Sudden drop in APCYL_DMA at 5500rpm

[rhit_rs]

First off, I'll say I'm relatively new to EFILive. I've been tuning with it for about a week now (bought it from wait4me about 4 months ago). The car is an 07 solstice with the LE5 (2.4L w/ E67 ecm) engine with an m62 and LSJ intake manifold and intercooler setup. Now to the problem:

Tuning the VVE and spark tables has been easy enough (especially once I took out the cat protection). I'm used to tuning GM tbi and tpi systems, so the VVE table makes me feel right at home :). The one problem I haven't been able to sort out is a sudden drop in the APCYL_DMA parameter (air in grams per each cylinder as I understand it) that happens at 5500 rpm and the subsequent jump in AFR, which I'm assuming is a result of the lower cylinder filling that the ECU thinks is occurring, when each cylinder is actually still consuming just as much. I've checked all of the parameters in the tune file looking for 5500 rpm, but the only parameter I've been able to track down is the Operating Zone RPM boundaries (B8020), the highest value of which is 5500 (my modification). On the stock tune, this value was 4200 rpm, but data logging the stock engine did not reveal the same event at 4200. Does anyone have any thoughts on why this is happening/what could be causing it? Am I onto something with the B8020 table? How does the ecu calculate the APCYL_DMA? I would think it should be the maf reading divided by rpm/60 divided by 2 since only 2 cylinders are drawing air per each revolution. This comes out to be about 1.01-1.03 for the entire pull.

I've attached a log file of what is happening. There were no sharp changes in the VVE table at 5500 - it was nice and smooth there. I've found other ways to compensate for the drop in grams/cyl, but I'd like to get to the bottom of why the ecu thinks the cylinders aren't filling as well as they really are!

Also, before someone suggests it, B8024 - High Speed Mode RPM Enter is set to 7000rpm.

Thanks!

Josh

[GMPX]

Josh, I wonder if it's something to do with the Intake cam movement, though it seems sudden even though the cam phase is a smooth change.

[rhit_rs]

Josh, I wonder if it's something to do with the Intake cam movement, though it seems sudden even though the cam phase is a smooth change.

That was my first thought as well, but I ruled it out once I realized it was moving smoothly over the entire region. On top of that, phasing the cams should improve cylinder filling (or at least keep it from dropping off quickly), right?

GMPX, how is the e67.APCYL_DMA pid calculated?

Perhaps I should try tweaking the upper RPM zone on B8020 and see if that shifts the rpm that it happens at. Is that table specific to generating VVE tables, or something that the factory calibration uses?

[scdyne]

Sorry I missed this for a day..

1. Start with the stock VVE and copy the values to Excel.
2. Set your B8020 to what you wish. I set mine to 2000, 3000, 4500, 5500.
3. Set your B8021 to the MAP range you will see. I upped my MAP5 to 97.
4. Save a new file and close it.
5. Open the new file up and paste the copied VVE values back in.
6. Generate Coeffs.
7. Generate VVE.

That should work better if you didn't already do that in accordance to the E-38 VVE Adjustment Tutorial.


I neglected this early on too and almost had a REAL bad DAY.

This is what concerns me about other tunes that can't directly regen the VVE like EFILive does. Just moving the zone boundary and not changing the coefficients could (AS I SEE IT) go down a dangerous path of an unknown VE result.

[rhit_rs]

Adam,

I did follow those directions when I moved to the 2 bar VVE table. I copied the table to a different file, then copied it back over when I changed the regions.

My rpm values are set to 1400, 2400, 4000, and 5500.
My MAP values are 27, 50, 80, 120, and 150.

Josh

[swingtan]

Looking at the log shows that MAF air flow was climbing through the whole log, while APCYL shows the drop at 5500RPM.

Given that the MAF reading is an actual measurement and the APCYL is a calculated value, I'd be trusting the MAF in this case. So you are not actually dropping any air flow, it's just that the ECM is calculating a drop from other parameters.

The 2 main factors that go in to calculating APCYL are MAP and the VVE. Your MAP signal has a slight rise in it which leaves me to believe that there must be an issue with the VVE.

Cross checking the WB AFR shows that at the same point, the AFR's leaned out from 12.9:1 to 14.3:1 while the commanded stayed at 12.5:1. This indicates that the VVE is not yet correct and is causing changes in the calculated APCYL figure. To correct this, lower the VVE below the 5500RPM mark and raise it above the 5500 RPM mark. Given how fast the change occurs, I'd say you have a decent "step" in the VVE. This is something you don't want.

If you get a chance, post your tune.

Simon.

[scdyne]

I had to bring up the VVE, but if you are even close to what I have then you are going from one Zone to another at that point.
In that - one poly is trending down and the next is a slope up.
That area you are passing through ends up being a real sharp change in direction.

http://www.scdyne.com/EFILive/vve.jpg

This may not reflect your VVE, but it should be the idea.

Smooth that out like swingtan says and see how that works..


