Can't Get Back Into Closed Loop

[cmitchell17]

I just put a TSP Stage 1/2 228/236-114 .635/.635 cam and ported heads with 015in milled in my LT1 Camaro and nothing is making sense.

I've been trying to tune VVE and I have raised the coolant temperature limits to turn the MAF off, put in the maximum seconds to the closed loop enable delay, and disables LTFT coolant temp enablers as well (There is no STFT disablers I can find in my tune). I was able to make some corrections to decrease the VVE table in the lower RPMS (about -20%) based on wideband feedback, but when I tried to make more adjustments to more areas of the table it kept calculating bad coefficients.

I turned my attention to the MAF calibration, even though I am a strong believer in the only time you need to change MAF is when you alter the air intake by the sensor and any other time you change it is just a "band-aid" fix. I did'nt really get anywhere with the MAF, especially since the stock scaling basically maxes out in the 8500Hz range and with my cam and heads I am exceeding this reading up close to 10000Hz. I tried to change the scaling to get more resolution and I tried re flashing and the car wouldnt even run on it. So I kind of gave in for now on the MAF even though I don't think I need to change anything (I have a Rotofab intake, which you are supposed to be able to use the stock MAF calibration on).

Basically, I am almost back to stock now, and although, at least today, it seems WOT is running a little better. What makes no sense is WOT feels like it is missing and falling on its face, even though AFR looks decent and isn't too far off of commanded. Timing is around 10-20 degrees, which should be enough to make it run decent, with less power of course, but not enough to make it feel like it is missing and stumbling at WOT?

Even though I have returned all the closed loop and MAF fail to stock enable values, it is still stuck in closed loop. When I use the special functions to force closed loop it goes extremely lean (above 20:1 and past that as the wideband no longer reads) and somehow keeps running. Yesterday it was running rich, although I was running a little more E85 in the tank yesterday and I am returning to 93 to try to eliminate some variables. Either way nothing is following convention or making sense here.

Theoretically, it seems like to me even with the stock tune it should have some room to compensate for the theoretically lower VE at lower rpms and higher VE at higher rpms that the cam requires, besides that I don't get what else you even need, or should, change for a cam swap?

Attachment 96538

[cmitchell17]

For some reason my log file isn't uploading, but the tune file is?

Here is a link:

https://1drv.ms/u/s!AkCUE89IhD0Ng91-...sQPxg?e=7NyV6e

[kevin87turbot]

Your DS oxygen sensor is dead. It's stuck at 1107mV all the time. Very likely a bad sensor. Once you fix that issue, you'll be good to go.
What codes are you getting?

[cmitchell17]

So I have my drivers side O2 removed and put the wideband in place, I was going to change the O2 configuration to "single bank" to ensure the ECM did'nt do anything weird. Instead I just turned all the codes off to "not reported".

So it could be that the ECM is going into some self correct fail safe? I figure if I do change it to "single bank" it most likely will assume the drivers side is the only operational one and won't work? I guess the only way to find out is to try it. I could move it over but I've spent all day messing with this stupid exhaust under the car since I dropped a socket into the manifold.

good.jpg

[cmitchell17]

Ok so changing to "single bank" didn't seem to work and maybe have made it worse. It's like you never know when its going to go super lean or super rich and when its going to die in a parking lot and when its going to idle down normally.

I just don't get how the wideband is reporting lambda in the .975-1.25 (14.0-15.0 AFR) and at WOT its reporting pretty close (12:1, 10.8 at the richest) and it just runs so bad and feels like its almost misfiring. Most of the time AFR looks to almost be spot on with commanded, and with 17 degrees it should run decent and you shouldn't be able to feel any stumbling:
good.jpg

Also I had to take off my downpipe today to get a socket I dropped in there, since I thought it was what was making the rattling sound, and it turns out my cats are cracked and it appears this is what is making the rattling sound.
IMG_20200209_165359_01_01.jpg
I feel like there's no way having the tune off so bad caused this? I hope my wideband isn't screwed up and reporting bad values and its actually really lean and killed the cat? I'm not sure of the failure mode of cats, if they just melt or crack? I think I have seen old ones eventually start breaking up and blow out the exhaust pipe.

I am about ready to give up on this and pay someone for a "canned" tune even though I know all these "tuners" do so little and just get it running somewhat and don't address drivability.

