When Widebands Lie. Or Don't Ignore Your Narrowbands!

[gmtech16450yz]

I've had an issue with tuning my LSA/E67/PLX wideband setup that I finally figured out. The wideband was giving me very misleading information.

For those that don't know me, I'm not a newb. (no offense, we were all newbs at some point!) I know about wideband offsets and voltage errors. I've done more tuning on factory widebands (E69/LNF's) than anything else. I know how widebands are supposed to work. This is something different than a simple wideband setup or scanner configuration issue. At this point I honestly can't say whether it's a problem that only exists in my particular hardware setup, or is something that only exists with the PLX wideband. (I just looked on the PLX website to see exactly what Bosch sensor I have, and see they have a new Gen 4 Bosch sensor. I bought my setup a year ago, I'll be calling PLX on Monday to see if mine is the Gen 3 or Gen 4. If it's the Gen 3, then the new Bosch sensor may solve my particular issue.)

What I've been dealing with is shown in these log screen shots. When I go full throttle, the wideband reads lean for a little less than a second, then reads exactly what it should. This will happen at ANY rpm and ANY airflow points, so this IS NOT something that can be nailed down in the tune. That's what's made this difficult. At first I thought is was actually lean like the wideband was showing. The wideband seems to react very quickly to other mixture changes, why would it react slowly in this one particular situation? I tried the usual MAF and transient changes to try to get more fuel during the first half second or so of a full throttle hit. I couldn't get rid of the lean indication. My butt dyno actually showed I was loosing power even though it was reading lean. I took it to the dyno and had them run an external wideband during the runs. IT SHOWED THE OPPOSITE OF WHAT MY ONBOARD WIDEBAND WAS SHOWING! It showed that I was way rich during the first second or so, not too lean like the wideband was showing. Readings after that first second were almost identical between my onboard wideband and the dyno's external wideband in my exhaust. If anything, the dyno's wideband in my tailpipe should have been SLOWER to react, not faster.

I stared at the logs for awhile and finally had the last piece of the puzzle. The narrowbands. Look at the narrowband reading in these logs. It CLEARLY shows that the mixture is indeed going too rich during the first second or so of a full throttle event. While the narrowbands are reading very rich (I'm talking about the difference between .900v and .950v. There IS a difference.), the wideband is reading lean! You can also see by the injection numbers that it is indeed injecting a ton of fuel during the time the wideband shows it being lean.

THE WIDEBAND IS WRONG!!! This damn lean spot shown in my logs doesn't actually exist! It's the wideband reacting too slowly to the full throttle event. At no other time in any of the logs does the wideband react slowly, at least from my eyes. Why it reacts slowly in this particular situation and not in others is beyond me. The wideband is in the stock LS9 exhaust manifold, right next to the factory narrowband. It's plenty hot and plenty close. The wideband was bought new a year ago and it's only been used for about 4k miles so far. Like I said, I'll be talking to PLX about this and whether this was normal for the sensor I have.

The reason I'm posting this is because I'm betting I'm not the only one with this situation. It's not something that would jump out at most tuners. And how many of you are using widebands that are more than a year old? I'm betting a bunch. And how many times have people said that narrowbands are useless for tuning? They're not! Sure, they're more accurate around .450v, but that doesn't mean they're worthless at .900v or 1.00v. The factory narrowband in these logs clearly shows what's REALLY happening in my engine, which is the opposite of what the aftermarket wideband is showing. Needless to say, after figuring this out and tuning my engine properly during that first second or so of a full throttle event, the car responds a helluva lot better than before. The dyno's tailpipe wideband was right, it was too rich until 4k or so during the dyno runs. The end result of all this is now I have traction issues that I didn't have before!!! lol.

Hope this helps somebody. I'll update this after I hear from PLX. BTW, this is in NO WAY speaking badly about PLX or their products. This is a Bosch sensor. If they've improved them or there was a problem with the response of my particular sensor, that in no way diminishes the awesome work and products that PLX provides. If this LSA/E67 was like the awesome factory wideband setup in the E69's, no of us would have any of these issues. The Direct Injection and factory widebands are SO incredibly accurate and easy to tune it makes this Port Injection stuff seem like caveman chit! Hopefully my next V8 swap will be with an LT4!!!

edit... I included a screenshot of the dyno runs where the onboard wideband was showing it lean until 4k and the dyno's tailpipe wideband was showing it buried rich until 4k or so. The dyno wideband was right. Leaning out the tune in airflows below 4k rpm or so resulted in WAAAY improved response and power. It was indeed too rich but my onboard wideband was telling me the opposite.

