Mustang and Ford Performance Forums banner

AFR variance between banks. Need opinions/help.

18K views 55 replies 3 participants last post by  Thunderbolt287 
#1 · (Edited)
Vehicle is a 2000 F150 4x4 with a Lightning conversion. Fully built 2V 5.4 with 3.4 whipple. 800 miles on my current build.

I am running into an AFR variance between banks of approx .5 (OPEN LOOP ONLY)

Bank 2 is .5-.6 leaner than bank 1 during part throttle open loop acceleration.

I am running an AEM failsafe on each bank. Yeah, I know... OCD

Fuel system is parallel flow. -8an feed to a y fitting, -8an to each rail, -8an from each rail to a y, -8an to the regulator, -6AN from regulator to fuel hat. Dual Walbro 465 pumps, one is on a Hobbs at 10# boost. ID1300 injectors. Fore F2i regulator. Currently running 93

* Had Injector dynamics clean my injectors

*Staggered injectors based off cfm to balance banks. Bank 2 avg flow is 1309 cfm, bank 1 avg flow is 1307.5 cfm

* Changed spark plugs (br7ef)

*Wideband sensors are just behind the longtube v-bands

* Verified compression (1/2 psi difference between each bank average)

* Swapped wideband sensors between banks

* Swapped wideband gauges between banks

* Replumbed my fuel system and Incorporated a y fitting with equal length hose to & from each rail. Fuel rails are plumbed in parallel.

*Cams were degreed using cam events, not the intake centerline method. Cloyes hex adjust gears.

*Short term fuel trims are perfect


Any ideas are appreciated!!!!!!!
This is driving me nuts...
 
See less See more
#2 ·
Before I finished reading your post I was going to suggest a potential difference in cam phasing bank to bank Matt but with your compression numbers are as close as they are that eliminates cam phasing. I am thinking that you might not have a real bank to bank AFR difference, it might be a perceived difference being created by a whoopse in the exhaust plumbing.

Because you observed the difference persisting after you swapped the wideband sensors bank to bank I am suspicious of a small air leak somewhere in the exhaust plumbing. Both sensors showing the same AFR difference on the same bunk makes me think you have a pinhole sized air leak somewhere in the exhaust plumbing.

At less than WOT a pinhole sized leak can allow small amounts of air to enter the exhaust system that the wideband will read as lean when you really are not. When the engine is under load the higher sustained pressure in the exhaust forces exhaust gasses out the leak so the apparent discrepancy disappears. Have you tried smoking it or otherwise leak checking it?


Ed
 
#3 ·
I have not smoked the exhaust yet. Shortly after posting this thread I began thinking about the exhaust side of the equation. My upstream and wideband oxygen sensors are mounted very close together with a partial overlap. I am thinking that the wideband is receiving dirty airflow caused by the upstream sensor.

I am using v-band flanges so I suppose there could easily be a small air leak at lower rpms.

Thanks for such a fast reply Ed,

Matt
 
#7 ·
Just made gaskets for my front v-bands. Used mr gasket part #5971 and cut them into rings. They come with 3" id

Taking it out in the am, hoping this nips it. I could see carbon along the entire inner edge of the v-band clamps. They were definitely leaking.

I bought two different gaskets. Ended up using the metal core graphite gaskets rather than the solid fiber gaskets as picture.
 
#10 · (Edited)
Hmmm, I'm starting to run out of ideas also, Matt.

This is a diagram with the four common plumbing models for a return style system. The manifold reference vacuum line is missing in all four illustrations.

Rectangle Line Font Parallel Technology


From your comments I believe the version you are using is version D. Normally I would expect the FPR to equally pressurize both rails with this configuration. If it did not at part throttle then I would suspect some type of flow difference from entry port to entry port on the FPR at low flow rates.

The model I prefer is Version B with a small balance tube (-3AN or -4AN) across the end of each fuel rail to normalize the fuel pressure pulses at the ends of the rails. Each time an injector fires it is an abrupt opening and closing event that produces a shot of fuel for the particular cylinder and pressure pulses that echo up and down the rails. The balance tube at the end of the rails mutes but does not eliminate the waves. I don't think the pressure waves however, are the source of your AFR differences. The commentary was more a way of explaining my reason for using the balance tube.

The benefit of the version B plumbing is that should the FPR exhibit low flow differences port to port they will be invisible to the injectors because only one port is fueling both rails. The other port is being used to receive fuel from the pump and the -6 is being used to return excess fuel. All injectors will see the same fuel supply attributes with this plumbing model.

I know replumbing is a PITA so my suggestion would be to use the replumbing option as a last resort assuming another fix couldn't be found.

Ed
 
#13 · (Edited)
Change of heart here, Matt. I had forgotten all about that thread!

