Page 1 of 2 12 LastLast
Results 1 to 20 of 22

Thread: Fix required for AFR difference between C/L and WB

  1. #1
    Tuner in Training IACS_HSV's Avatar
    Join Date
    Feb 2008
    Location
    Gloucester, UK
    Posts
    13

    Question Fix required for AFR difference between C/L and WB

    I’ve recently got my LC-1 installed in addition to my stock NBs (which are still feeding the PCM) but I’m seeing a 0.6 difference between measured WB and Closed Loop Commanded AFR and need some expert advice on how to dial my actual (measured by WB) in to my commanded.

    I want my LC-1 range set between 10 and 18 AFR giving a voltage output of 2.95V (2.9375V) at 14.7. So I temporarily setup the LC-1 to output a constant 2.95V and after configuring the matching PID, I’ve logged the EIO input in a histogram so I could get a nice average but it’s not really needed since the LC-1/HPT combination is absolutely rock steady at 14.7 with the engine running or stopped. No offsets, ground loops or drifts – perfect, and a testament to the performance of both HPT and the often maligned LC-1 (many thanks to Bill, SoundEngineer and others for taking the time to provide quality advice!).

    Next I decide to make some real measurements, so I re-program my LC-1 to output between 10 to 18, and fire-up the LS1. I have my closed loop stoich set at 14.68 and once C/L enables, I can log 14.68 AFR Commanded all day while idling in the driveway.

    Problem is, my WB is showing me I’m getting an average 14.1 measured. Now, I know I’m trusting the WB to be totally accurate, but after a couple of careful free-air cals. it’s very repeatable and probably a lot better than the stock narrow bands.

    My assumption is that the stock narrowband switching points are out causing commanded to deliver slightly rich and the voltages need to be tweaked. I’ve played around with these by +/- 50mV without making a great deal of difference – am I barking up the wrong tree? Any help appreciated...

    Paul
    Holden HSV GTS

  2. #2
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    This has been a topic for debate for some time. See the "Perplexing PLX thread" and "Another wide band controller on the way" threads. Yours is the first that I've seen where it reads richer than 14.7 in closed loop. Many have seen the WB report leaner than 14.7 when in closed loop. Without using another wide band it's impossible to know whether it's the first wide band that is eradign the AFR incorrectly or the narrow bands are switching at the wrong place.

    So thats exactly what I did. My AEM was reading ~15.5:1 in closed loop. I ordered a JAW DIY WB controller and connected it to the same sensor my AEM was and it still read ~15.5:1 even after a calibration (which the AEM does not allow). So then I ordered another Bosch LSU 4.2 sensor. It came in the other day. When I get off my butt and install it I'll know whether my other WB sensor is bad or whether the narrow bands are switching at the wrong AFR due to maybe reloacting them in the collectors of the stainless steel headers. If it's the narrow bands the trouble is that changing the switchpoints on a gen 4 GM car has no effect in the current version of HPT. It works fine on gen 3's though.
    Last edited by 5_Liter_Eater; 03-03-2008 at 03:36 PM.
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  3. #3
    Mine showed up lean as well when going back the closed loop mode, I will be messing with the switchpoints someday once I get my VE and MAF dialed in.
    2002 Camaro Z28
    236/236/0.590"/0.590"/112 lsa cam
    Longtube Headers, True Dual w/ X Pipe
    Air Lid, Filter, Ported Maf End, Bellows Hose

  4. #4
    I'm no expert on this subject by any means, but after reading these posts it seems the concensus is that the issue lies with HPT, the wideband controller, or the wideband sensor. Is it feasible that the discrepancy may be with the OE narrowband O2's? Specifically, I haven't read of anybody installing new narrowband O2's to eliminate this as a potential source of the problem. I would think that OE sensors with 50,000 + miles on them may be a significant factor.

  5. #5
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    That's the thing, you have to spend some cash on either a new WB sensor and controller or new NB sensors. I've elected to do the former.

    http://www.hptuners.com/forum/showthread.php?t=16029

    I'm almost done proving there is nothing wrong with the WB setup, just need to move it to a known good bung now. If it's still lean then it is something "wrong" with the NB sensor(s) or just the fact that they are now further down the pipe installed in SS LT headers has altered their switchpoints.
    Last edited by 5_Liter_Eater; 03-06-2008 at 11:04 AM.
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  6. #6
    Tuner in Training IACS_HSV's Avatar
    Join Date
    Feb 2008
    Location
    Gloucester, UK
    Posts
    13
    Thanks Guys - apologies for missing the earlier posts 5LE. I’ve been playing with the NB switch points some more and I can clearly see an offset being introduced when I log the O2 mV, but this doesn’t seem to be introducing any real effects to the idle fuelling and the measured AFR appears to stay the same...

    Am I missing something obvious?

    Should I be thinking of tweaking my flow rates rather than messing with the switch points?

  7. #7
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  8. #8
    Tuner in Training IACS_HSV's Avatar
    Join Date
    Feb 2008
    Location
    Gloucester, UK
    Posts
    13
    Thanks 5LE, have just read your thread. I'm a bit puzzled why my mods to my switch points don't seem to be affecting my fuelling since I'm running a Gen3 in the HSV, not Gen4 as you are?

