! So unless he has put the EGT into the downpipe or something similar then he hadn't disabled EGT in the map...

Should all be running the k04 stuff.

Only stuff that isnt there is VVT solenoid as i cant find one anywhere :( - CDNWS = 0 and turned off in ESKONF (havent touched any cam maps)
Rear lambda delete
SAI delete
evap delete
Charcoal canister delete
N249 delete

All those done as per the wiki.

Ah! thanks for the explanation!

I would leave the WG alone. I don't understand why you think it isn't working correctly. You are getting so much boost that you are hitting the hard boost limit at times, which is like 2550ish mbar or like 22.5 psi on a boost gauge, the reason for your limp mode. To get any higher boost with ME7, it is my understanding that you'll have to go the 5120 hack route.

He's trying to run MORE boost with a k03.

It's so far out of it's efficiency range, he'd make more power with less, but that's his call.

When I said boost is hitting the MAP limit in my prior post I meant the max boost the ECU is capable of registering, the hard MAP limit.
ldrxn is a guide for the max you ever want load to ever be.

 You are getting boost overshoot past the MAP limit and going into limp. You need to cut down on the WGDC% and adjust the PID more. ..and if you are running k04software on a k03s I would start with a stock DRL and LDRPID(yeap, every parameter) from a file calibrated for a k03s.

no worries, that's what the NO2 kit on like 3 fmics is for! gain back some of that loss you know.. ;D

oh. they are k03s?

wow. yea. don't touch the WG. you're wasting your time, get a k04.

Not they Nye, IT.

Whats wrong with a 21psi spike on a k03s and holding good midrange boost to the redline?  Its the way 99% of tuners in the UK run a k03s.   of course, anything too much will just cause high EGTs and stupidly high intake temps, but my setup is built for free flowing boost, little pressure drop and maximum cooling efficiency.

Handy to know about the hard limit though.  might pull back LDRXN and get that wanting a 21psi spike.

what's wrong is that you will have a turbos replacement expense in short order...

Got a golf Edition 30/ Golf R turbo to go on. :)  if this one dies (ive ran k03s at more midrange boost before) then the replacement will ahve to go on :D

What is wrong with you are doing? Nothing, you are just operating inefficiently so you will gain more torque if you will drop your boost. Running such silly high boost on small K03S you are just heating air despite having big and good intercooler.

Personally I think IATs are pretty good considering the air temp was about 14c on the day those runs were done.  If they were above 30-35c.,....id be more worried.  Water methanol is on the cards to further aid cooling and little octane boost.

It is up to you but still even after you waste money for water/meth injection you will only gain from droping boost to get into efficient operating range of your turbocharger...

What would you class as effictive operating for a k03s then?

check the compressor map :)

btw ecuxplot will plot PR vs flow.

I need to redo the pressure drop stuff though, its currently a rough estimate.

(http://s4wiki.com/images/thumb/7/75/K03-1870_EXA.png/724px-K03-1870_EXA.png)

At redline you don't want more than 11-12 psi.

That sounds a pretty good feature.  is the a guide on how to put a compressor map into ecuplot?  I can see the Load preset > Compressor map.  but unsure how to use it properly.

I've been meaning to add a way to import compressor map datasets for overlay, but it is a pretty big feature that would take a lot of work :(


you have to set up maf scaling and # of turbos.. that should be it.

So is this K03-029 map?

Does anyone happen to have K03-073 map?

Got a new load map which is capped at 180, since that was the value where my boost was hitting where I wanted it to...roughly.  Hopefully that will stop it wanting to boost over the map sensor limits.

Im currently running no VVT solenoid....coded out in ESKONF and the CDNWS set to 0.   Is there anything I need to do with the cam changeover maps KFNWSE?  Or does the ECU see that in ESKONF that VVT isnt a option and simply follows the first timing map KFZW instead of KFZW2?

Got myself a forge actuator.  Begins opening at 10psi and fully open by 14-15psi.

Whats the best way to scale it in to get the boost path I want?

Why?

Hold more boost across midrange.  They work pretty damn well.   Common upgrade for a bit extra is 10psi or 14psi actuator here in the UK.

I don't know of anybody who runs 100% WGDC in midrange, unless they are running turbos not suited for the application.

