NefMoto

Hi,

I've a little overboost problem. I tried to compensate it with KFLDIMX but the reaction is not as expected.
I've attached the starter KFLDIMX and the log for that(1st 2 picture)
After that I modified the KFLDIMX, subtract -3 at 3000rpm(3rd picture)
And I still have the overboost(not as much previously) and after 4500rpm it looks a shit.
What I can see from the log that fuel injector duty cycle is over 100%. Maybe my injectors are too small for that config?

Thanks for the help

PIDs dont like huge cliffs in the set point.

Why on earth is your boost request dropping like a stone, then coming up again, then dropping again.

What do you THINK a PID is going to do in response?

These set points created by the controller. I don't understand why it happened. The requested boost should be the same as it is on the first log. I always make the log from 1800rpm until 6500rpm WOT.

Thats because you didn't bother to log just about any of the important requested load variables.

They aren't "created by the controller", they're a direct result of whatever you put in the corresponding maps, all of which are clearly documented in the tuning page.

Sigh. I don't know what to say.

Set up a boost request that isn't completely retarded and you wont have injector issues either.

Please tell me why is this boost request retarded?

Between the 2 runs only the KFLIDMX table was changed. I haven't modified the load maps. I'm loging rl_w and rlmax_w all the time.

I found something really strange. Every run when the requested boost was a shit the lambda value was also strange.
The attached runs was after reprogramming the ECU and first runs. I've compared the logs to a "good" one where the requested lambda and actual lambda is similar and the injector timing curve is same.

Why only those two?

And as far as fueling goes, did you still not learn anything about bts? didn't we walk through ALL of it already? You can't get 11 AFR at those boost levels.

You're probably seeing slow intervention which is why your req boost is falling off a cliff.

Ok so a k03s with 317 injector cannot handle the my boost request.

post your  and original bin.
I think your load values are wrong, you are experiencing load restriction from ECU

Like i did with wrong tuned load values

Hi,

I've uploaded them.

Please log the rest of the load request path.

I'll do it just I need the disable some other, because I can log only 36 bytes. I read that disconnect the instruments cluster but I need that for log.

Ugh, sorry about that, that makes things significantly more difficult :(

I've attached what could work. I would like to log everything but my communication bus does not like too much data or I don't know what is the real problem.

Your KFMIRL and KFMIOP values are way off everything.
Can you log these vaules at the same time?
rl_w
rlmax
miszul
mrfa
mifa
misol
rlsol
and off course RPM and throttle plate angle

I've done 4 logs

i don't see intervention now, only that ECU is chasing the boost

i am seeing rlsol drop off from rlmax, looks like slow path intervention.

need to see mrfa and mimax to be sure

Good point,
At 6000 RPM and 90% load request He got low values in KFMIRL map

Need to get little higher values in this region,and in this whole row(bigger than LDRXN)

Yes currently I've the overboost. Yeah my turbo is not so fast/strong in lower rpm. What can I do with overboost problem? If I change the kfldimx I will get the intervention.

Is this a problem? I'm asking because the requested boost pressure is good at 6000 rpm. I don't need more.

Change IMX then do logs please, I want to see how that IMX have influence on intervation

Yes you get enought boost, but that is bad tuned KFMIRL/KFMIOP that is not the best solution. You can expeeience problems in some condition

It isn't a problem as long as the requested boost doesn't drop off a cliff or do a crazy dance. It needs to be nice and easy, no big changes.

Currently I have this log. The strange thing is the lamda value for me when I checked the log.
http://nefariousmotorsports.com/forum/index.php?topic=16562.msg128041#msg128041 (http://nefariousmotorsports.com/forum/index.php?topic=16562.msg128041#msg128041)

Ok, so rlmax should follow the rlsol?

i don't have these variables to confirm, but everething point that there is intervetion. But without these variables
rl_w
rlmax
miszul
mrfa
mifa
misol
rlsol

I can't tell you what change in maps need to be done

Ok I'll do the modification maybe tomorrow afternoon and run a test. I will change only KFLDIMX. Do you have a proposal where should I change it to prevent the overboost and what should be the values? I've uploaded the current KFLDIMX table

Thanks for the help you a nyet too.