[ingsteve]

Where have you put the wideband?
This may have something to do with tau.

[BigDaddyCool]

Sorry mate but this is actually a fairly normal thing. Like .... where to start here.

You cannot compare a tailpipe sniffer to a wideband in the header, while you think it should be slower your wrong because it slows the reaction response.......

I dont offer a solution except to say 'now you know' like everything tuning, now you can find a way to tune 'with knowing this'.

You go back and aim to move your spike so it levels and tune the spot you need to and the same again..... If you have a dyno you can really just select a long gear to resolve this problem.

There is nothing wrong with the wideband........ I tried solving this issue myself...... until I simply worked out the wideband on a wot spike had its moments where it was wrong or rather had a delay.

Doing wot runs from a standstill in 2nd is easiest, you just have to ignore that lean wideband read........ the only thing really annoying about it mostly is its hard to see the PE enrichment rate for a throttle spike.

Using narrowbands helps...... ultimately you just need to move the spike so it happens before or outside the area's you need to tune......

Dyno tuners simply have dynos and better widebands....... but personally I dont think myself many dyno tuners necessarily even use widebands as their main tuning method, most probably tune in closed loop with narrowbands and confirm with tailsniffers....... easier, they dont have to get under the car.......... either it makes power, knocks, runs rich or lean etc.

The wideband isnt lying........ it simply has that delay. Treat it no different then a VE MAP sensor.......... slow as buggery......

The only thing in a car that is fast is the MAF but even that ....... has lag to some extent. Hence why these cars run a mix of three airflow measurements, to combat this issue........ VE/MAF/Dynamic.

Dynamic is designed to compensate for the maf and ve readings ...... the lag of the map sensor, the quick read of the maf etc........ throttle spikes and transitions.

Much the same way there is the myth disabling Dynamic disables VE........ it doesn't. My guess just like going mafless, you need to go mapless to properly tune the MAF on its own.

There is alot to know, the main thing....... dont fight what you cant change, work around it, work with it......... the goal is always the same....... to get the afr where it needs to be........

[gmtech16450yz]

Sorry mate but this is actually a fairly normal thing. Like .... where to start here.

You cannot compare a tailpipe sniffer to a wideband in the header, while you think it should be slower your wrong because it slows the reaction response.......

I dont offer a solution except to say 'now you know' like everything tuning, now you can find a way to tune 'with knowing this'.

You go back and aim to move your spike so it levels and tune the spot you need to and the same again..... If you have a dyno you can really just select a long gear to resolve this problem.

There is nothing wrong with the wideband........ I tried solving this issue myself...... until I simply worked out the wideband on a wot spike had its moments where it was wrong or rather had a delay.

Doing wot runs from a standstill in 2nd is easiest, you just have to ignore that lean wideband read........ the only thing really annoying about it mostly is its hard to see the PE enrichment rate for a throttle spike.

Using narrowbands helps...... ultimately you just need to move the spike so it happens before or outside the area's you need to tune......

Dyno tuners simply have dynos and better widebands....... but personally I dont think myself many dyno tuners necessarily even use widebands as their main tuning method, most probably tune in closed loop with narrowbands and confirm with tailsniffers....... easier, they dont have to get under the car.......... either it makes power, knocks, runs rich or lean etc.

The wideband isnt lying........ it simply has that delay. Treat it no different then a VE MAP sensor.......... slow as buggery......

The only thing in a car that is fast is the MAF but even that ....... has lag to some extent. Hence why these cars run a mix of three airflow measurements, to combat this issue........ VE/MAF/Dynamic.

Dynamic is designed to compensate for the maf and ve readings ...... the lag of the map sensor, the quick read of the maf etc........ throttle spikes and transitions.

Much the same way there is the myth disabling Dynamic disables VE........ it doesn't. My guess just like going mafless, you need to go mapless to properly tune the MAF on its own.

There is alot to know, the main thing....... dont fight what you cant change, work around it, work with it......... the goal is always the same....... to get the afr where it needs to be........

Thanks for the detailed reply, I appreciate it!

I just ordered the Gen4 PLX sensor/module/gauge setup. I will be able to do direct comparisons between the Gen3 sensor I have and the new Gen4 sensor. It's supposed to be considerably faster in the response time. My Gen3 looks like the response is somewhere around 750ms. That's HELLA slow! Almost a full second until it shows accurate readings? If that's "normal" in the world of aftermarket widebands, that's f'd up. Factory Bosch widebands are damn near instantaneous on response, for what we need them for at least. I'm telling you on DI and a stock wideband it's super easy and common to have your desired and actual lambda lines be on top of each other. You go WOT and the wideband reacts instantly to the change. That's why they can run CLOSED LOOP fuel control even at WOT.

