accel enrichment tuning

[bk_]

I am having problems with accel enrichment tables and would like some input on how to tune it properly.

The main injection table - MAP vs RPM - has already been tuned on the dyno, and confirmed OK with another ECU - PowerFC

Settings that I have available:
1) out-of-phase injection time (accel vs RPM, uS units - 2D table)
2) accel correction (accel vs RPM, % units - 2D table)
3) accel correction % based on ECT (ECT, +% units - 1D table).
4) injection acceleration decay: % (one value)

The default settings do not help as they are suited for 4d-table tuning, rather than the typical tuning tables (MAP vs RPM main injection table).

This is for a 1JZ-GTE with stock twins at 14psi.

Thanks.

[thechuckster]

being turbo based you want to accel tune for changes in MAP x rate/speed-of-change of TPS position. Can you advise what the 'accel' variable is derived from?

Also, having a wideband sensor will help.

[bk_]

accel is throttle change (percentage) over time, which is sampled every 50 mS.

eg: 10% accel would mean ~0.30 V change in 50 mS.

wideband has confirmed it does run rich (on quick accel, AFR drops to ~12, even though low boost was tuned for 14.0-14.7)

[thechuckster]

problem you may find using a delta TPS+RPM only accel enrichment regime is that there is a significant band of rpm where the motor can be in boost or vacuum territory and thus have very different fueling requirements (e.g. end up with accel enrichment that's great not in boost but crap when at higher boost).

My experience is with the megasquirt ECU and some basic guidance on accel tuning can be found here: http://www.msextra.com/doc/ms2extra/...ual.html#accel

But in a nutshell, they suggest:
- get fuel table good (e.g. tune with accel enrichment off)
- tune for gentle throttle changes first (e.g. the cell(s)/bin(s) that enrich with minimal throttle movement):
--- start with a high figure then reduce until the engine stumbles/bucks (lean event)
--- if your starting figure stumbles/bucks, double it and then adjust downwards
- tune for increasingly more aggressive throttle changes (e.g. the cell(s)/bin(s) that enrich with significant throttle movement) with the same 2 steps
- tune in the accel taper time(s) (aka accel decay time)
- finally tune decel (can also referred to as decel taper and decel fuel cut)

[bk_]

thechuckster,
Thanks for the ideas and the URL.

I do also have access to altering throttle-vs-RPM injection correction table (throttle as in absolute throttle position, not throttle-change)

I can also perform target-AFR (MAP-vs-RPM) tuning onto the throttle-vs-rpm correction table which will probably make life a little bit easier.

This would probably explain the problems, and why the ECU must be tuned by a combination of speed-density (primary) and alpha-n (secondary / % correction), as intended by the ECU manufactucture (and ignored by most authorised tuners of this ECU)

[bk_]

OK... The best thing to do is to graph the logs... and here they are:

AFR=11+Y
RPM=Y*(7200/5)
AFRtarget was generated based on MAP, as main injector table was tuned on dyno.
throttle=Y*(100/5)

1500 RPM, 18% accel
1500 RPM, 23% accel
1500 RPM, 35% accel
2000 RPM, 5% accel
2000 RPM, 5% accel
2000 RPM, 8% accel
2000 RPM, 10% accel
2000 RPM, 10% accel
2000 RPM, 10% accel
2000 RPM, 30% accel
2500 RPM, 10% accel
2500 RPM, 15% accel
2500 RPM, 15% accel
2500 RPM, 20% accel
2500 RPM, 25% accel
2500 RPM, 25% accel
2500 RPM, 25% accel
3000 RPM, 15% accel
3000 RPM, 15% accel
3000 RPM, 30% accel

I can see (and did feel) hesitation where RPM oscillates due to too much wake-up injection pulse. But apart from that, there does not appear to be acceleration enrichment / decay involved?

And decel is quite a bit rich?

What's the ideal decel fuel cut off / return / delay time (constant)?

[thechuckster]

do you log files haver injector pulsewidth? watching that will show you what amount of enrichment is occuring - AFR is helpful but it's a bit after-the-fact.

however, looking at the log files, i doubt there is any substantial enrichment occuring. And yes, Decel should should show the AFRs taper then cut to free-air (pulsewidth = 0).

fwiw, I found the post accel enrichment taper to tune, and after much fruitless testing, I set decel fuel to to 1sec if RPM was over 1800rpm (e.g. TPS goes to 0%, after 1 second cut fuel if rpms are above 1800rpm and the motor is warm). Some ECUs let you do more clever rules for decel triggers (e.g. no fuel cut if >100kpa or road speed is xxx, etc).

[bk_]

Yes, I can log like 10+ different fuel injection correction parameters. But since acceleration (not out-of-phase / wake-up injection) was set to 0% at 80C, the result was all zero.

I have now realised that the default settings for acceleration vs ECT correction was set for: 100% at -20C, 0% at 80C.

So once ECT hit 80C... there is no accel correction (besides wake-up injection)... because the assumption made by default is that the engine would be tuned with a combination of alpha-n (throttle-RPM) and speed-density (MAP-RPM).

So last night, I decided to add a little acceleration correction (set accel vs ECT @ 80C to 10%) and it does feel different (better higher up, worse at smaller throttle, due to too much wakeup injection).

I will have to write my own logging and graphing tool because the as-is software is pretty crap at it (no realtime display of logging, and no auto-trigger). Not a problem for me.

Thanks for the ideas and assistance. Much appreciated!

[bk_]

Hmmm... conclusion:

1) out-of-phase / wake-up injection affects AFR within the first 0.1-0.5 seconds. I was initially off by ~10%. Could do a little bit more work but it's OK.

2) added acceleration correction between 1-4% and 1-8% which only 10% of that is applied (so 0.1 to 0.8% of ~5000uS with 500-1000mS decay), for 2-60% throttle acceleration.

3) between +0.3 and +0.7 bar boost, and 2750-3000 RPM, I added +10% more fuel. for 2500 RPM (and around the +10% block), added +5% fuel. This solved the strange high-AFR issues pretty much spot on throughout the boost progression. Automatic target AFR tuning also helped massively.

I dont know how it was tuned on the dyno, except for the target AFR table I used... perhaps it was thought the turbos boosted up after 3000 RPM, instead of 2500 RPM. Who knows. That was 12 months ago.

After all that was done, I real-time graphed (sorry, I didnt take screenshots of my custom-made app that I wrote yesterday/today) from ~1500 RPM in 2nd and 3rd gear to full boost. actual AFR and target AFR was within 0.3 for both scenarios, rather than 1.0-1.5! (take into account poor analog signal and exhaust-to-sensor delay).

Next stage is fine-tuning ignition. That's for next weekend.

[bk_]

And another problem caused by imperfect tuning software / firmware.

The software assumes individual injection, not banked injection.

Which means the following problems occurred:
* accel enrichment default is twice as much than it should have been, or atleast 1.75 times (I may have to re-add ECT-based accel enrichment correction, as it feels it needs more injection at lower than 80C)
* crank injection default is half as much as it should be.

A lesson to never trust tuning software for defaults.