Tuning cars is BULLSHIT

[Gh0st]

Cars runs PERFECT and error % is within 0 to 5% MAX on both my VE tune and my MAF tune when I run them alone.

Put the two together and BAM, cars runs like ****! Idle dips, car surges, fuel trims are apeshit.

****

[ringram]

OLSD and OLMAF run ok? or what?

If so perhaps you O2 sensors are shot if you mean CLOEM config is crap..

[Gh0st]

You got it dead on my man. OLSD and OLMAF run perfect. I'm in agreement with you that it's most likely my narrowband 02 sensors that are doodoo. I've noticed once my car goes in to closed loop it runs like ass and I do get some weird ass narrowband data when data logging.

My drive to work was very pleasant though. I only live 1 mile away but the car ran great for the most part, not sure if I even reached closed loop temp by the time I got here. It still surged but I think that's an idle air issue more than anything at the moment.

[eficalibrator]

It's also possible that:

1) Your wideband is off by either ground-loop issues, an aged or uncalibrated sensor, or other problem

2) You may have tuned around some other issue in your OLSD and OLMAF tuning exercises that "baked in" errors to the VE or MAF surfaces. This is common when changing injectors, but not getting the short pulse adder or offsets nailed beforehand. Even though both OL airpath models seem to work fine independently, they can each have their own "baked in" errors that don't jive when blending between the two. Your "idle air issue" leads me to believe this more than #1, since idle RAF depends heavily upon getting the airpath models right.

[Gh0st]

Thanks Greg,

My idle air issue is that I just haven't tuned it yet, I'm not having a real "issue".

[Patrick G]

I agree with Greg. Your narrow band O2 sensors could be gone. For what it's worth, if I'm going to be using the NBO2 sensors for closed loop operation, then I always tune my VE tables with them and not a WBO2 sensor. The narrow bands are MUCH more accurate than the widebands for VE tuning. Besides, if your pcm is using them to trim fuel closed loop, wouldn't it make sense to use them for the tune?

[Gh0st]

I agree with Greg. Your narrow band O2 sensors could be gone. For what it's worth, if I'm going to be using the NBO2 sensors for closed loop operation, then I always tune my VE tables with them and not a WBO2 sensor. The narrow bands are MUCH more accurate than the widebands for VE tuning. Besides, if your pcm is using them to trim fuel closed loop, wouldn't it make sense to use them for the tune?

How do you figure that a narrowband is more accurate than a wideband for VE tuning? Are you saying once I've done VE with wideband, I should then work with STFT's to finalize my tune?

[ringram]

This is common when changing injectors, but not getting the short pulse adder or offsets nailed beforehand.

How do you get these right? Does it mean asking the manufacturer? Or can they be calculated somehow?

[miami993c297]

The narrow bands are MUCH more accurate than the widebands for VE tuning.

Oups Patrick,

Could I say that's an incomplete information...depending of your goals.

The NB Sensors are accurate only around ~14.7 AFR, and that's it.
If you chase for 14.7 AFR all over your VE table it may be a good tool, not always reacting as quickly as a WB sensor nonetheless!

If you tune a big "overlaped" camshaft in OL and chase for a Pleased Engine and Drivability/Streetability on the complete VE Table your AFR will vary a big bit, let's say from ~17 AFR around idle to 15.5/16.5 AFR at cruise and somewhere in the 12 AFR on Throttle opening or in the Low Revs Low Torque situation (which is plenty of false reading that require interpretation)!

I never found easy to use the NB Sensors for the VE Table purpose with a consequent camshaft and good drivability/streetability as a result.
Might learn something...

Christian

[eficalibrator]

The narrow bands are MUCH more accurate than the widebands for VE tuning.
Perhaps more accurate than some of the el cheapo widebands (like the Innovate), but not true if you're using good equipment. Remember that these tables are typically built from open loop engine dyno data collected using an ECM wideband. Strangely enough, I use an ECM LambdaPRO for my own projects too. ;) Not all "wideband" sensors or controllers are created equal!
How do you get these right? Does it mean asking the manufacturer? Or can they be calculated somehow?
They are typically supplied by the injector manufacturer at the OEM level. But Bosch/Siemens/Delphi/Lucas/Denso are unlikely to give the public that level of data. Your best bet is to copy the same data from another OEM application that uses the same injector. I have done the work to convert from Ford to GM style units on all the popular injectors that were publically posted on the Ford Racing site. I will eventually publish my own "GM converted values" on my website during the next wholesale refresh of it.