Use values of IMX from this file i dont see here overboosting
But what i see is bad tuned KFMIRL

Your LDRXN at 5250 RPM is 175 % load
--> 175% load and 5250 RPM tru KFMIOP in your map is about 90% tourqe request
--> 90% tourqe at 5200 RPM tru KFMIRL in your map is around 160-165%

Soo at first you are requesting 175% load, but tru tourqe maps ecu calculated to request 165% load. And there all shit start to hapen, it has nothing to do with your IMX

So I should change LDRXN to 165 to limit rlmax?

Sorry but can't anwer you precise without log with variables I tell you to log
Also log ldrlts

Forget about this calculation, without mifa and misol it can't be calculated it this way

I've logged all of the variables what you mentioned.

Ahhhh....
Step 1 Change IMX to values you already have in you map when you don't have overboost
Step 2 Flash you file to car
Step 3 Take logs
Step 4 post your logs and bin here

Ok that was my fault. I thought that you need these variables to tell the LDRXN value. I will do a log with the modified IMX table.

HI,

I cannot reproduce the behaviour. Currently I've this logs with the bin.

I don't think your IRL is requesting enough load, rlsol is dropping off below rlmax

Why should I change IRL? What about change rlmax to lower? The requested boost curve is what I need.

Sure, if you like.

its better to implement taper with rlmax than IRL because IRL can get bumpy.

Thanks, so I should change LDRXN? If I know well this is the rlmax variable.

Yep.

Look at page before, i show you what values you need to change

Thanks. But I don't want to more load at 6000rpm, that's why I would like to change rlmax(LDRXN) a little bit down.

If you change rlmax, your misol will drop down. And that will result even worse intervention.
You will need to put higher values in KFMIRL. Than you can lover down LDRXN

Depends on his IOP which i haven't looked at, but yes, you are correct that could well be the result.

http://nefariousmotorsports.com/forum/index.php?topic=16562.msg128123#msg128123

LOL ya, there is zero reason to taper IRL.

Thank you guys. I will change kfmirl and after that lower the ldrxn. I hope that the boost request wont change.

The calculation is good with the new logs?

I've made the modification. from 2500 rpm offset +10 at the 85% line. I've checked with linear interpolation and kfmirl is always higher than ldrxn through kfmiop.

After change to 386cc injectors and clean the n75 valve I've got this log. There is no overboost(finally) and the values are not bad.

Hi!

Please help me. I've modified the kfldimx at lower section 2250 and 2500 rpm. I've made a log and there is an intervention at 3000 rpm. The actual boost is under the required boost. Which table should I modify?

Thank you

Thats D (Q2)

might want to review your PID basics ;)

Thanks. First I need Q2 in my xdf because currently I don't have.
What will happen if I change ldrxn at 3000rpm?  Set it to 180.

Leave ldrxn alone. It's fine. Your IMX is also a bit low, but you will trade undershoot with overshoot.

Ok thanks. Can you tell me why I have intervention? Is it connected to load? I've read again PID control theory and with D I can compensate the overshoots or accelerate the system. What I saw at logs that the rlmax is lower than the actual load.

You don't have intervention, but you have knock. And maybe you MAF is not scaled rihgt
What turbo do you have?

K03s. That was my previous log http://nefariousmotorsports.com/forum/index.php?topic=16562.msg129829#msg129829 the change betwwen the 2 softwares is that the KFLDIMX values are higher at 4 points. The reaction is faster ~400rpm but the new software open the wg at 3000rpm... That is because of knock?

Sorry I am not expert, but you are asking 12 deg ingition at spool. That valuese schould be lower.  That is only a hint
But it has nothing to do with your PID setings, and WGDC drift. And i don't know what to do there
That is a preaty good K03s,it can archive 185 gs

Stop calling it intervention.

Ok I will try lower values at ignition. Thanks a lot for help to understand it.
Under 3000rpm it is slow a little bit, but I'm satisfied with it. I saw 190gs from k03s.