I took these screen shots a long time ago when I was trying to show a guy that his wideband was lazy. The white trace is requested lambda, the green is actual lambda from the stock wideband. This is showing the ECM running an o2 monitor test, it goes way rich, then way lean and watches the sensor's response. Look at how a good sensor follows requested EXACTLY. The lazy sensor follows, but slowly and gradual in the readings. The lazy sensor screen shot is what my aftermarket wideband looks like. Trying to tune with data that delayed is pretty crude. Hopefully the new Gen4 sensor will be much faster. If it ends up being less than 200ms, I can live with that. I'll let you guys know.

Properly working factory Bosch wideband in an E69 ECM...



"Lazy" factory wideband in E69 ECM...

[kangsta]

Good spotting, I dont agree that a good wideband should be 'slow' particular at WOT where there is enough gas flow. It all comes down to the controller and the output filtering. The mixture is sensed by a PID controller on the sensor pump cell current. A slow controller is of course going to have a slow response and also inaccurate data on a time basis. Added to that this signal has then to be output, normally via an analogue signal which is conditioned and filtered at both the wb controller and the cable reading it. So if the net result of your filter settings are very slow or heavily filtered then you wont see a sharp transition in your datalogs. Best to check out all your settings to be sure. I would start with a few tests of the wb controller to determine the response speed and go from there.

[gmtech16450yz]

Good spotting, I dont agree that a good wideband should be 'slow' particular at WOT where there is enough gas flow. It all comes down to the controller and the output filtering. The mixture is sensed by a PID controller on the sensor pump cell current. A slow controller is of course going to have a slow response and also inaccurate data on a time basis. Added to that this signal has then to be output, normally via an analogue signal which is conditioned and filtered at both the wb controller and the cable reading it. So if the net result of your filter settings are very slow or heavily filtered then you wont see a sharp transition in your datalogs. Best to check out all your settings to be sure. I would start with a few tests of the wb controller to determine the response speed and go from there.

Thanks kangsta.

Yeah, like I mentioned, I have plenty of experience with factory widebands but very little with aftermarket widebands. So I have no idea if this is somewhat normal for aftermarket sensors and controllers. The sensors are pretty standard and being Bosch they should be plenty fast in themselves. That leads me to think the PLX controller is what's slowing it down? I hesitate to even put that out there without knowing for sure. I'll certainly be able to say if their Gen3 is faster or slower than their Gen4 once I get the sensor and controller in my car.

Does anyone on here have any logs showing the o2 tests being run or have anything showing the response time of ANY aftermarket widebands? I'd really like to know if all you guys are tuning with widebands that have a 500ms, 700ms or even 1000ms delay. We're working on cars that will do 8 second 1/4 mile times and the sensors/controllers we're using are inaccurate for the first 1/8 of that run? That can't be right. I think the easiest way is if someone has some logs with their wideband AND the narrowbands logged. Mine was pretty obvious to see the wideband delay when I compared it to the stock narrowbands, which react extremely fast.

[kangsta]

I dont have any logs but I will be doing some work over the weekend and will see what I can get for you. There is a big difference in the response of controllers, Id say its all in there. The sensors are the same unless you are using a NTK

[gmtech16450yz]

Ok I got the new wideband from PLX today and threw it in the car. Ok, I have to say this...

HOLY F ING CR@P!!!!

Either my old wideband was literally broken or there's a HUMONGOUS difference between the Gen3 Bosch LSU4.2 sensor and the new Gen4 Bosch LSU4.9 sensor. This new wideband reads like the factory ones I've worked with look. Super fast readings that show rich and lean spots that the other wideband simply did not show. You'd never know they were there if you didn't compare the wideband to the narrowbands.

