NefMoto

So, in the never ending learning curve of ME7.x, these maps are obviously very important for the car to drive/react happily at both part and full throttle in terms of boost control, etc.  Although I'm not sure yet but it might also have a role in load calculations as well?  I ask only because while I was able to get my car running smoothly by modifying the maps (specifically only the pressure axis) but the car for running the same and more constant boost levels feels slower (could be for the reasons below though)?

That said, I'm curious to know what variables we should be logging to properly dial in these maps (even if it's just to adjust the axis, which for the most part seems to work quite well) when bigger hardware changes are made.

Personal example:

I've always piss poor boost response and stability in my GLI.  The PO had upgraded the turbo to a K04 unit (unsure at this point if it's a real BW or china replica).  Everything else other then the intake silicon and DV is stock on the car and now I've added a 4 BAR FPR.  The car has a lot of miles on it (over 300k km).  I do think there might be some blockage with the stock catalytic as well at this point since I've had issues with the coils in the past as have a cat efficiency for the past little while code so keep that in mind.

I was able to solve almost completely the boost issues by reducing the pressure axis by 15%, here you can see before and after logs...

Before:
(http://i.imgur.com/iZH6ItC.png)

After:
(http://i.imgur.com/DTrhCsz.png)

No matter what I tried to do with the boost control maps, I was never able to solve the oscillation issues under part/full throttle until now so I'd like to see how people generally go about tuning these maps for their setups/situations.

One small note - I might be wrong but what you are referring as feeling the car being slower albeit running same level of boost might be only because it drives smoother so also the boost is built smoother.

That first graph looks like it is full throttle: you should have NO problem tuning the PID for that condition.

Changing the pressure axis just changed the way the PID responds, IMO.

I agree. I don't think this was the right approach here.

I've tried everything I can think of with PID changes and it never, ever turned out as nice as with the simple change to the axis that I made...

Also are you sure that pressure axis is even shared with the PID?  I don't think it is?

Nothing else in the two before/after logs have changed, only wgdc response.

exactly... so if everything remains the same but my only change was the pressure axis for the throttle maps, I'm having a hard time to understand how this is PID related that's all I'm saying.

Of course this is why I started this thread :P

My take on it is that the PID controller is all screwed up.

First run starts from 3k and ramps up boost very quickly overshooting the target and causing wild oscillation which PID controller cannot attenuate due to wrong parameters.
Second run starts from 2k and there is physically no possibility of overshooting there so the ramp up is gentle enough for PID to still work smoothly and not cause oscillations.

Post logs of P, I and D and we will immediately tell you what's wrong :)

It doesn't matter where I start if I put the pressure axis back to stock for the throttle tables the boost will oscillate all over the place until redline.

So how can it be that the PID is screwed up when the PID settings are identical between the two files?

Here is a log that starts right around the same time as the other run (around 2k):

(http://i.imgur.com/POKKXnm.png)

He's saying the overshoot isn't as severe, so the reaction isn't so dramatic. Telling the ecu the gates are a different preload is not the solution here IMHO. You're just allowing it to compensate for an improper PID with target throttle plate angle.

Still doesn't make sense to me.

Throttle plate angle isn't any different in either file.. Both follow my pedal accordingly and TB never closes.

So how can it be that PID is reacting correctly with just that change but I see no difference with the actual TB movement.

We're obviously looking at WOT only here in all of these runs.

When I played with these maps I found the throttle plate position just changed.. Instead of 100 % I got 85% .. Worked as a hack but maf readings and driving was very wrong...

This is exactly what the axis does. Basically if requested boost is below wastegate pressure, rather than allowing the PID to try to control it, it uses the throttle plate angle.

As for it being off, you still had to modify WDKMSN and MSNWDK properly.

Pretty sure I did modify them.. I need to spend more time on this as current methods of controlling boost by direct n75 control is leading to issues where actuators settle down over time and where say 65% dc gives 1.4bar one week, a few hot cycles and softening of a new spring leads to 1.1bar and the need for tweaking.. Starting to like the old load based tuning much more for this exact point...

