Wideband Reading Lean Despite Narrowbands OK ? Plus Intermittent Knock Retard

[Justacanuck]

Hi everyone,

I'm chasing two issues with my 5.3L Avalanche and could use some experienced eyes on this. Tune and log files are attached. Any help is appreciated.

Build Details:
Vehicle: 2009 Chevy Avalanche, 5.3L LMG
Cam: BTR Truck Norris, LS7 lifters, beehive springs
Intake/MAF: Stock intake, stock MAF
Exhaust: Texas Speed 1-3/4" long tube headers, full 3" dual Exhaust
Narrowbands: Factory O2 sensors mounted in 3" collectors
Wideband: AEM 4110 (MPVI2.1 > Analog input), installed post-collector just before X-pipe (driver side only)
ECM: E38, using HP Tuners v5.0.4

Programming/Tuning Notes:
LTFTs disabled
STFTs active in closed loop
Tuning VE using wideband above ~3000 RPM; idle and cruise zones still under review
PE, spark, and airflow tables gradually being refined

Issue #1: Wideband vs Narrowbands
At idle, wideband reads ~17.7:1 while commanded = 14.17 (shows ~25% lean)
At cruise (e.g., 1800-2500 RPM, 35?50 kPa), wideband shows ~15.5-15.8:1 (8?10% lean)
Narrowbands are switching actively and STFTs typically hover near zero
AFR error histogram confirms persistent lean WB readings in cruise/idle range
VE table is fairly dialed in above 3000 RPM (WB and EQ ratio agree well there)

Suspicions:
Possibly scavenging or reversion from the Truck Norris cam + long tubes causing wideband dilution?
Post-collector WB location may still be affected by low exhaust velocity?
Considering increasing O2 sensor delay (12470, 12476, 12477) and wrapping headers to retain exhaust heat

Issue #2: Intermittent Knock Retard Under Light Load
Seeing 4?7.9? of total knock retard at times during part throttle (~2400 RPM, 34?50 kPa MAP, ~15? spark advance)
Occurs at low throttle (30-50% TPS), not WOT
KR events are brief but pronounced, enough to disrupt part-throttle performance.

Suspicions:
Potential false knock due to exhaust resonance, or sensor sensitivity?
Fuel quality is 91 octane (~5-10% ethanol content)
No audible "pinging" or real knock as far as I can tell, except for a little "bark" on tip-in (see log/screenshot) which I assume is from spark retard knock reduction? No drivability issues otherwise

❓Questions for the Group:
Should I trust the wideband at idle/cruise or assume false lean from overlap?
Should I apply VE corrections based on wideband in the cruise zones or defer to NB + STFT?
Are O2 voltage switch points worth adjusting to improve STFT/NB response?
Could the knock retard be false? Should I adjust the knock sensor gain/filtering or the spark in that range?
Would wrapping the header or relocating the WB improve signal stability at idle or cruise?

I'm trying to preserve drivability while dialling in the tune correctly. The part-throttle KR and conflicting AFR signals make it tough to trust corrections. Any insights or known-good strategies for cammed long-tube setups would be much appreciated.

Attached:
Tune File (.hpt)
Idle + Cruise Log (including knock events and AFR error)
screenshoots

Thanks in advance!

2009_SANE_Avalanche_DFCO_OFF-LTFT-OFF_O2-ON_PE-Enabled_MAF-ON._CameroEOIT_VEandMAF-Cruising.hptkr.PNG25-06-10 10-30-03.hplknock.PNG

[Justacanuck]

The injectors were cleaned and very basic flow test but I'm also considering the possibility that I have some slow/lazy injectors so I've ordered a "matched set" of the same factory injectors. The O2 sensors are brand new.

[biholliday]

i have the same wide band and ive noticed that there are several options to choose from in the scanner wide band set up where if you choose the 30-0300-0600 it will give results that are slightly richer and appear more in line with narrow band reading due to voltage offset. I would just ignore the wideband and stick with narrow band and the proper filter using dynamic air to adjust your vve and maf tables. Go to YT and look up silversurfer77 and watch his method of tuning. He also has free vve and maf assistant you will download from his drive. This is the best way to tune in my opinion. He is Cringer on this forum and his avatar is a green tiger with yellow stripes but we all know it should be a GOAT

[Justacanuck]

thanks biholliday. I've read many threads from Cringer and watched most all of his videos too. I used his MAF and VE tuning guide , formulas and MATH/filters to get fueling setup appropriately. I've been ignoring the WB at low and mid throttle for the most part, but the persistent lean reading, even at cruising speeds, as well as the KR with small amounts of throttle tip-in has me worried.

[Justacanuck]

There has to be something I'm missing....

[Justacanuck]

I removed the headers to apply heat wrap; the gaskets show signs of multiple exhaust leaks. I'll update again if this solves my problems.

[biholliday]

when the scanner shows knock i just copy that table and then go to the editor and subtract from the high octane table and then smooth it out. Take it for another drive and repeat until no knock. Once there is no knock I can add octane booster and then add 4 degrees to the entire table and get 0 knock which is actually noticeable gain in power. then after i burn 1/2 tank I return it to the 91octane tune

[04silverado6.0]

The preset hpt wideband options are not 100% perfect. I noticed on at least 2 they skimped on the rounding and my math was a good amount different. Produce your own transform function using the data given by the wb manufacturer.

[gametech]

On the innovate widebands you can program what you want the voltage output to be during warm up when they are not giving an actual reading. Using a multimeter you can test what voltage you are sending to the scanner compared to what the wideband is supposed to be putting out. From there it is easy to see if you have an offset from your expected voltage and account for it in your transform. If you have a wideband with a serial output, you can also use your computer and test the analog out with a multimeter compared to what the serial output shows.

[Alvin]

Take a video of your wideband gauge face and the the scanner while logging. pause the video in a few spots. Make sure the two match.

Also, I rarely/next to never desensitize the knock sensors on these. It is likely real knock. Pull timing in affected areas.

[04silverado6.0]

Take a video of your wideband gauge face and the the scanner while logging. pause the video in a few spots. Make sure the two match.

Also, I rarely/next to never desensitize the knock sensors on these. It is likely real knock. Pull timing in affected areas.

I have came across a few with false knock but a lot of times they are giving audible knock and not pulling enough advance. Most are fine as is. I can say ive seen more get into audible knock without attacking it aggressively enough vs the opposite. E38 on cars with headers need the sensors desensitized 100% of the time.

[Alvin]

No they do not.

[Justacanuck]

Thanks for all the advice and ideas guys.

I messaged the flanges of the headers with a flat file. I finished wrapping the headers and re-installed with significantly thicker Felpro gaskets. This seems to have made a significant effect. The WB is much closer to 14.7 at light crusing ranges. I ran a few logs with very modest (stock Avalanche) timing tables and it is behaving MUCH better. I'm going to do more testing, probably modify the MATH to compensate for WB offset, and carefully re-introduce some timing again (stock timing table seems very conservative)

[04silverado6.0]

No they do not.

Cars, g8, 5th gen camaro, c6. Never had one with headers that didnt have the 0.5*-1* kr any time its making noise. Pull 2* still have 0.5*-1*, add 2* still have the same. Gto as well.