The balance (aka crossover) line at the front of the fuel rails solved the problem for Cammed03Whipple, I think his system was the Version B above. The pressure waves propagate up and down the fuel rail in a sin wave form that looks like this;

Rectangle Slope Font Parallel Plot


A horizontal line (missing) through the center of the pic would correspond to the base fuel system pressure. The trough in the pic corresponds to the reduced fuel rail pressure at the firing of an injector. The peak in the pic corresponds to the increased fuel rail pressure when the injector pintle slams shut. Each injector creates one of these waves each time it fires and they continuously propagate up and down the fuel rails while the engine runs. If you are unlucky enough a trough arrives at an injector just as it opens and the fuel delivery is off by a whisker - say 0.5 AFR?

The balance tube across the front of the rails does not eliminate the pulses but it does mitigate them. When Cammed03Whipple added his line, his lean condition (which was much worse than yours) disappeared. I suspect if you plumb like Version B and add a balance tube to the ends of the fuel rails your lean condition might disappear also - that's the good news. The bad news is plumbing can be a PITA but it might just be worth a shot. The poor fueling is not good for the engine or for your satisfaction driving the car because it is occurring in an rpm range you are likely to spend more time driving in.

Ed
 
#14 · (Edited)
Ed,
I don't mind the plumbing change. I really won't be that bad based on how my system is currently configured. Probably take me an hour to make the change. I already ordered the hose and fittings. Really hoping this nips it.

I am running an -8an balance tube so I don't have to replace my 180* fittings.

I assume the larger balance tube will be OK?

Thanks!
Matt
 
#15 ·
The larger tube will work fine, Matt. The smaller -4 was size makes it easier to"hide" the tube routing low in the front of the engine. Summit has -8 to -4 straight male adapters that are pretty inexpensive => -4AN to -8AN Adapter and it will make the front of the engine look much better.

I forgot, your engine is turbocharged. You can more easily position the larger -8 line and still have a good underhood appearance. Forget my comments about the smaller diameter balance line...

Ed
 
#18 · (Edited)
Looking at the diagram for version B. If a balance hose was added to the deadhead of the rails wouldn't air get trapped between banks?

I re-plumbed my rails to version B with a crossover. Bank 2 is leaner by .5 AFR until around 580 maf counts, then the banks switch and bank 1 is lean by .5 AFR throughout wot.

Starting to question my AEM wideband o2 sensors.
 
#20 · (Edited)
No more so than a pair of rails w/o the crossover, Matt. At first start there will be an air bubble for a very brief period of time. The start up process will flush it out of the fuel lines and it will not appear again, unless the system is disassembled - or has a leak.

It is possible the AEM Wideband sensors are in the latter stages of their life cycle and beginning to misbehave. If they are when you source a replacement sensor I would give the NGK sensors a look. NGK's exhibit a different behavior as they approach end of life. Instead of giving close but progressively less correct readings or behaving like the AEM sensors are possibly currently doing (assuming no real AFR change is taking place), the NGK sensors just shut down. They go completely dead.

I first discovered this working with AJ (Mofasta) on his engine. He used NGK sensors specifically for this reason. Rather than chasing a drifting sensor in it's later stages of life the NGK's just shut down completely. Another nice feature they have is the ability to accurately read down into the 0.5 to 0.6 lambda range which was important for him because he used methanol as a fuel.

If you have good reason to believe the sensors are at the end of their lives then I would consider the NGK's as an alternative. The other thing to keep in mind is, I believe, you said this was happening in open loop. Did that mean that the AFR drift did not happen in closed loop? If that is the case, unless you have locked the ECU in open loop, does that mean the normal operation of the engine in closed loop eliminates the 0.5 AFR drift?

If the only time the engine is seeing open loop is at start up and after warming goes to closed loop with stable AFR's you could just be chasing AFR ghosts that once warmed up disappear from the engine's operation. I know the 4.6L Terminators use an FRPS in the fuel rails to measure delta pressure across the injectors to allow the ECU to calculate the proper injector pulse width to deliver the commanded AFR. I don't know with certainty if the blown 5.4's do or not. If they do and you have disabled it in the tune or disconnected it, then the ECU can not see the pressure variations in the fuel rail and does a best guess pulse width calculation based on base system fuel pressure that was programmed into the tune.

Because of the time delay and initial boost reference signal strength attenuation from the manifold to the regulator through the boost reference line, the actual fuel rail pressure to vary from the steady state base pressure the FPR was set to. The use of the FRPS eliminates this variable because it monitors the delta pressure in the rail in real time and delivers the information at electronic speeds to the ECU where as the FPR is reading manifold pressure changes at the much slower signal propagation speeds and attenuated signal strength present in a boost reference line.

If the FRPS is still present and you allow the ECU to go to closed loop the AFR perturbations introduced by the plumbing should disappear and you should see two essentially similar AFR readings for both banks. Is this what happens if you allow it to go to closed loop?

