PorKen

The measurement is not the problem, it's how much air the engine is capable of pulling through the MAF.

I think the max MAF units for a regular MAF is over 600?

My S3 could pull into the 420s on a cold day.

PorKen

Ignition timing wise, after hours of logging/adjusting, the profile of my 3D WOT table has ended up looking a lot like the stock 92-octane-cat WOT table.

Although it has a bit more detail than stock, my map is still too coarse, especially at peak HP.
There are areas that could handle more advance, but the row breaks are such that I can't get to them.
EG. the last three rows are 5600 | 5900 | 6300, which is barely useful.


I have not had time to work on the EZ RPM scaling (which is more complicated than the LH).
Eventually, I hope to wrangle a bit more average power through 'micro-targeting'.

Dave928S

Loving this thread ... I think your title needs to be changed from 'Inventor' to '928 Grand Wizard'

gruffalo

Do have any idea how much power you have gained over the stock chips?

The X-pipe alone has been known to give a power boost in the region of 20-30 hp....

PorKen

At this point just ~9 peak vs. stock EZ/stock-ish LH, both Ott-X.
~20 at 6300, but that's not very useful.


There is a big dip in ignition advance around 3900 and a smaller one around 4800 because those areas are knock prone.

Might be able to dump fuel in just those areas, sub-12, then up the advance.
Up to this point I have been trying to make a flat AFR. Now I can try selectively enriching problem areas.

More columns around 3900 may help too. I have 3600 | 3900 | 4200 to work with now.
100 rpm or 50 rpm breaks would allow more total advance.


The notch at 3700 can be filled in by opening the flappy a little sooner, as seen in the blue line.
(The first TQ peak is created by closing the flappy.)

gruffalo

Keep up the great work! This is the thread I look for every time I log in

PorKen

Long ago, I looked at a 3D table and assumed it was the knock sensor sensitivity table, from what I read about Volvo EZKs.
Nope. It was the coil dwell time table. Doh! In testing I had reduced the values in the table by 10% and forgotten about it.
The dwell map compensates for battery voltage and rpm, increasing the charge time with less system voltage.
I had been running with reduced coil voltage all this time...

Anyhu, I found the real knock sensitivity table (along with most all the other knock settings, there's one for each of the 3 octane codings).
This map reduces the volume of the knock input as rpm goes up, because there is more mechanical noise.
On Volvo etc., this map is 3D, using coolant temp as an additional factor.

I reduced the knock volume at WOT just a touch overall, a little more at the TQ and HP peaks so the EZ won't retard on tiny knocks.

After smoothing out the fueling before the last dyno run, I went about adding fuel where it knocked and upping advance where the fueling was rich.
3000 rpm, for example, needed a LOT more fuel. In turn I was able to increase the advance there quite a bit (>4°).
Generally speaking, it's OK to have rich spikes. Lean spikes will usually cause (or indicate) knock points.


Just like with the S3 chips, with all the extra fuel in certain spots, you might expect of a lot of rich dips, instead, the AFR graph (purple) became more and more level and consistent.
You can see the big peaks in the injector duty graph (red). Judging by the injector duty and advance, it's making a bit more power that the last run.


12.3 on my LC-1 is about the AFR limit for keeping knocks away. Reducing the overall advance makes the car slower.

I'm hoping that a SLC-X will let it breathe better over 5700, make it less random in the process. (May have the new X to log/dyno in a few weeks.)
Occasionally the AFRs from 5800-6600 are nice and flat.


The red line, injector duty and the WOT map kind of approximate the TQ graph on the dyno if you turn your head a bit to the left. (Compare to the run above.)
Note at 3100 there is a dip. I like to think the flappy closed (first) hump is a funhouse mirror reflection of the later, flappy open, torque peaks.

Sometimes, depending on intake temps, voltage, sunspots, whatever, the engine can't use all the fuel at the peaks and there will be deep rich valleys.
It's tempting to take out fuel when you see this, but sometimes you just have to wait to the next log run.