Here's a screenshot from my first log. Look at the rich spike right after I go WOT. That's the spot that always looked LEAN in my old logs. Look at the difference between these readings and the ones in my screenshots above. Pretty much the same tune between those logs. Amazing! Now I can actually tune this thing properly now that I have some accurate data. If you guys are wondering about your widebands, do some logs with your narrowbands and wideband shown and compare the readings. If you have any doubts about how yours is working or just want one of the newest, fastest and most accurate widebands, I can now vouch for the new PLX Gen4 Bosch LSU4.9 wideband. I'll post up more after I get a few more logs and tunes looked at. So far this first log looks completely different than every one of the other 400 or so logs I've done in this car.




edit...
Found a couple pages on the Bosch 4.2 vs. 4.9 sensors...

http://www.ecotrons.com/technology/b...su_42_sensors/

http://wiki.efihacks.com/index.php?t...4.9_vs_LSU_4.2

[10_SS]

the fast response you see is also what I see with the NGK wideband. Your older logs of AFR look way to smooth to be real.

[BigDaddyCool]

Unbelievable. The newer Bosch sensors react like 10 times faster then the previous gen. I'm going to order a 4.9 to replace my 4.2 ....... damn should def help

[gmtech16450yz]

the fast response you see is also what I see with the NGK wideband. Your older logs of AFR look way to smooth to be real.

Exactly! That's what I though when I saw the very first log from this LSA swap. Like I said, I have almost no experience with aftermarket widebands so all I had to compare to was factory wideband setups. So what I want to know, or better yet, what all of you tuners on here should want to know is this... Is this "normal" for the average aftermarket wideband that most guys are using? If it is, there's no way you can tune precisely with that kind of wideband data.

I'm on my third or fourth tune revision with this new wideband and it just keeps running better and better! Look at these two screenshots. The first one is with the old wideband. The wideband is showing hella lean (.87 Lambda) but the narrowband is maxed at full rich (1.003 volts). Something is wrong there.



In this second screenshot with the new wideband the results are swapped. The new wideband is reading hella rich (.69 Lambda), and the narrowband is still showing plenty rich (.943 Volts). This makes perfect sense. I'm starting to get the mixtures closer to desired as you can seen in this screenshot. I'll lean out that initial rich spike a little more and slightly lean out the rest of the range and I should have a tune that is dead on. Amazing! Let me know what you guys think and post up some comparisons of your wideband readings. I'm really interested in what others look like.

[Quoll]

I'll post up some pics from my new TE 2C0B w/ Bosch 4.9 as soon as I get a chance, which given its going to be 40C / 100F this weekend...we'll see...

First though, and this thread served as a good reminder, I need to reprogram the 0-5V scale to something more desirable ... is 0.7-1.2 going to be a good choice, or do I need a wider lower-res range ?

[07MontRedcp]

Just a heads up for those interested in the new PLX SM-AFR Gen 4, PLXDevices.com has a Black Friday special on these with $50.00 off.... $159.99 instead of the normal $209.99.

BJK

[Iam Broke]

John, are you running E85? Wondering why your target is .78 lambda if not. Thanks.

[gmtech16450yz]

John, are you running E85? Wondering why your target is .78 lambda if not. Thanks.

Hi Tom!

Nope, not running any Ethanol other than the ~E10 in the Cali premium. Yeah until now my target lambda was just a number anyway since I'm not really sure how often it was even hitting that number. The best data I really have is when it was on the dyno and I got those wideband readings. They actually made sense and looked accurate. That is a really good question though.

Hey everyone, what PE would you run on an LSA with stock pistons at a max of 13psi and on crap California pump gas?

Seriously, what would you run Tom? You know I know a little bit about LNF's, but I haven't done a whole lot of work on supercharged LS engines. This LSA is definitely making good power for the mods I've done, and both dyno sessions were with it running really rich. Hella rich actually. That screenshot above was from my last dyno run, below 4k rpm was junk because of that old wideband's false readings but the over 4k stuff was somewhat accurate. It showed ~11.0 to 11.8 afr or .75 to .80 Lambda. That's horribly rich for an LNF, but isn't that decent for a boosted LS? Lemme know now that I can actually obtain the EXACT mixtures I want!

[Iam Broke]

Well, I asked because I was targeting .80 to .82 at 10 psi on the LS3/E85 before it threw the rods. Seems rich to me on pumpgas but I have no experience on 91 with boost. I picked .80 as a point in the recommended range of .71 to .86 on E85 and it was reachable with my fueling setup (pwm Livernois twin, factory fpcm) and 80% idc on 95 lb/hr injectors.

Max power rich on gasoline is .85 lambda and max power lean is .90 so adding 10% ethanol will lower it a point or two. I'd probably target .85 lambda and go from there. As I said, I have zero experience with a boosted LS on pumpgas so I'm eager to hear what the pros say.