[ringram]

Widebands are 0-5v over say a range of 10 AFR maybe more thats 1V per 2 AFR ratio's.
Narrow goes 0-1v over a very tight range, say 0.5 AFR

Problem is FI setups etc that need to be richer, a narrowband isnt ideal for WOT either. But you can probably do 80% of your VE with narrow quite accurately.

[eficalibrator]

A couple interesting points here:

1) widebands are actually AFR vs pump current in mA not volts. It takes a controller to condition and translate the sensor's raw output into something that outboard dataloggers and ECUs understand like voltage. Many widebands are 0-1v linear, but 0-5v linear is more common. Some other ones (ETAS) aren't even linear in the volts vs AFR output. Moral of the story is: Understand exactly what the controller's output is supposed to look like before believing the data!

2) Narrowbands by definition must cross some threshold and register a change before indicating a valid "switch" to the ECU. As such, you end up with a deadband right around stoich that the narrowband has the ECU toggle across. This is why you rarely see trims any less than +/-2% even on completely stock vehicles. This deadband represents the most accurate you could ever hope to be in closed loop tuning with a narrowband. Good widebands are accurate to +/-0.1%, or 20 times more precise!

3) Even with a naturally aspirated engine, the narrowband is useless at WOT. They are only indicators of "somewhat rich" or "somewhat lean", but never really an exact measurement of how far at any given moment. (It takes an ECU correction and subsequent measurements to determine % error) So both 13.9:1 and 10.0:1 AFR look "rich" to a narrowband sensor. Temperature will also greatly skew this reading.

These are all topics covered in my book and training classes. :icon_bigg

[Patrick G]

The simple fact is, if your PCM is using the NB sensors for closed loop operation, it makes sense to use them when VE tuning...otherwise, you'll be chasing your tail. I have found good NB O2 sensors like the Bosch 13111s to be very accurate when trying to lock in a 14.7:1 A/F ratio. Also, if you're trying to vary your A/F ratio (like lean for decel and cruise and fatter for power), you can always adjust your O2 switchpoints.

My statement (about accuracy) applied to common WB O2 sensors. There's no way a cheapo WB sensor is going to be as accurate getting locked in at 14.7 as a good NB sensor. I have gotten superb results doing VE tuning using the NB sensors for all loads at and under 100 Kpa. Once you get into boost, the WB sensor gets the nod.

[wkmcd]

The simple fact is, if your PCM is using the NB sensors for closed loop operation, it makes sense to use them when VE tuning...otherwise, you'll be chasing your tail. I have found good NB O2 sensors like the Bosch 13111s to be very accurate when trying to lock in a 14.7:1 A/F ratio. Also, if you're trying to vary your A/F ratio (like lean for decel and cruise and fatter for power), you can always adjust your O2 switchpoints.

My statement (about accuracy) applied to common WB O2 sensors. There's no way a cheapo WB sensor is going to be as accurate getting locked in at 14.7 as a good NB sensor. I have gotten superb results doing VE tuning using the NB sensors for all loads at and under 100 Kpa. Once you get into boost, the WB sensor gets the nod.

Interesting. Ed ended up putting my car back in OL MAF because he wasn't happy with the switching speed of the new stock NB. We installed a PLX WB and sim a NB for the PCM but the passenger side stock 02 wasn't up to the task. I may throw another PLX on the passenger side but the car runs pretty damn good now so I really don't know the merit of doind it.