Ed
 
#19 · (Edited)
Orange line is bank 1, green bank 2


Looking at my logs I see a trend with bank 1. Around 500 maf counts it begins to go rich, around 630 maf counts it begins to lean out. Lines cross at 11.35afr

I bypassed my second fuel pump Hobbs switch and compared logs to rule it out.

Starting to wonder if I should replace my wideband sensors...
 
#21 ·
Ed,

You are correct. Closed loop I have no issues with afr drift. The issue is only present during open loop. My stft are spot on while driving around town. I seriously believe I've been chasing a ghost AFR signal or signals.

My pcm does not support fuel rail pressure sensors. The factory rails had a vacuum controlled fuel pressure regulator.

In regards to the NTK sensors... I've read online that they are not compatible with the AEM failsafe gauges. Apparently the configuration is incorrect? I wish these Bosch sensor would react the same as the NTK at the end of their life. My current sensors only have 800 miles of use "but" they were tuning miles which saw rich afr's and starts after heating them up for cold start logs. I understand that condensation will damage them if heated before firing the engine.

The sensors are UEGO 4.2 "not" 4.9

Do you know if there is any truck to the NTK incompatibility?

Matt
 
#22 ·
I am not sure if the NTK sensors will work with the AEM hardware either, Matt. My best bet on that would suggest a call to AEM although they might be less cooperative than hoped for. The next best check point would be the NTK guys because they would be motivated to know everything and everywhere their sensors could be deployed.

With a no drift outcome in closed loop the issue increasingly sounds like ghost sensor signals although it could be real variations that the ECU is correcting for once it goes closed loop. I would expect the STFT should show what is happening and if the ECU is correcting the AFR to bring it back to commanded. Your comment however about the STFT's being in tip top shape increasingly leads me to suspect some type of signal ghosting in happening in open loop.

The most important thing is the engine is properly fueled in closed loop which it seems to be. The open loop time is essentially limited to warm up so I wouldn't be concerned about engine damage occurring. The moment the engine is up to temp you are on closed loop and the fueling issue disappears so you know the engine is safe.

I might take some time to investigate the NTK sensor compatibility issue. BTW I just noticed I keep flip flopping back and forth between NTK and NGK naming conventions. FWIW they are the same sensor as far as I can tell. I think NTK is the more correct naming convention.

In the FWIW bucket I recall reading somewhere that the Bosch 4.9 sensors while an upgrade to their earlier 4.2 sensors also brought some corrections in sensor operation under certain conditions. I want to say they are interchangeable IIRC however, Bosch could know for sure.


Ed
 
#23 ·
Response from NGK/NTK

Hello,

*

Thank you for your E-mail.* I don’t have a cross for the those Bosch OE numbers.* If NTK has a sensor in aftermarket that is for a Bosch wideband application,* the sensor in the NTK box is going to be a Bosch.* Our NTK wideband sensor is more resistance to contamination, but you must use a Bosch in your application to work with the AFR unit.
 
#24 ·
Well, not exactly what we were hoping for but certainly definitive in its response.

If I were going to stick with the AEM hardware I think I would upgrade the sensors to the 4.9 units, Matt. They represent an improvement over the 4.2 generation units. Depending on costs I might consider replacing the existing AFR stuff with NTK compatible hardware if and when I decide to replace the 4.2 sensors.


Ed
 
#26 ·
It is worthwhile remembering that the open loop operation is essentially your warm up and a few other situations, Matt. At warm up a 0.5 difference between commanded and actual AFR is not a significant event. As long as the AFRs look the same under load in closed loop you are probably good to go. The difference in temp rise during the warm up phase is for all intents and purposes inconsequential. Once warmed up and under load it is a different story and that story, for your engine, sounds like it is OK.


Ed
 
#27 · (Edited)
What's killing me Ed is the fact that something changed a month ago, around 8-18.

Below are two afr logs.
Top log is from 8-8 and bottom log is from last night.

Orange FireWire is bank 2.

TPS counts were adjusted on 8-11. That is the only change that was done prior to the issue surfacing on 8-18.

 
#28 ·
Like yourself, I am running out of ideas as to what is causing the differences, Matt.

I would expect any changes in TPS counts to affect the measured AFR for both banks similarly. What is the X-Axis in the top pic and what is it in your lower pic — time perhaps? Your top log is quite tight bank to bank while the bottom log is reporting the 0.52 difference in measured AFR you had commented about earlier.

Have you checked the "Y" fitting that you use to split the fuel between the two rails for machining imperfections / inconsistencies leg to leg or burrs that may have been left behind in one of the legs after machining and not removed prior to sale? If you have the ability (two analog pressure sensors and 2 unassigned analog channels) it would be interesting to data log the fuel rail pressure in each fuel rail at the point in time that the AFR's diverge. If the pressures are similar the only other way the difference can occur is differing injector pulse widths being sent to the two different cylinder banks — which begs the pregnant question, why?

The first and easiest check point would be the "Y" fitting for uniformity leg to leg. Apologies, I wish I had more.


Ed
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top