Hessitation in Late Model Subarus

[GARY]

Hello,  

A lot of people have had a problem with a stuttering acceleration in
Subarus.  I offered up an hypothesis along with a fix for the problem. 
My hypothesis was first rejected by Subaru of America, then later
supported in a Technical Service Bulletin.  

Rather than go into detail here, I'll supply the URL for my website on
the matter. When you go there, scroll down for a lot of links.
http://home.earthlink.net/~hobiegary/hesitation.htm 

Well even though Subaru claims to have a fix for the problem via a new
ECU, I have yet to hear of anyone who could jump through all the
necessary hoops to get a new ECU *and* find out that the new one
actually cures the problem.  So I am not holding my breath on that note.

What I am doing however, is planning on an improvement on my primary
fix.  The ECU is a OBD2 type with EPROM.  It can be "reset," to loose
stored trouble logs, by disconnecting the battery for about a half
hour.  I need to electronically dampen or trim or filter (or whatever it
is called!) the signal from the knock sensor to the ECU.

What is happening is that the ECU is interpreting non detonation noises
as detonation and reacting accordingly.  This leads to piss-poor
performance to say the least.  I have gone the mechanical route,
buffering the signal by padding the mechanical interface between the
sensor and the engine block.

What I would like to do is simply reduce the sensitivity of the sensor. 
That is what I did mechanically and it seemed to work pretty good.  The
problem though is that it is a little iffy and esoteric and problematic
and lacking consistency.  It is not solid state enough to suit me.  I'd
rather dampen the signal electrically.

I tried hooking up a variable capacitor in parallel with it.  I thought
that this would let any voltage spikes bleed (partially) to ground and
lessen the signal strength arriving at the ECU.  I had trouble with this
method because as soon as I turned the capacitor a certain amount, I'd
get a trouble code and would have to shut down and restart.  I'll guess
that the ECU sends a monitor current through the sensor and detects when
it is not at the proper impedance.  It didn't seem to help either.  I am
not convinced that I was doing the wrong thing either.  I just could not
get any positive results.

I also wonder if I rather should be using a variable inductor in series
with the sensor.  Wouldn't this choke a signal without greatly affecting
monitor current?  I guess it would depend if the monitor is looking at
impedance with a/c or looking at resistance with d/c.  

Comments?  Any other ways I should consider reducing the amplitude of
this signal?  What about the ECU, is there anything I can do there to
help the situation?

I thank you for your attention Fellows and thank you in advance for any
suggestions.

-- 
GARY       mailto:hobiegary at earthlink.net