How can we trust wideband or narrowband O2 feedback with big cams?

[cmitchell17]

I just put a set of headers and recently a LT2 intake on my LT1 with a 228/236 - 111/114 cam and ported heads. I decided I would adjust VVE knowing from memory over a year ago that you may be able to get it initially running from stock, once you hit a certain point, given the numerical errors in generating the coefficients along with the confusion in EQ ratio, AFR, FAR, and lambda you just end up basically having to give up. Even when only pasting the errors by half you never seem to converge, one day you see it rich in one area, fix it, then the next day its like its fixed, then the next day its richer than before even though you leaned it out.

However, I started thinking, at what point do we have to ignore the readings from the wideband and consider them invalid? My cam will begin to lope below about 950 or so RPMS, I would think wideband readings are completely invalid under lope conditions? It certainly could explain why no matter how many iterations of VVE correction I do, it just never seems to converge in the low rpm cells and idle cells, because maybe its getting a false rich or false lean reading?

I guess my VVE was probably good enough given my trims usually stayed under about 10%, but I also think if the wideband readings are invalid when the cam is loping, so will the stock narrowband fuel trims at low rpms when its loping?

[Jonblarc7]

I fought that alot and what ended up working the best for me is to use both seperatly. And make sure I don't do any logs until the truck is warmed up. Mine are now within 2% and holding steady at .87 to .88 under WOT which is what I'm commanding

I ended up just using the narrowband for anything out of PE. So I would ride around doing everyday driving and try to hit ever cell I could without going into PE. Once all that was dialed in then I would log a WOT pull using the wideband error only and copy and paste just that section. But I would start the log right before the run at 2,000 rpms and end the log as soon as I safely could after the pull.

[Alvin]

It takes a really really big cam to fool a good wideband. Narrowbands on the other hand will not give good data on cams larger than say 230ish on the intake side. At the very least in closed loop they will correct themselves to running poorly.

Car I did this week had a 24x/26x on a 112 I believe. Wideband still worked well for idle. It did have some problems on higher RPM very very light load.

Exhaust leaks and misses play a bigger role when it comes to widebands.

The fueling will never be perfect day to day.. If your getting a running average of +/- 5% in the driving parts of the VE table your doing good. Wide open will usually hold a little tighter but it isn't perfect either.

[smokeshow]

If I could give any single piece of advice on calibration from what I learned in industry, its this... Every algorithm in an ECM that makes a stock engine work as advertised is designed for that vehicle with little room for deviation. It's not just the calibration parameters themselves. So at some point between bolt ons and a cammed/turboed/nitrous/etc frankenstein setup, those algorithms are going to fall short in their ability to properly control the engine with those mods. In that case, no matter how hard you try to dial in something like the VVE, it cannot be accurate. It's no reflection on you - the software just isn't capable of it.

One example I remember that is very similar to your VVE issue is the last engine I worked on before I left Chrysler. It had the ability to use variable valve actuation to enable exhaust scavenging on the fly, both for emissions and spooling the turbo. But scavenging literally drags unburned o2 out of the exhaust, so without the algorithm that was introduced to account for that additional air passing the O2 sensors and entering the catalyst, it would never pass emissions while scavenging was enabled. And the fuel trims would always be wrong. Same thing goes for your car. At some point, it's just out of your hands without being able to actually redesign the algorithms themselves.

[cmitchell17]

Thanks everyone for the input, I guess I am moving on from this one and giving up haha A lot of this is so "black magic like" just because you never can see it with you're own eyes and to your intuition, it seems in no way logical or plausible.

It's just as frustrating as trying to make sense of the stock GM calibrations, like wondering why no one could make up their mind weather less or more timing during PE was better or worse, or wondering why anyone would make a lookup table with huge jumps and discontinuities and think these same jumps and ragged edges exist in real life.

I remember someone saying that the calibrations are generated from algorithms? I would think for a fact at least one human puts their eyes on all parts of the calibration, but I guess if the car passes all emissions and federal regulations, then no one cares to change it even though it might look like crap, I mean I get it if there's no more budget the engineer will get yelled at trying to make it better.