[Patrick G]

Interesting. Ed ended up putting my car back in OL MAF because he wasn't happy with the switching speed of the new stock NB. We installed a PLX WB and sim a NB for the PCM but the passenger side stock 02 wasn't up to the task. I may throw another PLX on the passenger side but the car runs pretty damn good now so I really don't know the merit of doind it.Kevin, when I tune with the NB sensors, I typically have to get them nice and hot...especially if the vehicle has long tube headers and no cats. Cold NB sensors (or any O2 sensor) are pretty inaccurate. Once I have the VE tables tuned to perfection, I typically leave the long term fuel trims turned off. Otherwise short trips with cold O2 sensors will submit bogus info to your L-Trim cache and degrade your tune over time. This does not happen when operating off an accurate VE table and short term fuel trim corrections only.

[Gh0st]

Patrick,

Can you confirm that you are indeed using STFT to tune via narrowbands?

[Patrick G]

Patrick,

Can you confirm that you are indeed using STFT to tune via narrowbands?It's not my only method, but I've had success using the NB sensors for doing VE tuning via STFTs. Whether it's the method I use depends on the application and total mods.

[redhardsupra]

Ghost, have you tuned MAF and VE completely separately or used one set of data for both?

[Gh0st]

Ghost, have you tuned MAF and VE completely separately or used one set of data for both?

I have a tune that is VE only and then I have a tune that is for tuning the MAF with VE lookup disabled. Both tunes have the same paremeters.

I run VE only, car runs perfect.
I run MAF tune(with no VE lookup), car runs perfect.

Put the two together in to a finalized tune with DFCO, fuel trims, and all the standard stuff and it sucks. I get surging. I didn't get surging in my VE or MAF tune at all, car ran like a dream. I don't get it.

To further elaborate. I spent three hours tuning OLSD for my VE. I then spent one more hour tuning my MAF and it only had about 1-3% error from the last time I tuned the MAF tables.

[Patrick G]

Have you logged to see if your dynamic airflow matches your MAF airflow at all different load cells and rpm? Because if your closed loop speed density airflow matches your MAF airlfow, you should have perfect (or close to it) street manners. If it does not match, then you need to rescale your MAF so they do.

[ringram]

Do your narrow bands agree with your wideband?
Like Greg says ground offsets and dodgy sensors can cause bad wideband data. Mine went very sick at the end of its life and gave me totally bad info.

[Gh0st]

I'm going to try something tonight. I'm going to disable VE lookup and have the car only reference the MAF tables. This will help me narrow down the issue.

[Goatkart]

The simple fact is, if your PCM is using the NB sensors for closed loop operation, it makes sense to use them when VE tuning...otherwise, you'll be chasing your tail. I have found good NB O2 sensors like the Bosch 13111s to be very accurate when trying to lock in a 14.7:1 A/F ratio. Also, if you're trying to vary your A/F ratio (like lean for decel and cruise and fatter for power), you can always adjust your O2 switchpoints.

My statement (about accuracy) applied to common WB O2 sensors. There's no way a cheapo WB sensor is going to be as accurate getting locked in at 14.7 as a good NB sensor. I have gotten superb results doing VE tuning using the NB sensors for all loads at and under 100 Kpa. Once you get into boost, the WB sensor gets the nod.


thats something i've been fighting over for some time patrick....

what ever accuracy a WB can bring to the table in regards to true AFR.. if the narrowbands dont agree when they are enabled, they are the ruling factor.... the computer doesnt care what the WB say's is going on... it cares what the NB say is going on...


its for this very specific reason, i struggled when deciding to turn closed loop back on..... Open Loop MAF seemed so much nicer....


the only reason i went back into closed loop was i was unsure of typical "daily variance" so without having time to check day to day over various conditions i didnt want to leave it open....

i think its the old problem of resolution..... the car tune when done with a WB is tuned to precise resolution.... with the NB that resolution is suddenly greatly reduced and so error is automatically introduced....

now in my personal car, the NB's tend to agree well enough that i get about 3% error between WB and NB readings when in closed loop...

however, something very interesting ive noticed.... When trying to run a lean cruise using the holden left over..... i have the VE and MAF literally dialed in with a WB to be spot on... i see variance in the range of 14.5-14.7 at cruise stoich according to the WB....deffinatly with in the range of error for the WB...

however, when it drops to lean cruise in closed loop.... the AFR's are all kinds of wrong.... i see + - 0.5 afr error.... which i have to assume is due to narrowband effective range.... more or less the 15-16:1 afr being off stoich so the accuracy, or lack of accuracy of the NB is showing through when trying to estimate fueling....



just some thoughts...