Also try logging GM.ETCTP over SAE.TP

[rhit_rs]

Swingtan and Adam, thanks for the advice. There was a dip in the VVE table right after 5500rpm. I think the last time I made changes it was smooth, but when I recalculated the VVE tables based on the updated coefficients a bit of a dip showed up where I didn't notice it on the VVE table (I didn't rotate it around and look at it from different angles). In the mean time we used B0148 (Open Loop Fuel Load Multiplier) to fix the problem. Now I'll be able to put that table back to all 1's and fix the VVE table appropriately.

I would have posted the VVE tables but I'm away from the car and flashscan for the next few weeks (back at college).

Thanks again for the help everyone!

Josh

[scdyne]

Josh,
I'm glad you found the REAL issue and didn't just keep B0148 modified to compensate. I have been composing a VVE Primer for the LE5 guys for a while and haven't posted it because of how contrary it is to what other LE5 tuners have been preaching for a long time. In fact there is a very well known 'dip at 3200 rpm' documented in both EFI tuning and dyno runs that exposes a zone transition 'flip-flop' like what you saw. Other tunes have done exactly what you suggested and just changed the B0148, but I contend that having an accurate VVE to stat with is a better way to go.

When all is said and done you'll have by far the most complete tune for an LE5 equipped vehicle around..

[Mileage]

I like the idea of adjusting the zone boundries for increased resolution and it would seem to be a must when doubling MAP but does this typically increase the discontinuity at zone boundries? I need to look at a few stock .tun files to see how "pretty" they are at the boundries. Thnanks for the help everyone. Ken

[scdyne]

Ken,
Due to the mathematical aspect of the coefficients in relation to the VVE it is not beyond reason to induce bad areas in the VVE by just changing the zone boundaries.

By changing the zones you potentially allow a given coefficient group to extend out further and if a curve is already trending down it will continue to do so. Additionally since the polynomials are second order its conceivable that the trend will double it's bad direction quickly.

Along comes the next zone and a new group of coefficients likely more in line with what the stock VVE is and you see an abrupt change in the numbers.

Everyone talks about a spot around 3200 RPM as some weird, unidentifiable drop in power. That's nothing more than the stock VVE table crossing from one zone to another. Once you add an intake it becomes evident and with a header even more so, but rather than properly address it in the VVE it's taken care of in other ways.

We have the advantage of having a solid software package that allows us to visualize these potential conflicts in the VVE path and fix them before even getting on a dyno. Other options are stuck trying to clean up the unknowns by adjusting other tables.. that often compromise the closed loop operation of the vehicle.

[swingtan]

scdyne, you are correct. The whole idea of tuning the parameters in the tune is to optimise the tune and make it the best it can be. There is no use in "fudging" tables to cover up problems in other areas. The VVE can be difficult to work out and it can get frustrating at times, which is why I made up the VVE Adjustment Guide (http://download.efilive.com/Tutorials/PDF/Virtual%20VE%20Adjustment%20Tutorial.pdf) in the first place.

When I adjusted my zone boundaries, I looked at the where I needed the most constant settings for both MAP and RPM. With RPM, you get 5 zones while for MAP you get 7. So the air is to put your required steady states in the centre of a zone to allow for the curves to give the required values. For example...

Idle: 750 RPM and 45 kPA
cruise: 1400 to 1600 RPM and 30 to 35 kPA
Gear Changes: 2500 to 3000 RPM and 20 kPA
Loaded up: 1400 to 1600 RPM and 75 kPA
WOT: 4500+ RPM and 100 kPA


So I set my boundaries to try and have those spots in the centre of the zones.

I've mentioned a few times that you simply cannot adjust the VVE with lots of little test drives. All you end up doing is pushing the values up and down and not ever really getting the right shape. There are 2 ways to gather data for the VVE,


On the dyno: Obviously the quickest, easiest and safest method. You can gather lots of very specific data for the cells and zones. It really should be classed as cheating it's so easy.
Long drives on the road: It takes longer to gather data on the road as you can't hold the loads like you can on the dyno. You also loose a lot of data accuracy because the the driving conditions are changing all the time. But with enough data, everything averages out and you get some good results. A 1 to 2 hour drive in varying conditions, changing up and down gears to vary the RPM's will give good results.


Once the data has been applied to the VVE, you also need to manually "massage" the shape to ensure that the curves go the way you want. After a while of playing, you can estimate what the generation of the coeffs. and VVE will do to the shape of the VVE and you can allow for that in your adjustments.

Simon.

[Mileage]

Thank you Adam and Simon.

Simon, the first paragraph of your guide references "Virtual VE Tutorial" By EFI Live. When I go to the EFI Live download page all that I find is your guide. Is it the only documentation now? I found the "AutoVE Tuning Tutorial" but your guide references that as well. Thanks

[swingtan]

Open ScanTool or TuneTool and then click on Help - Tutorials - Virtual VE Tutorial.pdf.

It should come with EFILive.

regards,

Simon

[Mileage]

Thanks. I will look tonight. I was looking on line. Just when I am getting accustomed to everything being "on line" this is with the software!

[Mileage]

Found it. Thanks. Ken