Ok sorry.

Changed the ignition tables. I've 2 logs with the second I dont have retard but the reaction is similar.
What could be the problem?
Thanks

What does ignition have to do with boost?

I don't understand your question.

Also, you probably don't want negative timing.

BlackT said that the previous log there were retards at 3000rpm where was the boost issue and 12deg was little bit too much at this section.
You mentioned the D part. I have not changed that this is the original value.

I sad that is yust a hint, and I have said it has nothing to do with WGDC and PID control

Thanks for the hint. Now there is no retard but something controls WGDC at 3000rpm.

Yes, that is the PID. Your IMX is too low and Q2 is reacting to the slope of lde.

Question: what is de/dt (where e is process error aka lde)?

If you can't answer that, we're in trouble.

If I know well that is how fast e is changing in time

Exactly. So if the slope of lde is very negative, you'll see a large cut in WGDC.

What is the slope of lde when you see WGDC drop?

lde=("actual absolute pressure" - "requested pressure")
From the log the starting point of cut at 5.3s. The requested boost is constant(2300).
lde at 5.1s 2139-2300=-161
lde at 5.2s 2194-2300=-106
lde at 5.3s 2221-2300=-79

 I've changed kfldimx a little. Lets see what will happen

It seems it helped to the cut. The strange thing is that the maf goes a little bit lower.
The first 2 logs is with the yesterday software. The last 2 is with the modifed kfldimx ans I set kfldrl from 100 to 95 and LDRQ0s from 2.1 to 2.
I've add p i d parameters to the log. Please comment if you find anything wrong.

Thanks

That IMX still dips too low.

Can you tell me which section should I raise? Thanks

You tell me.

I should raie at 2500 and 2750 to near 95 and raise a little at 3000

No. You have way too many things going on in IMX. Just find a good start/end value. Don't have huge jumps in it. Don't have lots of random high areas and low areas and drastic changes. Try to keep it mostly flat. PIDs are hard enough to tune without you giving it a bunch of random crap that moves around a lot, especially in non-linear weird parameter scheduling stuff. Keep it simple.

something like this.

but really if you need a huge imx slope to make sure it stays high enough, that means your DRL is bad.

Ideally IMX should stay flat, and let DRL deal with increasing DC if you need more DC at higher rpms.

Also you are maxed out up top. Might want to pull some req boost

Now I put together wgdc and imx. From that picture what I can see is that from 5250 I need to close the wg so it needs 95% from that section. So it could be a good end point. Can I use I result to define imx?

Yes and no. IMX should really be more or less flat, and you should use DRL to compensate for needing more wg for a given boost. That is its purpose. Tuning this stuff blindly w/o really understanding at least the basics is going to be problematic. And if you are hitting 95% ANYWHERE (other than during spool) that means your req is likely too high.

Thanks for the help. I checked ME7-functionsrahrmen to understand the pid but it is not so easy(or I don't understand it). What is the puposeof imx? The output for N75 is comes from kfldrl.

Understanding PIDs is a general engineering/control topic. You won't find a PID explanation in the FR. IMX is there to prevent integrator windup, and generally in ME7 it can be used to provide feedforward control. Much of this is covered in the Tuning wiki and in lots of other places here.

Long story short: your WGDC will (mostly) follow IMX post processed by DRL.


I sense you don't really understand what DRL does either :(

In general I understand PID controllers. But I just used P I D parameters to tune the controller. Here there are more tables. I need to read more the FR.

The reason IMX is so important in ME7 is that in the steady state, P result is near zero and D is scheduled out with B_lddy. This means that I dominates.

On a WOT pull, I is maxed out due to integrator windup (accumulated error) so it is simply following IMX.

Also in the steady state, you would expect the output of the PID to be flat for a given flat boost set point. However, DC isn't linear with output boost, so the output of the PID is post processed by DRL.

Again, if you truly understand PIDs, you should already know the critical points here

1) integrator windup effects when there is long term error
2) linearizing the control output to compensate for varying process output gain.