[gmtech16450yz]

Well, I asked because I was targeting .80 to .82 at 10 psi on the LS3/E85 before it threw the rods. Seems rich to me on pumpgas but I have no experience on 91 with boost. I picked .80 as a point in the recommended range of .71 to .86 on E85 and it was reachable with my fueling setup (pwm Livernois twin, factory fpcm) and 80% idc on 95 lb/hr injectors.

Max power rich on gasoline is .85 lambda and max power lean is .90 so adding 10% ethanol will lower it a point or two. I'd probably target .85 lambda and go from there. As I said, I have zero experience with a boosted LS on pumpgas so I'm eager to hear what the pros say.

IDK I think .85 Lambda for a ~750bhp 13psi setup seems a little scary lean to me. I'm on a stock LSA cam so my dynamic compression is kinda high compared to the big cam setups. I had my PE as a slight ramp from a rich low end of .77 to a little leaned out .80 for some over-rev. With this new wideband I'm going to try requesting a flat .79, which is 11.6 afr or 1.251 EQ in the tune. From what I've seen I think that's a safe PE setting for my particular setup on pump gas/E10.

edit... BTW Tom, I forgot that I had my stoich at 14.2 in the tune so that's why my desired Lambda was so low. You had me thinking I had to compensate for the E10 in the PE, but I forgot that the stoich change also moves the desired PE to where it needs to be automatically. So yeah, I was commanding a little leaner PE in the tune but the desired moved lower to compensate for the E10 like it's supposed to. It's all good, unless some more experienced guys on here tell me I'm nuts. I think I'm pretty close to a good, safe WOT mixture. If I was running straight gas, my desired PE Lambda would be showing .80 instead of .78. If nothing else, I know I'm on the rich side instead of the lean side.

[MikeOD]

At that much cylinder pressure, with the stock pistons, I'd probably be nervous going much leaner than 0.77ish lambda...and I'd feel better around .76-.75. It might be ok with a pull on an inertia dyno (dynojet) at .79, and might show great power...but out on the highway, or at the top end of the drag strip when it's actually working, think about the heat you're making in the chambers. Hopefully your knock sensors are working...but I don't like relying on them to make up the difference between a dyno pull and a real world pull...I like them to be there for the random bad fuel or other variable beyond my control that could hurt the engine.

[gmtech16450yz]

At that much cylinder pressure, with the stock pistons, I'd probably be nervous going much leaner than 0.77ish lambda...and I'd feel better around .76-.75. It might be ok with a pull on an inertia dyno (dynojet) at .79, and might show great power...but out on the highway, or at the top end of the drag strip when it's actually working, think about the heat you're making in the chambers. Hopefully your knock sensors are working...but I don't like relying on them to make up the difference between a dyno pull and a real world pull...I like them to be there for the random bad fuel or other variable beyond my control that could hurt the engine.

Thanks for the input Mike, that sounds like good advice. Tom (Iam Broke) and I are from a tuning world where .88 or even .90 Lambda is fine on a turbo engine at 30psi boost! Not only that, but retarding ign timing or richening the mixture usually results in KR on those engines! Guys would freak out when they were getting a ton of KR and I'd tell them to lean it out and advance the timing. The Direct Injected engines work more like diesels than gas motors sometimes.

And yes, I totally agree on the dyno vs. street thing. The only time I've used a dyno specifically for tuning is on a race car where it was cheaper to run the car in the team's $100k dyno room than it was to rent a track and a driver to run the car in it's natural environment at race speeds. After a couple days in the dyno room we still did track tuning, mostly because the car made so much power the Pro driver testing it said he needed me to tone it down!

On this particular LSA car I have about 5k miles on it since the swap and I'm on my 543'd tune revision and my 477'th log! I've had it on the dyno twice, only two or three runs each. I didn't do any tuning on the dyno, it was mostly just to satisfy all the guys asking how much power it made! I'm gonna richen it back up a tad to where I had it, .78 Lambda. It's been happy there for most of that 5k miles so I think it will be fine. Out of those 477 logs I'm betting I have at least 400 full pulls on it!

[BigDaddyCool]

Well I bought an AEM version of the new sensor........ I still have the same lag(nowhere near as bad tho). Overall sensor is much faster generally speaking. I log via Dell Venue Tablet and I suspect it's somewhat 'slower' in terms of hardware for logging.... Plus my wire to the HP Tuners device is longer.... I think like anything the longer the cable the more lag you'll get.

As it is my offset is 9.835 .... which probably relates to the length of cable as well so. But yeh, still it's faster. Changes actually improve things instead of overshooting. You can have all the tables in the world histogram wise....... but sensor performance is paramount.