[Gh0st]

I thought lean cruise is open loop bro

[Goatkart]

I thought lean cruise is open loop bro


dont think so..... never checked to be 100% on that though.... o2 sensors still report back though way low, trims dont zero out, at least they dont on my car....

so.... not sure....

i think there is a pid that could be logged to see if it actually drops into open loop when lean cruise is active, or not....


all though, if it used Closed loop trim data and fell into open loop, that might make sense(well make the bad kind of sense) as to why the afr's end up off....

not sure....


ironicly, that means if it is infact an openloop setup... that if i just setup an open loop lean cruise with the open loop EQ tables... it would be far more accurate, than the holden lean cruise which incorporated trim data....


that would also mean NicD was right when he told me i should just run Open loop if i wanted a lean cruise because it works better..


DAMN DAMN DAMN!

[eficalibrator]

Narrowband sensors can't work in closed loop with lean cruise by definition. They switch at lambda=1.0 Start a new thread to discuss if you wish, but there is only limited benefit here.

[Patrick G]

Check your O2 switch points. 450mv is supposed to be 14.67:1 A/F ratio. Look at the scale. Often, they are at 350-380 at very low loads (leaner than 14.67:1), then rising to 430mv at light loads (still slightly leaner than 14.67), then they are at 450mv for cruising, then going richer to 500-550mv at heavier loads. If you are trying to chase a constant 14.7 A/F ratio, don't necessarily blame your NB sensors, try making all your O2 switchpoints 450mv for all loads. You might have more success matching your NB and your WB sensors.

[Goatkart]

i set my switch points to 451 across the board :)

[redhardsupra]

i set my switch points to 451 across the board :)
i never understood the logic behind mucking with switchpoints... Greg, could you enlighten us to whether there is method to this madness?

[Goatkart]

i never understood the logic behind mucking with switchpoints... Greg, could you enlighten us to whether there is method to this madness?

my understanding of it... just to toss this out there... was to set the median afr in closed loop more or less... well median as the NB sensors understand it..... obviously they read lambda... but when it comes to closed loop operation.. they still need SOME VALUE to switch around in mv that is called "stoich" the switch point sets that value...

understanding that at any switch rich to lean, the afr has to push the narrowband so far either rich or lean to turn it around some value... in other words producing the sinusoidal wave form of the NB switching...

so at any given time you will either be rich or lean, and very seldom will you actually be at 14.6... more often you spend your time in a rich or lean condition...

now the idea of moving the switching points is to center the switch point back and forth over the stoich value....


MAINLY the only time i have seen this matter, is when the o2 sensors are moved from the stock location.... ie Long tube headers shoving the o2 back 6-10 inches from stock...

and even then mainly at or around idle with the velocity of gas in the exhaust is minimal....


just my .02

[eficalibrator]

Let's understand a couple things about narrowband O2's first. One of the biggest misconceptions is that they detect air-fuel ratios. That's not really what happens. What they really sense is the partial pressure of the oxygen molecule on the zirconium dioxide ceramic element. Based on whether the detected oxygen is either positive or negative in balance from a stoichiometric (even burn) reference, the sensor output is pushed either low or high respectively.

There is no magical detection of "14.6:1." They're really lambda (excess air ratio) sensors. A narrowband doesn't care if you're burning gasoline, E85, pure ethanol, natural gas, whiskey, rum, or coal. It just looks at what's left over from the reaction and determines if you've used the oxygen. The stoich point of various fuels will change, but the switching point of a narrowband follows the fuel's stoich point anyway.

There is some deadband around stoich where the PCM has trouble determining if the sensor has really switched. To cover for this, the PCM waits until a significant enough change has occurred to make sure that it really saw a switch. This means that +/-1-2% around stoich is pretty much the limit for narrowband precision. If you tune in open loop to right on stoich and get trims in closed loop of ~2%, consider this to be right on the money. In practice, getting inside of 4-5% is probably close enough. (Remember that the tolerances on your fuel injectors from the manufacturer were +/-6% anyway!)

