WBO2 is off from NBO2

[RADustin]

Guys,

Finally set my truck to run SOL on a COS5 tune. AutoVE'd to <=1% change across both VE tables.

STFT are trimming -3% and making me run lean under all closed loop.

I set NBO2s to switch at 450mV across the board.

I don't believe I have an exhaust leak- but will 100% verify in the next few days.

Is this a common issue? Any ideas?

[joecar]

Post log/tune files.

[RADustin]

See attached.1855718558

[joecar]

There's no WB pid in that log file... post your calc_pids.txt (it looks like you got WB connected to EGR input).

[RADustin]

I'm using a/c inputs.

18561

[RADustin]

my WBO2 is an innovate LC-2.

http://www.innovatemotorsports.com/products/lc2.php

[joecar]

Ok, thanks, I'll be able to take a closer look later today.

[RADustin]

Thanks Joe. I haven't verified exhaust piping integrity but I will this afternoon.

The WB02 if wrong is making the truck too rich, so at least under boost I am safe according to both sensors. I have parked the truck though until I can come up with what to do next.

Something else worth looking at is my MAP reading. Does this seem high for the speed? My truck idles at ~60kpa which just seems very high.

[RADustin]

I'm also loosing timing somewhere. I need to investigate more(as usual!)

[RADustin]

Anyone have any ideas?

Exhaust pipe integrity looks great.

The NB02s are mounted in the collectors which feeds directly into a welded y-pipe where at the exit of the y is the WB02. Following that is a couple feet of a pipe and a muffler.


If any air(free 02) was entering the exhaust it would enter in front of the NB02s. So I am at a complete loss.

[joecar]

I'm still looking thru your files...

[joecar]

An airleak up to 12" past the the NBO2 can effect the NBO2 (the low pressure behind each exhaust pulse can sharply suck air in).

[RADustin]

the only place air could enter would be about 1.5" ahead of the NB02 where the collectors meet the headers.

so my thought is that if I do have a leak(doesn't look like it) the NB02s would see this air as well as the WB02.

[ACCLR8N]

My LC-1 doesn't agree with my NBs either. I added a 1.8% fudge factor to my calc PIDs to compensate. I chalked it up to sensor tolerances.

[RADustin]

I'm like 3+ percent off.

I had to pull a header as the coating wrinkled, but when it's back together I am going to use the innovate software to compare the WBO2 to itself reading it thru the ecu and thru the innovate software.

I really don't want to fudge things, but I do admit I believe the NBO2s over the WBO2.

[ACCLR8N]

How old are your NBO2s? It is not so much believing them over your WB, they are running the engine and your correction factors will be thrown off by the deviation. How come you are not taking advantage of the LC-2 serial connection directly into EFI Live?

[joecar]

...
How come you are not taking advantage of the LC-2 serial connection directly into EFI Live?+1

see here:
serial-port-cable-that-goes-to-v2-->post#3 (https://forum.efilive.com/showthread.php?9340-serial-port-cable-that-goes-to-v2&p=81205&viewfull=1#post81205)
FlashScan-to-LC-1-Serial-Cable-Diagram (https://forum.efilive.com/showthread.php?8115-FlashScan-to-LC-1-Serial-Cable-Diagram)

(LC-1, LC-2, LM-1, LM-2 serial comms are all the same)

[RADustin]

How old are your NBO2s? It is not so much believing them over your WB, they are running the engine and your correction factors will be thrown off by the deviation. How come you are not taking advantage of the LC-2 serial connection directly into EFI Live?


I am running LC-2 into ECU through the a/c port so that I can setup something like the app Torque later on to monitor everything. The vehicle platform is an old truck and I'd like to keep the cockpit clean without the additional boost and WB02 gauges. Probably dumb in your opinion but I'm only slightly OCD.:banana:

The 02s are what came with the motor when I got it. Its a ~60k mile pull out. They could be original but they read similar to each other so I would assume they are reliable. All 02 sensors are mounted at 12 position so moisture isn't an issue.


+1

see here:
serial-port-cable-that-goes-to-v2-->post#3 (https://forum.efilive.com/showthread.php?9340-serial-port-cable-that-goes-to-v2&p=81205&viewfull=1#post81205)
FlashScan-to-LC-1-Serial-Cable-Diagram (https://forum.efilive.com/showthread.php?8115-FlashScan-to-LC-1-Serial-Cable-Diagram)

(LC-1, LC-2, LM-1, LM-2 serial comms are all the same)

I will look into this.

Thanks.

[RADustin]

well I have the null modem connector already so it looks like it is simple plug and play and change my calc.pids file.

edit-

if it is off I should be able to chart my two different AFR readings to 'scale' the one run through my ecu.

[ACCLR8N]

Your AC pressure is averaging 2.7V. Isn't that lean? 1.05 lambda

[RADustin]

I'm thinking this could be a rounding error deal. - it isn't but I left my rationale below anyways.