AFR is initially rich as the airbox sucks down a bunch of hot air. Look ma! No knocks (blue)!


If only there were columns for 5700 and 5900! This contributes to the randomness above 5600?


Stock knock sensitivity table. All three octane levels have the same values.

Mrmerlin

Thanks for sharing all of this great information Ken.
And taking the time to make a great product

FredR

Ken,

Great stuff- you are explaining some of the things I have observed with ST2 but have no clue as to why they seem to happen- especially the randomness of the AFR at higher [5k plus] rpms. My motor seems to want to run at about 12.4 average AFR up there but the scatter seems run at about +/- 0.4 AFR units no matter what I seem to do with the settings. I tended to think the problem must be related to my ineptness/inexperience but perhaps not. When I tried to richen things up a bit more [average AFR 12.0] it did not seem to help the tendency to knock at about 5300 rpm so I reverted back to 12.4 average AFR -remember I have only my assometer to go by plus whatever ST2 is telling me.

Similarly I found that the stock mapping seems to like additional fuel at the 3k mark and then you can dial in more advance. During a recent problem episode when I suspected I had an LH issue, I ran the car with my spare units- I have to change the LH to the PEM module but I ran the car the with the stock EZ temporarily. Forgot about this, thought the car felt somewhat flat then it dawned on what the problem was. Replaced the EZ with the PEM in it and it was interesting to note the wake up factor in the motor so yes- tuning does make a noticeable difference on EZ alone.

It is also interesting that you have been able to tweak with knock threshold sensitivity. Not sure I fully understand what is going on here but I suspect that is something we cannot adjust within the current configuration of ST2- I am sure John or Jim will correct me if I am wrong on that one if it is a question of terminology.

Keep up the great work!

Regards

Fred

John Speake

You can't alter the threshold for knock detection in ST, and I'm not sure that is something I would want to expose to the general ST user. You can adjust the retard step size per knock and also the retard recovery time. Plus several other parameters such as the degree of retard applied when you suddenly apply a bootful of throttle.

PorKen

FWIW, here is the stock coil dwell table. (These are the raw, unconverted, numbers.)
This table controls the coil charge time in milliseconds before the spark is triggered, according to rpm and system voltage.
There is a small range where the coils are happy. Too much current, coil overheats, too little, not enough spark.

You would need to modify this table if you were using non-stock coils.


Here is a nice writeup on what the table does - Ignition Coil Dwell Calibration
Kinda makes me want to buy a USB oscilloscope.


Not sure if it means anything, but note the jump after 3000. (3000 is prone to knock.)


I am guestimating the voltage numbers 10, 13, 16.

fraggle

USB scopes are cheap, Ken! Try here:

http://www.saelig.com/

ptuomov

On setting the knock volume threshold: I think that this would be absolutely useless for most users, an opportunity to do something stupid. However, if someone say builds an engine with 2618 alloy pistons with centered pins, the noise level is going to change and one would presumably want to increase the knock threshold. Even that would be a difficult calibration exercise, I can't think of a way to do that without cylinder pressure sensor equipment. For example, this: http://www.tfxengine.com/ (The cheapest TFX model is $7k so it's probably only for people who own dynos.)

EMan 928

Do we have an estimated availability date yet?

PorKen

Really just waiting on a SLC-X. (Hopefully, Dave will have time to fab one soon.)

The tune is stable at various temps. I have stopped making changes.
I don't expect that it will require much tweaking from the Ott-X that is on there now, but you never know.

I have to
- log/dyno with SLC-X
- recheck custom code for errors
- code a smoother rev limit (recovers too quickly, especially with NLS)
- decide whether to detune the EZ for 91 octane (no multiple octane ratings)

Eventually, I would like to enable the CO adjust and maybe make a separate WOT adjustment.


One advantage with the S4 chips is I have kept the cruise and idle maps essentially stock (with a few tweaks).
With so much extra code and map space (vs. S3), I was able to make completely separate code/maps/constants/etc. for WOT.