BTW the two graphs in the first post have drastically different boost ramps.

the PID should be tuned so it can handle both ...

You need a LOT more D in the RPM area where the boost nears reqest in the first graph..

in the second graph, the D is sufficient at that RPM.

I've also seen the N75 itself start to act differently depending on engine bay temps (i think), which makes things even more unpleasant :/

thinking about relocating it to behind the firewall to see if it makes a difference.

the runs were started at different times in the RPM band, and slight differences in PID tables since the log was from a much earlier revision where I was still trying to calm everything by editing the PID tables.

I'll put these two together where the runs started almost at the exact same point, no throttle opening differences at all.  The main difference is that I actually numbed the PID/DRL tables to try to avoid overshoot after not getting anywhere with PID adjustments, so the ramp is longer and boost builds slower in the first graph.  But I'm telling you that if the only thing I change between file logged in the bottom graph is putting the axis back to stock, I will have oscillations throughout the RPM band from start to end.

throttle press. axis stock:
(http://i.imgur.com/POKKXnm.png)

throttle press. axis stock minus 15% across the board:
(http://i.imgur.com/DTrhCsz.png)

The pressure axis is a ratio of desired manifold pressure : ambient pressure.

vpsspls_w is a desired pressure ratio pre-throttle pressure : manifold pressure.

vpsspls_w is used to convert desired manifold pressure pssol_w to desired boost pressure plsol_w.

It is also used to regulate pressure pre and post throttle via the throttle plate. If vpsspls_w is > 0.95, then the throttle plate is forced to open that percentage towards WOT. Anytime  vpsspls_w is 1, then the throttle is forced to open 100%.

It has no other function and no direct influence on the boost PID controller.

Target boost pressure is lower in the second one.

Is it switching back to dynamic mode or remaining in steady state?

We really need to see the PID terms at least.

The PID response looks crazy aggressive to me.

Log WG pre-lin.

something is very wrong with your LDRL perhaps?

Hey nyet, how off are the logs if the PID parameters aren't set up correctly in xPlot?

The PID parameters are only for the CALC functions (if you want to simulate a PID given a set of data to see how it would respond)

As an aside, this "simulation" isn't really complete because it is stateless; it only shows what the pid value would be given a set of inputs... it can't manipulate the DC and predict how the system would respond.

I"ll get some more data soon... in any case modding just the axis made the car found like it's boosting but something else is going on (looks like the car trying to correct the boost by taking away timing).

anyway, I've reverted back to a much tamer file, I'll throw up some logs soon for it.

Still according to Phila logging the vpsspls_w variable is enough to setup the maps/axis properly in these two maps, correct?

It's a pressure ratio. No logging required, just set it based on your wastegate pressure.

The axis variable is vpssplg_w and can be logged itself.

Keep in mind that the pressure ratio, vpssplg_w, is based on desired intake manifold pressure pssol_w, so it's post throttle and will always be less than or equal to desired pressure at the MAP sensor plsol_w which is pre-throttle.

Assuming:

a) the MAF and scaling is reporting true airflow
b) you're running a proper 5120 or not maxing out the pressure variables
c) no other parts of the model are messed up

ps_w should follow pssol_w and rl_w should follow rlsol_w.

To explain it as simply as possible,

The map is used to request a 1:1 ratio pre and post throttle, i.e. full throttle.

Thanks guys... I'll play with it all again once I have some free time :)

Sorry to bring up old topic but I have some weirdness with dwdksut_w and was thinking of updating KFVPDKSD/KFVPDKSE to all 1s. Will that help me to overcome that dwdksut_w "swing" (please look attached log screenshot)? Currently dwdksut_w makes my throttle plate go 100% -> ~80% -> 100% in 1000-1520 rpm zone.

Anyone who has tuned K04-015 or K03s what have you set in these maps (KFVPDKSD/E)? Screenshots would be most welcome :) I kinda am thinking of setting everything to 1s (as it is done in 2.7T A6) or just copy RS4 values here as a starting point...