Trust STFT's or WB for VE?
Thread Starter
Joined: Jan 2002
Posts: 4,072
Likes: 76
From: Casselberry FL
Trust STFT's or WB for VE? I'm trying to go back to closed loop. I installed new ACDelco O2's and a new LM1 sensor also. I'm getting conflicting results in the HPT histogram. How do I isolate which to trust, and why the other sensor is not reporting similar fueling?
I can't bring myself to trust the histogram, there is
no differentiating between the "sensible" (steady
state) and the "garbage" (transient throttle) and a
street log is highly contaminated by the latter. I've
been meaning to ask them if there's a way to filter
for the same "steady MAP" criterion that is used to
switch between MAF mode and dynamic airflow but
haven't gotten to it. I bet the datum exists but don't
see it, or delta-CylAir, as selectable PIDs meaning a
custom one that spans frames, would be needed.
Need a sense of history.
Anyway, lies, damn lies and statistics. If you see a
large (like several AFR units) difference between histo
"+" and "-" views (min and max) then your data has
too much noise to trust.
So I kind of stick with doing it the hard way; take
steady-throttle pulls, grab the snippet you really
believe in, and work the table by hand with the
real data, then feather it in (also by hand) to fit
the un-touched cells to be smoothly continuous
with the real.
No automatic magic answer box, but I see a lot
of guys keep going 'round and 'round that mulberry
busy and never converging on the answer, because
they're chasing a cloud of fog.
no differentiating between the "sensible" (steady
state) and the "garbage" (transient throttle) and a
street log is highly contaminated by the latter. I've
been meaning to ask them if there's a way to filter
for the same "steady MAP" criterion that is used to
switch between MAF mode and dynamic airflow but
haven't gotten to it. I bet the datum exists but don't
see it, or delta-CylAir, as selectable PIDs meaning a
custom one that spans frames, would be needed.
Need a sense of history.
Anyway, lies, damn lies and statistics. If you see a
large (like several AFR units) difference between histo
"+" and "-" views (min and max) then your data has
too much noise to trust.
So I kind of stick with doing it the hard way; take
steady-throttle pulls, grab the snippet you really
believe in, and work the table by hand with the
real data, then feather it in (also by hand) to fit
the un-touched cells to be smoothly continuous
with the real.
No automatic magic answer box, but I see a lot
of guys keep going 'round and 'round that mulberry
busy and never converging on the answer, because
they're chasing a cloud of fog.
