[s-cars] Re: Motronics MAP - some fuzzy logic included

[Robert Pastore]

Scott:

Generally agree with all you've said WRT boost controllers; several years
ago I invested in the Apexi AVC-R, and wound up removing it after I learned
how to write a boost map for the motronic.  

But I disagree that the fuzzy logic (FL) is what is putting your engine at
risk. 

As I understand it, the FL is looking at the delta between the actual boost
experienced (as per the MAP sensor), and the desired boost (per the
programmed boost profile), and then adjusting the controller's output to the
actuator to minimize that delta, whether it be changing the duty cycle of a
FV, or a stepper motor.   If you want more on how it works, I think Julian
Edgar from Autospeed did a great job of explaining it in the series of
articles on the intercooler/water injection controller they built.

The integration and interaction of boost control with the ignition and fuel
control of motronic provide a much more sophisticated management system than
an aftermarket controller can provide. Just like you said, it is a function
of the number of inputs, data points on the maps, and the tools the
controller can control as outputs.   The boost controller can only vary
stepper motor or FV duty cycle. The motronic can reduce ignition timing,
dump extra fuel to cool the cylinder, and shut power to the WGFV to reduce
boost, all as safety measures to protect the engine.

I see the primary shortcoming of the boost control system on the AAN to be
bottom chamber only boost control (the WG is  slow to react and leaks boost
into the exhaust via the WG valve-to-guide clearance),  as opposed to the
earlier Audi (and most aftermarket boost controller) top chamber control
systems.  I think the change was probably due to the regularity with which
WG diaphragms tore under the earlier system, because the pressure delta
across the diaphragm is greater, thus faster acting and more violent.  The
change probably solved a warranty/durability issue, but was IMO an
engineering compromise that resulted in less actual boost control, [
strictly my theory] and the creation of "overboost" as a great marketing
gimmick to describe the period of time it take the inferior actuator system
to reel in a rapidly spooling turbocharger.  

The other problem is the speed of the stock WGFV.   It is slow and soft.
It is interesting to see that ECS has taken this pig of a valve ( or
derivative thereof), and marketed it as an upgrade to the b4 s4 crew.   It
seems that the B4 s4 has a faster active WGFV, so by putting in a slower
one, boost levels climb and you go faster.      

rgds,
Bob
 

-----Original Message-----
From: QSHIPQ at aol.com [mailto:QSHIPQ at aol.com]
Sent: Thursday, January 22, 2004 1:38 PM
To: CaptMagu at aol.com; FvAMI at aol.com; pizzoman at yahoo.com
Cc: s-car-list at audifans.com
Subject: [s-cars] Re: Motronics MAP - some fuzzy logic included


Let's explore this a bit Hap.  Computers can't really learn, these 
controllers you speak of are referred to as fuzzy logic boost controllers.
Fuzzy logic 
as applied to a boost controller will only be as good as the S car Motronic 
when it can accept as many variables in the boost profile.  Currently we
have 
MAP, engine temp, dual knock, throttle position, altitude and CAT.  IMU,
none of 
the boost controllers on the market can account for these variables.  They 
can *assume* their is something wrong within the parameters of the fuzzy
logic 
program, but it can't *identify* it if it's not an input.  Then what
happens?

Well it can't code anything or flash you a ck engine light to advise you
that 
the reason for fuzzy logic interrupt to a boost profile is for a specific 
reason.  So, by definition fuzzy logic can't *learn* anything.  It can only 
sense, via it's specific and only inputs, that it can't give you what you
have told 
it to.  So, one could argue that a fuzzy logic could be better than a non 
fuzzy logic external boost controller (see below), but I'm not at all
convinced 
that this is even close to a Motronic boost controller.  

The best part of ANY external boost controller could be argued to be the 
stepper motor function.  It's quick and it's accurate and it's failure rate
is 
almost nil.  That doesn't at all equate to the controller itself being
better, 
only the mechanics of the valve actuator of boost being better.  

Further, I'd ask the question in "fuzzy" logic, how does the controller
learn 
that you've fixed or changed some parameter to allow 'full' boost again.  
Does it continue to apply your desires, then accept a lower value?  That
could be 
scarey.  When you climb into the mountains (by definition a boost reduction 
in the motronic >10,000ft = no WGFV function), what's fuzzy logic think of 
that?  I don't believe it thinks at all.  I personally believe fuzzy logic
boost 
controller are looking ONLY for turbo surge line as the indication of
profile 
adaptation.  Add in a limited number of inputs, it's primary adaptation is 
surge line based.  Even then, by my thinking, it will fail at it more times
than 
it will succeed.  Why?  Cuz FL have to constantly learn.

I argue that if indeed this is true, then I'd rather have a manual boost 
controller that let's me do the fuzzy logic (fair and reasonable knobben
dialer 
argument).  Cuz my argument is that a FL boost controller is constantly
putting 
your engine into the danger zone, by definition, it has to to apply FL to
it's 
adaptation.  Your hi and low settings are based on your estimation, not
based 
on the actual engine function or the parameters motronic is conservatively 
monitoring.  

Point of reference 012204, I make the claim that FL boost controllers are 
still second to Motronic in terms of input parameters and safety in
attaining 
boost levels.  My suggestion is to take some of this seemingly unlimited
budget 
dollars and getting yourself a good motronic programmer to really explore 
options within the motronic box.  Bosch made a darn good integrated boost 
controller, it needs a little tweeking to make it great.  IMO, it's within
the 
capabilities of motronic, and hence, is money better spent.

Audisport never put boost control in the hands of the driver.  EVER.  I 
believe for very good reason.  The driver is in charge of driving the car,
his crew 
is in charge of making it faster.  T

My .02

Scott Justusson


In a message dated 1/21/2004 11:32:14 PM Central Standard Time, 
CaptMagu at aol.com writes:
Feico

I had the HKS EVC IV. It was a great controller and had what is known as the

industry standard in stepper motors. It was a snap to program(meaning even I

could figure it out) but could not program individual gears. I now run the 
Apexi AVC Type R that is fully programable. It learns the gear ratios of
your 
transmission and you can program individual boost curves for each gear. The 
Apexi 
does require a software engineer to program. I am able to limit 1st gear 
boost 
to 20 psi and 2nd gear to 24 psi. I can also switch between low and high 
boost and do a blanket increase on all curves. 

If my memory serves me the Apexi does tap into the throttle position signal 
going to the ECU. Both of these controllers apply boost pressure to the top 
of 
the wastegate to help with creep(no Pizzo I'm not talking about you or any 
relative of yours). Somewhere I've got the manual for the Apexi and will 
confirm 
the inputs//// hold it I just checked and it does in fact have a throttle 
position input.
_______________________________________________
S-CAR-List mailing list
S-CAR-List at audifans.com
http://www.audifans.com/mailman/listinfo/s-car-list