NefMoto

I have the PID pretty well dialed in as far as on road impressions, but wanted a couple of second/third opinions. The hardware is 60lb EV14's and frankenturbos on an 01 Allroad. Stock wastegate actuators. I have some ignition correction, but I am only running a single meth nozzle. Waiting on my two matched nozzles to get here before I finish timing.

Your wg duty is a mess.

Why isn't it 100% during spool?

You are leaving a TON of torque on the table.

Why did you disable P?

Also, your I-Max is a mess.

Nyet, Can you explain what you mean by "Disable P"?

My whole reason for making changes to the PID were that the frankens use a lot of spring preload and the gate openings seem to be undersized. I was getting a lot of overshoot initially.

My initial response was to drop the P (LDRQ0S to 1.8 from 2.0) and I (LDRQ1DY 20% drop above 2750rpm) terms to slow boost build and lower my IMX to keep the spike at bay. Those changes and increasing Q2 made a world of difference. I also followed what has been posted to set KFLDRL to match the DC response. Not sure I used the spreadsheet PRJ posted correctly though.

It has worked. Boost may build a little slower, but it only does what I request. Killed the overshoot.

I do see that my WGDC is crazy on spool, but a little more stable once boost is built.

My approach may not be as aggressive as you usually set cars up. Its my daily and has a baby seat in the back.
I really appreciate your help. I have read a lot on different PID approaches for ME7, but there are limited examples out there to compare notes with. What would you recommend?

I like strong bottom end TQ( I miss the hit of my K03's a little), but don't want to get too aggressive and make the car an on/off switch or stress the rods.   

Either way, Thanks for any and all comments. No better way to learn than with feedback.

http://youtu.be/SefKQb9y_B4

Excellent Video. I am 2/3 of the way through it and have already made some revisions I will try on my morning commute. It also stresses the need for proper DRL tuning. I think I will go back through that. Thanks for posting that link.

Your P contribution is zero in the log you posted.

You should be at 100% DC during spool, and use Q2 and an appropriate i-limit (preferably a relatively smooth one) to prevent spikes.

The way to limit torque is NOT by screwing up the PID control, but to have less req torque... and no, I don't see a good reason to do that either.

no problem! it helped make alot of sense out of confusion for me.   tune P, D then I.. now I understand why. 8)

One thing.. With 100dc you divert boost away , thus creating a "boost leak" as the boost is drawn away from the charge system... 0dc obviously flows boost to actuator but allows waste gate to open.. Have you tried a valve on the charge side of the n75 set to crack at 15psi (or similar) to limit boost loss on spool?

Interesting point... never thought about this. Is that leak really significant? I've never noticed that it was.

Its not a leak to atmosphere. Just back into the spider hose assembly.

You're describing an MBC in parallel.

The amount of air routed from the compressor housing into the spider is inconsequential IMO.

I have pretty much the same set-up as you same set-up (A6, not AR, though) and made things much smoother by decreasing WG spring tension to 7 or 8 lbs opening pressure and linearized KFPED in the region where boost comes on.  Not having the throttle snap full-open at 40% pedal (or whatever) was key.  I'm sure there are other ways to do this also.

I set my WGDC to 0% until close to wastegate pressure for the exact reason you said. I'm sure it doesn't matter that much, just a matter of personal preference.

Problem with that approach is it reduces the spring's ability to keep the gate sealed at high boost levels. Lots of pressure build up in the small turbine housings.

I'm showing the same 2300 mbar/315 g/s flow as you.   My numbers are very similar to yours except that mine peak sooner.

The RS4 WG is set at about 8 psig IIRC.  But as you say, you're happy with it, so don't take this as trying to convince you one way or the other...

Huh? 315g/sec? The log I posted peaked at 375 g/sec... typo? or talking about a specific rpm?

Do you have a log to post?

I was told by Doug at frankenturbo that to run 25psi+ the gate preload must be kept high because the turbine housing internal pressure would start to push the flap open. I just checked to see if the gates moved in sync and at the same ~15psi cracking pressure before installing. Base WG pressure with the N75 unplugged is roughly 18psi. I left them at the high cracking pressure incase I wanted to run more than map limit at some point. Not something easily adjusted in car after all. In hindsight, setting the gates to less preload would have made my boost control much easier. No big deal. Its working now. 

As for drivability, it is excellent. I scaled the KFVPD axis to match the base gate pressure and its very well mannered at part throttle. No surging or hesitation. I found that info in the big turbo tuning post where Julex had said it worked well for him.

I am not finished with timing by a long shot, but my boost curve is where I want it. Will see a lot better torque curve once the second meth nozzle is in and I can finalize timing. Right now I am running the same settings I ran on 93 pump with my old K03's and a single small nozzle mounted in the driver's bi-pipe.
 

IMO, this is what you should shoot for (I still have a touch of overshoot here).

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

I agree. That DC looks great. Little to no adjustment from the PID.

 I need to redo the DRL map. I think its just starting with the wrong base wgdc and the having to adjust alot to get it right. Second meth nozzle here and i have a free weekend ahead of me to get it setup.

Why is everyone complaining about the boost curve though? I like the current profile. Its a personal prefrence.

Again, i'd rather keep pressure off the wastegate during spool. maybe its just me.

I agree when the spring preload is very low, but in my case with 14psi cracking 18psi control pressure, its not affecting operation.

I understand that, but that isn't how PIDs are supposed to work, even one with parameter scheduling.

can you post your P, I, D tables/values.

Your D is really bouncing like crazy initially below 4k when in dynamic mode, most likely due to I not being adequate.
Then you go into static with "P" active and it is a calm period but due to too small of "P" quotient (or maybe KFLDIMX being too small?, or your KFLDRL being all screwed up - most likely this) you deviate below the dynamic regulation threshold again (UMDYLDR) and the ECU switches in dynamic load regulation for PID controller.

I spent too much time tuning this stuff on my car, as you can tell...

edit: I am now almost 100% sure you KFLDRL needs major overhaul... it is stock which with 18psi wastegates will never work smooth. You need to follow "KFLDRL calibration" thread and do some runs with fixed WGDC to figure out how different levels of fixed WGDC result in boost profile which you can then linearize.

My DRL is not stock, but needs a major overhaul. Sunday is log and flash day. I am going to redo my DRL. I used values I logged from a friend's GT2552 car. Figured that would get me a lot closer than the factory stuff. I know that's the wrong way to do it, but this is pretty much a base file.

I agree, but there is a limit.  Heavier the spring less you should use to minimize air loss.  For instance I use a 21# tial gate, its true spring flow rate is 23#.  At 3.5% duty cycle I make 30 Psi, at 15% duty cycle I make 36, at 24% I make 42.  25% becomes my max dc for spool region.  0% until spool is possible.  Otherwise its just a line that flows like a boost leak or even vac leak.