Temperature has a HUGE impact on narrowband performance. Cold sensors don't switch as readily and have generally lower output voltage. This is what requires us to wait a little bit before going into closed loop on startup. Once running, the switchpoint (measured in mV) may move slightly due to temp, pressure, and other factors even though the stoich point of the fuel remains constant. WE DO NOT MOVE THE SWITCHPOINT TO GET A DIFFERENT AIR-FUEL RATIO. The OEM calibrators are targetting lambda=1.00 everywhere except cold start, power enrichment, or component protection. We aren't biasing the swtichpoint to get "slightly richer" at high loads, we're just telling the PCM exactly how the sensor behaves under those conditions.

"Tuning" switchpoints to get a different AFR is more creative art than scientific application. Sure you can get something that looks good, but you're tricking the system to do so. There are other options out there to more precisely change the target AFR.

[Patrick G]

WE DO NOT MOVE THE SWITCHPOINT TO GET A DIFFERENT AIR-FUEL RATIO. The OEM calibrators are targetting lambda=1.00 everywhere except cold start, power enrichment, or component protection. We aren't biasing the swtichpoint to get "slightly richer" at high loads, we're just telling the PCM exactly how the sensor behaves under those conditions.

"Tuning" switchpoints to get a different AFR is more creative art than scientific application. Sure you can get something that looks good, but you're tricking the system to do so. There are other options out there to more precisely change the target AFR.It's interesting that many older OEM calibrations have O2 switch points with a sliding scale (going from mid 3xxmv to mid 5xxmv), but most of the newer E38 PCMs have switch points locked in at 450mv across the borad. The PCMs that have O2 switch points at a sliding scale seem to give tuners more headaches when trying to match NB sensor data to WB sensor data. Setting all O2 switch points to 450mv across the board seems to make tuning a little easier. This is the point I was trying to make.

[redhardsupra]

In Your Face, 5litereater And All The Other Hacks On Hpt Board! ;)

[Goatkart]

double post

[Goatkart]

In Your Face, 5litereater And All The Other Hacks On Hpt Board! ;)

roflol, that almost made my sig quote right there....

so just to be clear greg... are you saying the switch point table, when used in a sliding scale setup as patrick mentioned above... is merely supposed to be adjusted for the point at which the pcm should be swinging for a specific temperature? and as they heat up so do the mv reading for the pcm to swing around?

if thats the case is it still not a correct adjustment when the o2 sensors have been moved further away from the engine, ie when a collector on a header moves it further back, so its running either cooler or hotter....

also, if it is the case that it is a temperature varriance adjustment that gm is using with the sliding scale, why would they change up to a static value later... which i think they did in the 04+ ls calibrations (dont quote me there)...


interesting info!

[Gh0st]

Rocking 23 degrees of timing from 4400-6400 lol...gotta love high compression.

[lsx427]

Perhaps more accurate than some of the el cheapo widebands (like the Innovate), but not true if you're using good equipment. Remember that these tables are typically built from open loop engine dyno data collected using an ECM wideband. Strangely enough, I use an ECM LambdaPRO for my own projects too. ;) Not all "wideband" sensors or controllers are created equal!

They are typically supplied by the injector manufacturer at the OEM level. But Bosch/Siemens/Delphi/Lucas/Denso are unlikely to give the public that level of data. Your best bet is to copy the same data from another OEM application that uses the same injector. I have done the work to convert from Ford to GM style units on all the popular injectors that were publically posted on the Ford Racing site. I will eventually publish my own "GM converted values" on my website during the next wholesale refresh of it.


Greg i am having trouble finding prices on the ecm stuff-

Any info-

[eficalibrator]

I'm working with ECM right now to make it easier for the performance guys to get the ECM equipment...

[lsx427]

Thanks greg- is afm the same brand as the ecm as it looks like the same...???

what ecm stuff have you used apart from the lamda pro?

is the afm range the same-

http://shop.airfuelratio.com/product.sc?productId=7&categoryId=3

Thanks