Narrowbands Reading Super Lean at WOT ($EE / LT-1)

[kur4o]

I have to disappoint you. Pump shot can`t be disabled.
My theory is that you have really lean condition at high rpm. Maybe you have 15.5-16 AFR. What saves you from engine blow is restrictive intake. Really low spark advance and high gears, low weight. I guess engine power suffers alot too. You get some constant 3-4 degree retard, where it comes from.
As you can see your AFGS from 5000 to 6600 rpm change only from 190 to 220, which is really low. At 6400 you should see 320-500 AFGS depending on engine power. Either your Maf is shot or there is some tune issue. Try scaling upper maf table by 20% and see what happens.
It is really important to keep the spread between cells from stock table. So you dont adjust single cell but scale the whole region.
When you tune MAF with wideband at steady throttle, you will get nasty tip in stumble at low rpm when you nail it. Narrowband tuning with closed loop enabled, masks that condition and considering lower weight is normal that you don`t get that lean. I have monitored wideband at tip in and jumps to 16 afr momentary, When you get around 17 you can feel the stumble and misfire. ARound 15.5-16 is not felt. At 13 AFR, power suffers. I am still trying to find the perfect point where you make max power.

[spfautsch]

I have to disappoint you. Pump shot can`t be disabled.

What if I zeroed the adder cells from 240a to 2412? That was going to be my angle of attack. Unfortunately I had another project fall into my lap and haven't had time to look into creating definitions for the pump shot data you posted. By the way - awesome work finding that!

Either your Maf is shot or there is some tune issue. Try scaling upper maf table by 20% and see what happens.
It is really important to keep the spread between cells from stock table. So you dont adjust single cell but scale the whole region.

As I mentioned, I came to a baseline calibration on the MAF without considering the pump shot adder. Since I was tuning with narrowbands at stoich I didn't want to get it out on an interstate and run for long periods above 5000 rpms, so all my higher airflow data was skewed by the pump shot decaying out because it was a lot of accel / decel. I feel more comfortable with the lower airflows because about 85% of my logging data was cruising with fairly tame accelleration.

I always look at the wireframe / graph for tables like this and hand smooth the cells in raw hex but I didn't pay much attention to the size of the step from cell to cell. I guess I'll have to take a look at that. What I did with this table was arrive a baseline that was about 6.5% below stock calibration at lower throttle openings, but at higher airflows the graph diverged quite a bit lean of the factory cal. I guess that should have been a red flag but I didn't give it enough thought.

What I've done at this point is restored the upper 60% of the MAF table(s) to stock and then scaled those cells back by 2% and smoothed into the curve from the lower airflows that I'm more confident is less tainted by pump shot. It looks like the weather is going to be bad this week but I'm hoping to get one day of logging so if I can't get time to work on the definition I'll probably log with the tune I'm at today and see if WOT passes settle out to steady numbers in the neighborhood of 900mv. At least that will tell me if I'm on the right track.

[spfautsch]

I just re-visited this thread and saw a couple details I wanted to address.

You get some constant 3-4 degree retard, where it comes from.

This is burst knock retard that happens at tip-in and decays out (similar to pump shot). I'm not sure it can be tuned out but I wouldn't even have tried until I was sure the WOT fueling was relatively close.

It is really important to keep the spread between cells from stock table.

Would / can you explain better? Knowing how TunerPro works, if it's hyper critical to have the spread between cells consistent then the only way to scale this table correctly would be to shift cells by copying & pasting up / down. Scaling a high precision table like this there are all manner of rounding errors that the conversion formulas exacerbate. This is why I've gotten into the habit of smoothing tables in raw hex.

[kur4o]

Posting your current bin and the stock bin from your pcm will help clear that knock retard issue. Maybe it is normal for manual cars, but for automatic it is strange.

You should not zero that values because it is a scale factor or a multiplier and when you multiply by zero you will get zero AFGS as result. So if you want to play with this make it at least $01 value in hex.

Keeping the spread between cells for ve and maf is really important. Of course there will be some break points where the spread will be increased.
At ve the calculated VE% is an interpolation from 4 adjacent cells, so you always get some mid point as a result. If the spread is way off very small increase in map or rpm can lead to huge change in calculated VE%. And you start chasing tails to tune it.
At maf calculations the raw maf frequency is converted to two bytes value. First points to the maf cell number in the tune, second byte is an offset that divides the spread to the next cell by 255 and adds the result to the first cell value. So there is always interpolation between two cells in the tune.
Since the signal is not linear the curve is very important. You can see it in tunerpro when you open the maf table as a graph.

For scaling the tables use the multiply function. Maf is 16 bit value so rounding is not an issue. You can multiply a whole region with 1.05 or 0.95 to get 5% increase or decrease of the value.

Here is the most recent xdf with all the values entered.

[spfautsch]

Posting your current bin and the stock bin from your pcm will help clear that knock retard issue. Maybe it is normal for manual cars, but for automatic it is strange.

I'll try to post bins later but I haven't changed anything relevant to knock retard but to zero the lower cells and reduce the higher cells of the max retard cruising table at 1233A. This is all my own work applied to the original bin so you could easily compare any factory M6 to a factory A4 bin.

You should not zero that values because it is a scale factor or a multiplier and when you multiply by zero you will get zero AFGS as result. So if you want to play with this make it at least $01 value in hex.

I may try that tonight after I get home. I have to make it 60 miles so I'd rather not get too wild into experimentation until she's parked in my garage. There's nothing fun about having to reflash an older bin on the side of the road. :-)

Thanks for the .xdf - I'll add it to the collective.

[kur4o]

So i took another look at the supposed pump shot adder. After I examined some logs of data, I finally figured out what is does.
At rapid tps change, the maf readings jumps alot too fast, the pcm use this pump shot as a scale factor to gradually increase or decrease the AFGS.
For example the AFGS jumps from 20 to 100 g/s at one frame interval. PCM blends the increase to 30, 50, 80, 100 over a period of couple of frames. That way you don`t get erratic jumps in fuel bpw and engine runs smoother and more consistent.

So this factors how fast the transitions happen. I guess the lower the value the faster the transition. Still it needs to be confirmed.
It turns out maf reacts faster than the time needed for the air to get to cylinder. That`s why there is no ae build in the code.

[spfautsch]

Well I wish I had something intelligent to contribute.

I just looked at the log from tonight's WOT passes and it's a lot of the same.

What I can't seem to square up with common sense is the pattern of rising airflow and RPM coupled with falling commanded AFR, but falling IPW. It would stand to reason if the engine is trending towards pumping more air and the PE AFR targets are calling for more fuel to air as RPM increases, IPW should be heading towards "more"?!?

I'm starting to wonder if my wife has been lacing my whiskey glasses with LSD.

[jthompson122183]

Would adjusting the transient fuel tables help? It's like a "pump shot" also.

[kur4o]

What I can't seem to square up with common sense is the pattern of rising airflow and RPM coupled with falling commanded AFR, but falling IPW.

That can happen because rpm are risng, but the airflow stay almost the same. If the rpm increase with no change in airflow bpw will go lower.
At 220 AFGS at 6600 rpm, I can estimate the engine is around 200 hp. Which don`t make sense with your build.

Post the MAF table to see if there is something wrong with it. The way I see it you need 20-30 % more AFGS from 190 AFGS up.
Shot MAF is also a possibiity.