[5FDP]

The single bank thing isn't going work as you found out, you need two primary o2 sensors.

[cmitchell17]

I decided to start over and go back and try again at VVE. I used a little different method to kill the MAF and closed loop enable, but there is still no way to kill STFT, only LTFT that I see? Anyway I feel like I got a really good histogram and got a lot of good cells with a lot of hits, however, getting the coefficients calculator to calculate something meaningful is another story.

There still looks like there is some problems with the histogram data though, generally it looks like the error values are lower (negative) in the lower rpms and higher in the higher rpms, which is what you would expect changing to a bigger cam from stock. However, there is still a lot of places with sharp discontinuities. Filtering out the cells with lower counts helps a little, but this makes me wonder is my wideband giving me bad data? What is the source of these discontinuities?
good.jpg
I mean there is no way the next cell over requires a 100% increase in VE while the cell next to it requires a 7% decrease in VE? I have adjusted the count filter up to 15 now I guess I can keep going but I lose a lot of data. Has anyone been successful in making a filter to eliminate or search for this erroneous noise?

Instead of copying, then pasting and multiply by %-half, what if I just create another histogram that looks exactly like the one I just made, but just lower the resolution to something very simple like say a 10x10 table to start off with? I feel like this would allow you to calculate the coefficients easier and more stable, then you could further increase the resolution each time you did a run and calculated the coefficients?

The way I am doing it now I feel like everytime I make a correction to the VVE it gets more unstable or worse.

The good news is with closed loop and MAF disabled just using SD it seems to run a lot smoother and a lot smoother at WOT to, it only seems to cut out a stumble just a little, however it feels extremely slow and makes it feel like all the hours of work I've been doing installing and tuning the cam is so not worth it for the power increase it looks like I am going to get.

[kevin87turbot]

You can just zero the O2 Sensor readiness voltage to in order to force open loop.
You need to increase the number of hits required for your histogram...and go for a longer drive to get it dialed in properly. As you've surmised, you don't want adjacent cells with a difference of 100%...

[cmitchell17]

Ok so I feel like I might be getting somewhere. I have significantly reduced the resolution of my histogram:
good.jpg

I am filtering count values out below about 100 or 150 counts, which I feel like is a lot, but obviously I am capturing error somewhere.

However, I've done this with 3 or 4 iterations and I feel like I am not converging the error towards zero. It seems like its not doing much. I am just copying my histogram error and pasting and multiplying by half %. I am assuming the error is setup correct as far as sign convention goes (positive or negative error) (I am using the default HP tuners EQ SAE Error PID)

The problem seems to be that when I go to calculate the coefficients, HP tuners gives me a warning about not including all my zones, but I can't find anywhere where the zones boundaries are defined. I see the boundaries that are defined in the dynamic section of the tune file, but those are for RPM and P-Ratio, which I don't think have anything to do with the SD parameters? or do they?

I feel like there are some filtering techniques we could use, AEM quotes their wideband with a typical response time of 20ms, I can't narrow down what is going on to make some of these cells swing from massively different error values.

Also, does anyone know what the numbers in the virtual VVE table represent? They are in the 1000-3000 range? Anyone have any good histogram/VVE table setup boundaries and ranges they have found that worked good. With these 3 or 4 iterations I have done I feel like, especially at idle, I am still at around 11:1 AFR, so either I am doing something wrong when I go to update the VVE in the tune or it just takes a lot more iterations?

[cmitchell17]

Ok so I've done 3 more iterations and I am still getting the same errors and I am not converging towards zero error. I think I know why though, the coefficients are not being calculated correctly, whenever I go to update the virtual VVE table then calculate the coefficients, the HP tuners software is giving me the exact same coefficients?

So what am I doing wrong here? Is it because I don't have all the zone boundaries included in my virtual VVE table? Where are these zone boundaries defined?

[kevin87turbot]

Make sure you make those changes to all four VVE tables (technically, you can get by with just doing the non-DOD tables.)
If you're just making the changes to one table, it won't work.

[HeavyChevy305]

150 counts has always been the minimum and as mentioned, you have to make changes to both Manifold switch closed and open.

[cmitchell17]

Ok thanks for the help again everyone.

Now I have a different but kind of the same problem.