I've read this topic http://nefariousmotorsports.com/forum/index.php?topic=11701.30 it was very useful.
Does my boost request is ok? I know that around 6000rpm I should lower. Could be a problem if I running 95% wgdc at the end?

You'll eventually send the I-trims to max and in extreme cases get a pos deviation code.

I found an another good topic with the tool http://nefariousmotorsports.com/forum/index.php?topic=12352.0title=

Yes, I was directly referring to this when I mentioned both feed forward and the purpose of the DRL table above. It is also linked in the tuning page (i think? If not i should add it).

I've read tuning wiki a several times and every time I found new useful sections. Thank you to update it all the time.

This kfldimx has big leak but it is working very well. The turbo(k03s) is same as mine, and also 386cc injectors.

This is mine log and dyno picture

Hi,

How can I handle the spool of the turbo? D is not enough to do that. So when the system reaches the target ->goes to steady state and it will run from IMAX. SO If I change Imax it wont be enought to stay at steady state. Should I change KFLDRL at the critical sections between 3000 and 4000?

Thanks

What happens if you try more Q2?

I don't know. Currently 25% added to the default values in 100 and 200mbar lde. I don't want to go into instable. Currently I think about change KFLDRL at 80% which is 1450mbar at the current configuration. I won't request that pressure so I know that if the input value is around 80 percent-> the turbo is in spooling and I can open wg a little bit more.

In my experience, pretty large changes in Q2 are required to get it to do anything, especially in the spool/steady state transition region. Also keep in mind that b_lddy cuts Q2 out pretty aggressively as lde approaches 0.

Yes the I found the cut out. After reach the target Q2 is zero immediately. I will check more Q2 at 3000-3500 rpm where is always the overshoot in WOT. At the dyno day that was a car with me7.5 ECU and there the required boost was maximum(2555mbar) in WOT. I don't know how does the pid working at his situation, but I think they just programmed a curve with imx and drl tables and depends on the world sometimes it generates more boost sometime a little bit lower.

Hi,

I've added more Q2 to PID. I've still overboost...  Should I add more q2? Or maybe my n75 valve is not working properly?
The first log contains different starting rpms

Other question. How to set UMDYLDR? I read at the wiki that set it to 5% of the maximum desired value. SO If  I would like 1370mbar to max set it to ~ 68?

Add more Q2. Also log I result.

How much? 10% at 3000-4000rpm?

more, 50%

You want to see a dip in WGDC just before lde hits UMDLYDR

To do that, it will have to overwhelm the fact that P is still pretty high, since max D (well, most negative) will be while P is still influenced by lde>UMDLYDR

Thanks. Now I've added 80% to the original(stock) values.
Can you tell me how does B_lddy work? At my software UMDYLDR is 100mbar. Sometime from the log I can see that the D part is working but lde is lower than 100mbar.

I don't see d nor lde nor b_lddy in that picture.

The cursor is set to the point where lde is lower 100 and d part is 9. You can see the data at the left side.

lde has to drop below 10 or above 100 to trigger lddy off/on. See also hysteresis. i don't see what is actually going on with lde so i can't tell.

Either way, D has to overwhelm P and I such that you see a drop in dc well below the I limit and also *well before* lde hits 0.

Orange circle... why is your imax jumping here?

Added yellow line should be your dc.

it should come down well before lde=0 and should settle on imx soon after lde=0

It's jumping because my boost request was not stable. It changed from 2350mbar to 2330mbar.
Yes I would like to see that curve on dc.

got it, didn't see that.

A big big thanks for the advices. I added more Q2 and the curve is nice for me.
The last log for the different starting rpms.

fantastic, glad you are getting what you want out of the PID finally.

Really thanks for the patience and for the help. I've learnt a lot about how does the boost pid work. At higher rpm's it has a little overboost but I think it is in the acceptable range.

I've just checked the logs and foung something strange for me. How does it possible that rlmax is lower than the rl_w?

They're not directly related and they're never really cross checked.

Ok so my log seems good and normal.

Yep.