Been chasing this for a week or two and had great comments on it (thanks engineermike). Has anyone else seen knock on cruise (or LSPI) on the ranger, or more specifically, the 2.3?
I seemingly get it around 9th and 10th gear, which sure makes sense because of the gearing and load, but it just seems aggressive that i will randomly see up to 3 KR. Once the octane ratio gets to about .90 it seemly gets pretty aggressive and i will see spikes where it adds timing, and then where it is retarding it.
Essentially - is this acceptable? this area of cruise is completely stock, nothing touched - and I am wondering if i should just remove a few degrees from the borderline maps (as this is always the spark map that it is using).
See attached log for a little cruise on route 80, there is a small pull at the end - which is tuned (only higher loads are) and something i am working with.
This is what your knock retard looks like for all 4 cylinders. It looks like it's doing what it is supposed to, to me. Knock learning appears to be working as it should, but that just affects the "Borderline" value, or the starting point. It will still advance the timing until it either hears knock or hits MBT. That's why I said to log and plot Borderline and MBT alongside Timing in the other thread. If you remove timing from the borderline tables, it will just start at a lower number but it will still advance from there until it hears knock.
I hear what you are saying (in the other thread) about any knock being bad, and I don't disagree. Ford, however, runs their engines into knock regularly as part of the control strategy. My guess is they are very confident that their knock sensing strategy is good enough to hear and correct it at very light knock that doesn't do any real damage. Personally, I'm still uncomfortable with this because as soon as we start tuning we are pushing the engine outside of the envelope Ford intended it to run in. I spent months tweaking my borderline tables to avoid knock completely on 93 octane. Once done, I subtracted 2 deg across the board then limited knock advance to 2 deg, just for a little extra safety if I get a bad tank. I rarely ever get any knock at all, as a result.
Notes on your WOT pull:
It shows your spark source as cyl pressure (though it still picked up knock). Many raise the cylinder pressure limits and use the borderline tables and knock advance limit to set the WOT timing.
Also, your highest load in the log is 2.3. Your spark tables are all undefined past 2.0. You are in dangerous territory! You need to re-scale the spark tables so you can populate them up to 2.4 or 2.5 load. Your tune is applying the same spark values at 2.3 as it would at 2.0, which is bad because timing needs to come down as load increases. This task requires some extrapolation, but not too bad. Also, when you re-scale a table you need to write entire.
It looks to me like the borderline tables are pretty accurate, as the knock advance hovers around 0 +/-3. As an experiment, set the max knock advance (ECM 44684) to all 1.0 (where it's currently non-zero). Do the same highway test and see if it sits at a constant 1.0 advance. That may get you close to the result you want to achieve.
The old paradigm of knock is bad is less of a thing these days...
More powerful controls, far more sensitive knock sensors, and quite most importantly, direct injection.
Yes, knock is still bad, but in small amounts, it's important feedback for fuel quality and tune self refinement.
I doubt you're getting LSPI, unless you've raised the LSPI table max loads in low RPM too much.
I tend to leave those stock at 2k RPM and lower and stock above 150 degree charge temps, Ford did a pretty good job setting these up to keep them safe from LSPI... Though not saying it can't still happen, it's not perfect of course.
Reply With Quote