http://www.e90post.com/forums/showthread.php?t=416787
Hi guys,
Just another thing we've been working on during the past few months and now ready to release to the beta list.
As most of you know, the Procede autotunes both fuel, timing and boost target curves. Which makes things very easy for the user. Just plug it in and drive.
However, there was one thing that the Procede didn't self-adjust: Global Boost Control Gain (formerly called 'Future Use 2' in the user adjustable menu). This value is essentially a scalar that is applied to the boost control Start Duty Cycle table which represents the initial/start DC% applied at any given RPM/Throttle combination. This can be as low as 20% (low rpm, light throttle) or as high as 70% (high rpm, full throttle). By "start/initial" DC%, I mean that this is the DC% value BEFORE any additional closed-loop PID corrections are applied. This PID system jumps into play when actual boost (dictated by the Start Duty Cycle table) is higher or lower than the Target boost pressure. So in a perfect world, where the Start DC table is perfectly mapped, the PID system doesn't have to do much or anything at all. But we've all learned by now the world aint so perfect due to variances in conditions, wastegate spring stiffness, turbo condition, exhaust back-pressure, intake restriction, etc,.
So you may ask.... "Who cares if the Start DC table is perfectly dialed in for my car? Isn't that what the PID system is for? To fix any boost errors?"
Not quite. The PID system is, of course, supposed to do that. But like all reactive systems, it takes a finite amount of time to fix the error. If the error is positive (actual boost is greater than boost target), you will experience an overboost condition/spike which makes the more knock prone, less stable, jumpy and more suseptable to throttle closures. If the error is negative (actual boost is less than the boos target), you will experience an underboost condition which makes the car feel like it is gradually "ramping" up to the desired torque value instead of following the throttle instantly.
This unwelcome positive or negative error exists until the PID can respond and reactively changing the wasgtegate duty cycle. But since we are dealing with pneumatics, fluid dynamics, etc,. there is a time delay before boost actually response to the change in wastegate DC. This is because the reactive change in DC signal first changes the open/closing rate of the boost control solenoid(s)
-which then bleed off/increase the amount of vacuum in the wastegate signal line
-which then then pulls closed or pushes open the wastegate actuator
-which then increases or decreases the amount of exhaust flow
-which goes to the turbo compressor which has to then slow down or speed up
-which then pushes more or less air, through the intercooler, past the throttle body and into the intake manifold
Phew. So depending on the size boost error and the effectiveness of the PID system, the "fix" can take anywhere from a 1/10th of a second to more than a second. Which for an engine (or driver) seems like a lifetime where you are not getting the torque output you want or worse, getting too much which can cause knock, unwanted throttle closure or an unstable and sustained oscillation condition (ongoing up and down and up and down boost patterns).
It is these boost control (and resultant throttle control) anomalies which play a huge part in making a car feel "smooth" or "on/off" or "inconsistent" or "responsive with the throttle".
Historically, with our v4 maps, we've always played it safe and set the default Global Wastegate DC Gain low enough (45%) so that the Start Boost Control DC table resulted in a slight underboost situation in most cars and an overboost situation in NO CARS. Those who have datalogged Debug Word 4 (boost learning aka Integral correction for those who understand PID systems) know this because they have logged positive corrections. And positive corrections is the response to a negative error/underboost condition.
However, in some cars with older softer-sprung wastegates (BMW has released many different n54 wastegate designs/fixes for our n54!), this intended slight initial underboost condition became a moderate initial underboost condition. For example, in my early production 2008 car, if I run the default Global Wastegate DC value of 45%, my debug word 4 values hit 10-15% at WOT. Which means that the PID system needs to dole out an additional 10-15% of wastegate DC in order for my boost target to be reached. This would look fine on a dyno since that only takes 1-2 seconds to happen after going full throttle, but on the road it would be a different story. Upon sudden throttle opening, my boost wouldn't jump to the desired boost target. Instead, it would jump to 2-3psi shy of it and then gradually ramp up from there. Which made the car slower and less responsive to the throttle. So to fix this, I had to conduct datalogs and see which user adjustable value resulted in debug word 4 values of 5-10%. So I ended up running a user adjustable value of 50-55%. I'm sure the same went for many others with older cars. Especially those at altitude or in super hot conditions where boost is naturally slower to build.
So now, we've put the last feather in the Procede Autotuning hat, so to speak. By having the Procede keep a track of ongoing PID corrections and automatically adjust the Start DC table scaling in order to keep the reactive PID corrects minimal throughout the life of the car and regardless of operating conditions. So every time you put your foot to the floor and run on boost, it will monitor these corrections and apply a constantly updating and PERSISTENT (non-resetting) correction to the BC Boost Control Gain setting. This correction is viewable (it's debug word channel) AND unique to every car.
And because the tune is now capable of customizing itself to the hardware in the car (and not just fuel/conditions), we now target a more aggressive gain value. Which means that boost will rise as quickly as you put your foot down. This also means that boost will be even more linear with the throttle. Which not only makes the Procede the "smoothest" tune on the market. But, all things equal, also the fastest.
I'll be emailing this firmware/map upgrade to the beta list tonight. Just in time for the weekend! However, only the 6MT map are currently available. The 6AT map will have to wait until this upcoming week. Big changes with that too. But that's for another thread
Cheers,
Shiv