IMHO uprated actuators are useless if you have electronic boost control!

How so? Can quite happily hold an maintan most boost profiles with an uprated actuators and electric boost control..

This thread is filled with so much fail at this point :(

But you can maintain same level of boost without uprated actuator if your ECU is properly tuned... And it is much easier to regulate boost if you have stock actuator :D

+1

So when the waste gate is being forced open due to high back pressures and a weak stock actuator how do you maintain boost?

No Ecu tuning can over come a physical barrier!

Then you replace your turbocharger with something suitable for boost levels you require.

100% DC was to see that the actuator was weak.   No different if someone ran a manual boost controller set to 21psi.

The forge one with yellow spring works awesome.   Duty cycle right down into the mid 70% on a stock k04 boost map....just needs tweaking to get it how I want it.

Every k03s that has made over 225bhp over here in the UK is running a uprated actuator.  One of the leading UK VAG tuners got 267bhp out of a k03s and other setups are regulary making 250bhp with uprated actuators and a good flowing setup.  Same with some of the K04 with relentless manifolds and uprated actuators going on cracking 300bhp on a stock turbo.

The OEM actuators springs go so weak over time, they can no longer hold back the pressure flowing across the wastegate so they tend to have high DC% and get blown open at higher RPMs.

Ive put one of the logs of my old AGU k03s setup that ran 250bhp (not on that map variant as we changed load due to a faulty gauge during mapping).

First of all - it is not a diesel so power peak @4500 RPM is not normal - it just shows that turbocharger is too small for your application :D I would like to ask how long will it survive running such software.
9 years ago I had Golf GTI (AUQ engine) which had ~230HP @~5700RPM at the crank with stock K03s running pump fuel. I believe it is highest possible power you can "safely" have from K03s in daily driven car.

Its not a fudged plot.  Ill guarantee you 100% that.  1.8T tuning is still jumping leaps and bounds here in the UK.  All the big power k03s builds over here boost to 21psi peak and thats it.  everything else is down the the hardware and mapping. 

Littco on here will back me up 100% on this as he's working 1st hand with the tuners in the UK that are making record figures from OEM turbo applications.

Still - it is IMHO useless to have peak power at 4500 RPM in petrol engine. Such a car is a "dyno queen" for me - made only to show numbers, not give fun when driven...

Trust me it was fun to drive.  Having power nice in the midrange is where the k03s is alive.  Pointless trying to make it at 6000rpk when turbo rubbing out of puff.

Exactly, and why you get bigger turbos...

Personal preference at end of the day.   Car was epic fun to drive to 120mph (where conditions allowed).

I have a motor thats built for big turbo setup....however only going TSFI k04-064 as the chassis wont be able to cope with the power I want and how I want to drive the car.  Could easily have a blower on there to make 550bhp, however 375bhp will be more suited to the style of driving ive built the car for.

Bigger isnt always better...

Id much rather have a car I can enjoy from 3000rpm to redline around country roads, occasional track day than a car which only has power 5500rpm- redline and can only go in a straight line.


Swings and roundabouts.

I am not saying "insanely large turbo" but something a tad bigger so that it doesn't choke at 5k rpms. If you want fun in low rpms just get turbo diesel :)

Since I don't have to funds to build a second car, let alone have room to store it, I thought about this also and decided on a "quick change" turbo setup. Everything will have v-band clamps and quick change fluid couplings where applicable to minimize downtime between turbo swaps. Exhaust/boost plumbing and turbo intake to fit both turbos. I'm thinking swap time would be under an hour.

Done a couple logs today with the new actuator (yellow spring) and a STOCK KFLDRL.    Got a overshoot of boost at initial spool up I need to try solve.

Any tips?

Stock KFLDRL definitely won't work with that wastegate. Either calibrate it correctly from scratch (a good amount of work) or start with a linear KFLDRL. Youre running WAY too much DC. I mean, you aren't even in the general ball park.

Yup,  far too much.  Linearised KFLDRL tonight and started to work up kfldimx from 10% to get to where I want the boost.

got to 55% in 3rd gear which gives about 17psi, but then in 4th its 20psi.

Thats the log i done.   Not sure where the increase in boost and duty is from though.

Bit more fiddling today :)

Looks like boost is somewhere close.   Still mighty close to the map sensor limit though.