    I wouldn't get too worried about running 15.5. I'm outside the US so I've got Lean Cruise enabled. I've attached part of my log from my trip home tonight and you can see I'm up at ~16.5+ when cruising on the motorway.

    My log also shows the commanded vs. measured WB offset I'm trying to eliminate (~0.5 AFR rich).

    I know my VE is well dialled in - the HPT dashboard in my log shows minimal LTFTs after 7 minutes of pressing on...

  9. #9
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    I'm not terribly concerned about running lean in closed loop as long as I can know that my WB is accurate.

    I'd need to see a tune and a log to see whats going on. The gen 3's should respond very well to altering the switchpoints. Did you use my method/histogram?
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  10. #10
    Tuner in Training IACS_HSV's Avatar
    Join Date
    Feb 2008
    Location
    Gloucester, UK
    Posts
    13
    I didn't get too detailed at this time - I just wanted a quick test to see if I could introduce a step change in the fuelling so I simply "added" -200mV to B1 & B2 for all air flows.

  11. #11
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    I would think that would have made a difference. When I logged mine with that histogram I found that the average narrow band voltage was up over 750 mV when the WB showed ~14.65 and 14.75.
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  12. #12
    Супер Модератор EC_Tune's Avatar
    Join Date
    Apr 2003
    Location
    Almost 2000 feet.
    Posts
    7,876
    Here's another wrench to throw into this mess: Have you checked your highway mileage in closed loop compared to open loop with the wideband?
    If fuel economy is better with the NB's active (Closed loop) you were too rich with the OL cal and your wideband is suspect.

    If at freeway cruise speeds your narrowband doesn't agree with the wideband, the wideband is suspect because at freeway speeds the NB's are being well heated by the exhaust system, effects of exhaust leaks are minimized, and the NB's are in their "natural habitat". All inexpensive widebands are suspect under these conditions if they don't agree with the NB's. The NB's are a switch and they go off to on at 14.7 AFR with gasoline.
    Always Support Our Troops!

  13. #13
    I don't have access to high quality analyser to prove my theory however I am incredibly confident that the calibre of widebands we are using lose accuracy at AFRs much leaner than 13.5 ish AFR (lambda ~0.9 ish). I believe it is the sensor more so than the controller also. You can buy lean burn sensors but I dont see the need.

    Here is something to try, in fact it is how I tune fueling (VE/MAF) as a general rule of thumb. Disable CL and set commanded AFR to 13.0 (best way is to temporarily set OLFA to 1.13) and dial the AFR Error % as close as you care. Then go and re-enable CL (AFR 14.6ish) and I'll nearly bet a testicle that the STFTs are very close to zero if your AFR Error % previously was close (less than 2%). If you then have the ability to log WB AFR at the same time as CL NB AFR I'll bet that the AFR Error that you verified at 13 AFR to be close to zero is no longer but the NBs are telling you that you are close to stoich.

    I am aware that changing the AFR can affect the completeness of the burn etc and therefore alter the "VE" or "airflow" you are trying to model which is why I then verify non-PE areas with NBs prior to finishing the tune whether it be a full OL tune, CL or hybrid including Lean Cruise. The further lean you go say 16.0 or leaner the typical WB sensor simply wont hack it in my findings and you'll end up will a falsely rich tune. The exception is very large cams at low airflows are difficult to get NBs to provide truly accurate data.

    In essence I am of the opinion that the few widebands that I have used are quite accurate at 11-13 AFR and not too bad at stoich (but still not as good as we need for tuning) and completely suck at anything leaner than stoich.

    Narrowbands are quite accurate and useful under certain circumstances, don't always dismiss their accuracy. All that said sometimes they are not too.

    Took me a long time to realise the strengths and weaknesses of methods of measuring fueling. This may not be true for all setups although it has worked well for me now in quite a few different cars and worth trying if for no other reason than to disprove the theory.

    Cheers

  14. #14
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    Doug, I'm not sure your theory holds true, or maybe it does but it's biased toward faith in the NB's. Of course my mileage will be less in OL because I tuned my airflow tables in with the WB, so it will read 14.7 instead of 15.5 or greater like it does in closed loop.

    The LSU 4.2 sensors are definitely less accurate the leaner the AFR. In the JAW Edit program (used to program the JAW DIY WB) the table you see to plug in different AFRs per voltage step is not linear. I asked the programmer about this and he said it was intentional because the 4.2 sensors are less and less accurate the leaner the AFR, but it's still pretty accurate around stoich.

    I had to think about it for a while SS but I understand your thinking about setting the OLFA to 13:1. Since the WB is more accurate the fatter the AFR, then dial in the airlow table(s) there.