I kept messing with VVE and experimenting with smaller resolution tables and only calculating zone by zone and it seemed I couldn't get it any better than about 7 or 8 percent plus or minus. The good news is I could tell from the wideband feedback idle AFR had improved and was a lot closer to stoich and also WOT felt really good and commanded was matching actual from the wideband at WOT as well (even though I have heard VVE isnt used past around 4000 rpms, it still somehow magically fixed WOT for the most part). I then adjusted timing some, since I mainly had to add back the timing lost from the VCP timing tables since I have a zero degree phaser lockout.

I decided to pull my wideband out and put my stock O2 sensor back in. I then re-enabled all the DTCs for the O2 sensor on the drivers side.
So then I went and tested it out and it actually felt worse, like it was more rich but it didn't seem like fuel trims were indicating it was that far off. So when I went WOT it fell on its face again (I can't see AFR since I took the wideband out and put the stock O2 back in).

So something is screwed up about closed loop, I tried resetting fuel trims (which I think it does anyway when you reflash) but it didn't work. It was running so bad it threw a misfire code. Now I am assuming it is running lean since the stock O2s at some points were even below 450 V. The only thing I changed when I put the stock O2's back in was enable all the O2 DTCs, so that made the ECM command closed loop, so before I must have been in a pseudo closed loop that was really open loop because of the missing O2 sensor. I just don't get what could be so different about open and closed loop. If it was running fine in open loop, then I reflash the ECM and it goes to closed loop. The fuel trims should reset to zero and it should still base fueling on VVE and dynamic air from the MAF until it gets some feedback from the O2s, which didn't really seem to be far off. I just don't get it?

I have included 2 logs, one after I put the stock O2 back in and re enabled the DTC's where it then commanded to go into "no fault" closed loop and ran like crap, then another log after where I just forced it into open loop by the O2 readiness voltage where it runs somewhat good (still has somekind of torque reduction on WOT downshifts (although this may be due to hitting the rev limiter since WOT shift MPH may be off or need adjustment)).

Log 21 Put Stock Narrowband Back in and Turned DTCs Back on and Started Running Bad at WOT Again.hpl
Log 22 Disabled Closed Loop and Runs Good Now.hpl
Fifty Tune Modified ECM_TCM_FICM.hpt

[cmitchell17]

So I think i'm pretty sure I found the problem. ECM table 277 says to "set to 0 to disable" for the "number of variable camshafts", after I put my zero degree phaser lockout I set this table to "none". Apparently you have to leave this table at the stock values (1 in this case for LT1 single cam) or else it will put it into some weird fueling mode that makes it run awful. So it set it back to stock value of 1 and then zeroed out all the cam timing tables and now everything seems to be working and O2 sensors are switching.

Now that that is fixed, it is doing some weird timing pulling:
good.jpg

But you can see in the place I circled it is pulling 19.6 degrees of timing at WOT. You can see below where I circled the transmission is not slipping clutches anymore so the shift has completed (and there was a torque reduction event just before from the WOT downshift to 1st gear). There is also a TPS torque reduction in that same WOT run, but I know why it is there because I am hitting the rev limiter and I need to adjust the WOT shift points or raise the limiter. During the torque management event both predicted and immediate engine torque sources are axel so I don't understand what else could be requesting this reduction? All the torque values and sources and pedal/driver/engine modifiers make no sense and don't seem to correlate to anything in the tune file.


20-02-14 17-38-50.hpl

Fifty Tune Modified ECM_TCM_FICM.hpt

[cmitchell17]

So I have looked over the torque PID's some more and I may have found some type of weak correlation:
good.jpg
You can see in the points where I drew the arrow where the "Actual Axel Torque" starts to exceed the "Immediate Axel Torque Cmd" PID, it then pulls timing, the "Actual Axel Torque" drops, about 400 lb-ft below the "Immediate Axle Torque Cmd", then starts climbing again.

So is the ECM trying to get "Actual Axel Torque" as close to "Immediate Axel Torque Cmd"? If this is true, it seems the control parameters are pretty slow and actual axel torque is overshooting immediate command pretty bad, about 400 lb ft, this dosent seem logical since timing can be changed almost at an instant by the ECM.

Is it more logical to think that "Immediate Axel Torque Cmd" is acting as a limiter, and timing is being pulled when actual axel exceeds immediate?

Either way none of these PID's seems to have any correlation with anything I can change in the tune, so I guess the only thing I know to do is keep increasing the driver demand.