    I tested my WB bung and there are a few pinholes in the bung. Not sure how much they would affect the WB reading but I've begun the teardown to put my blower cam and Paxton kit in so the car will be down for a while. I'm honestly a little curious to see if it still reads lean in CL with the new cam since it will be a negative overlap cam.
    Last edited by 5_Liter_Eater; 03-09-2008 at 10:27 AM.
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  15. #15
    Senior Tuner Russ K's Avatar
    Join Date
    Dec 2005
    Location
    Regina, Sask.
    Posts
    4,213
    Any exhaust leaks upstream/near the wideband will show as a lean AFR.

    Russ Kemp

  16. #16
    Супер Модератор EC_Tune's Avatar
    Join Date
    Apr 2003
    Location
    Almost 2000 feet.
    Posts
    7,876
    I don't know if many of you know that the AFR sensor used by the AEM, DynoJet, and AutoMeter widebands are used on the Cadillac SRX as a front o2 sensor. If there were issues with sensor drift GM would not use it. But then again they can calibrate things much more accurately than we can as they have the tools, materials, laboratories, and cubic $$$ to do so.

    The DynoJet wideband uses the same sensor as the above and doesn't have the AFR drift issue that you guys are reporting. So I'm inclined to think that it is a wideband MFG based issue and not an actual issue with the sensor or at worst it is an air leak issue.

    I'm more interested in accuracy than brand name when it comes to measurement tools so if there is an issue with a tool, it's time for a different tool.

    In reality I would like to see a real test of inexpensive widebands done with lab grade equipment (Horiba/ECM and the like) then we can put all of this to rest. Unfortunately I don't have access to the lab grade stuff and my wallet won't support buying it (Horiba is 20g's!). So until we get a real evaluation, all we have is empirical evidence and speculation. Not to sound harsh but I think that is the reality.

    However maligned, the NB's truely are accurate at stoch. I had a document at one time (lost it in a HD crash) where GM actually doped some sensors to read slightly richer or leaner than stoch (+/- .3 AFR) as an emissions research tool. They did exactly as they were doped.

    If there is any air entering the exhaust air from outside sources (AIR injection systems can be suspect even with the flapper valves intact) it will bias the WB and NB sensors.
    Always Support Our Troops!

  17. #17
    Tuner in Training IACS_HSV's Avatar
    Join Date
    Feb 2008
    Location
    Gloucester, UK
    Posts
    13

    Smile

    Boys,

    I've discovered the reason why my changes to the NBO2 switching points were having no effect...

    I was still in the "learning period" following a reset on my fuel trims. Although my LTFTs had shifted following the change to the switch points, my measured AFR remained the same 'cos the PCM wasn't using the trims to calculate fuelling!

    Following learning period expiry, the measured suddenly jumped as the fuelling began to dial-in on the commanded.

    So, some operational experience - if you are tweaking your NB switch points to align with your wideband, make sure the PCM is actually using the trims!

    5LE - my cunning plan now is to reset the switch points back to stock and try your method of slewing the switch points vs airflow mode (only crude mV step change used across all modes so-far).

    Many thanks for all suggestions - will post up results when finished.
    Last edited by IACS_HSV; 03-24-2008 at 04:03 PM.

  18. #18
    Senior Tuner 5_Liter_Eater's Avatar
    Join Date
    Nov 2005
    Location
    Houston, TX
    Posts
    3,968
    Really? Thats it? I have to say I'm a little skeptical. I will have to try this out when I get my car back together.
    Bill Winters

    Former owner/builder/tuner of the FarmVette
    Out of the LSx tuning game

  19. #19
    Супер Модератор EC_Tune's Avatar
    Join Date
    Apr 2003
    Location
    Almost 2000 feet.
    Posts
    7,876
    Remember the switch points are for the "running average" of the o2 mVdc to generate the STFT's which then alters the LTFT's.

    If you have the LTFT's turned off and STFT's tuned, make an o2 crossings change, then test the change, the STFT's should move in the direction of the o2 crossing change. If you made the crossings lower the "average STFT" should go negative and vice versus. Make sure that the air temp is the same on both tests because as we know that will bias the AFR around if it is left stock and untuned.

    As always start with a known baseline and work outward from that. Otherwise it becomes an exercise in "chicken and egg".
    Always Support Our Troops!

  20. #20
    Quote Originally Posted by IACS_HSV
    Boys,

    I was still in the "learning period" following a reset on my fuel trims. Although my LTFTs had shifted following the change to the switch points, my measured AFR remained the same 'cos the PCM wasn't using the trims to calculate fuelling!

    Following learning period expiry, the measured suddenly jumped as the fuelling began to dial-in on the commanded.

    So, some operational experience - if you are tweaking your NB switch points to align with your wideband, make sure the PCM is actually using the trims!
    Few questions here...
    What do you mean as Learning Period (STFT havent lined out yet, or time)
    If your LTFT changed and in closed loop, how does the PCM not used trims for fueling, Isnt that the point of CL. Just trying to get the grasp on this, cause I am about at the point of tweeking the switchpoints
    2002 Camaro Z28
    236/236/0.590"/0.590"/112 lsa cam
    Longtube Headers, True Dual w/ X Pipe
    Air Lid, Filter, Ported Maf End, Bellows Hose