I have B3601 set to 14.63.

so say my truck is running at 2.7V on this A/c line. 2.7V is fed into my calc/pids and 2.7*3+7.35= 15.45 AFR comes out. To calculate lambda(to get BEN) take 15.45/14.7=1.05(1) lamda. BEN factor is then the same as LAM as GM.EQR is 1. But should it be?

EQR of 1 is really for 14.63 according to B3601. Therefor my LAM/BEN should be 15.45/14.63= 1.05(6). Ah! still works- rounding error isn't there.

Now, voltage to AFR should be using 2.7*3.008+7.35 per innovates numbers. So back to 2.7V- AFR would be 15.47. Then making this LAM/BEN- 15.47/14.63=1.05(7) Ah! still works- rounding error isn't there.


I'm going to keep chasing this. I get my header back on(had to take it off and have the coating fixed) I will reprogram LC-2 to output 2.5V all the time- then verify what I am getting in the a/c circuit. I did connect the a/c pressure ground directly with a wire to the innovate ground to try and avoid any ground potentials. Possibility the voltage coming from LC-2 is scaled slightly.

[RADustin]

Your AC pressure is averaging 2.7V. Isn't that lean? 1.05 lambda

yes that is corect. my WBO2 is reading leaner than my NB02s. or my NB0s are reading richer than my WB02. Either way...if STFT was 0(not negative or subtracting fuel), my WBO2 would read stoich or very close to it(the lean condition would get richer).

I have a feeling my entire VE table is 3-5% lean because of this.

My WB02 will average 15.442(2.7V) or 1.05LAM, and my STFTs will be ~-4.5%. So I think my VE is tuned very well to my WB02. I just need to figure out why it is off.

If I export the log and sort by cylinder air- and I average common cylinder air BENs and STFTs- it is painfully apparent the fuel the STFTs subtract is showing up in the WBO2 nearly perfectly.

[joecar]

EQR = 1 regardless of 14.63.

[joecar]

If you run WB while any trimming is gong on, you can't apply the BEN correction from this.

[RADustin]

Coming back to this- finally. Just want to follow up with my solution.

Got my WBO2 hooked up through serial to the V2. Then compared the signal from serial(considered most reliable) to my signal from the the ecu reading through the A/c Pressure line.

I do have the a/c pressure ground connected directly with the LC2- to try and avoid a ground offset. Either way- I'm getting some kind of offset.

I compared the signals in excel, then back calculated the offset by using a/c pressure voltage and serial AFR at same time stamp. I assume this to give me the most reliable figures. I only did a static idle log, so I will continue to play around with a driving log when I'm not so hungover. My error does go down around frame 900. Not sure why.

Attached some images and the excel file for those who are interested. Also if my EFI log file. I would highly recommend anyone using this style of WBO2 input to calibrate their setup before tuning- it could have been disastrous.

[joecar]

Do you have a calc pid to convert ACP voltage to AFR and/or Lambda...?

[RADustin]

yes.


18848

[RADustin]

my formula was ACPx3+7.35 Now it is ACPx3+6.92.

I could rescale the LM2 output, but this should work for now.

[joecar]

The LC-2 serial comms emits the Lambda and AFR that it's microcontroller calculates from the wideband sensor...

it calculates the AFR from it's determination of lambda and the stoich AFR programmed into it (using LM Programmer software).

[joecar]

And the LC-2 scales the AFR to fit within 0-5V and outputs this its analog output.

[RADustin]

Yes I understand.

The problem I had though was the ACP voltage and AFR was off from the direct serial AFR.

So what I did was scale my equation to align the AFR through ACP, to the AFR from serial. My assumption is that the data direct through serial is most correct.

[joecar]

Do this: with a DMM (set 20VDC scale) measure the voltage at the ACP connector (backprobe using a needle probe, or use a piercing probe (and fill the hole in with lacquer (wife's nail polish), not RTV))...

see how this voltage compares to the pid GM.ACP.

[joecar]

...

My assumption is that the data direct through serial is most correct.Yes, correct, this is a view of what the LC-2's microcontroller is doing internally.

[RADustin]

I'll check it with a voltmeter in the next few days.

I've read of others having to fudge the equation a little to work. I have common grounds so I know it isn't that. I may play with a bench ecu and my power supply and check the GPIO pins to see if there is any kind of bias within the ecu.

[joecar]

Also check if there are any voltage drops across the various ECM grounds while engine is running.

[RADustin]

that will be very hard to do while the truck is running, but I will try. The ECM is under the truck, under the drivers floor board. I can ohm the grounds for sure.

[joecar]

( yeah, I know, easier said than done, sorry... )

[RADustin]

I can't find anything weird.

only thing I can find is that other people who do the same have to 'offset' the input voltage.

I have not probed the output on the LM2. That is the last thing to do.

[Gelf VXR]

The LC-2 serial comms emits the Lambda and AFR that it's microcontroller calculates from the wideband sensor...

it calculates the AFR from it's determination of lambda and the stoich AFR programmed into it (using LM Programmer software).

Hi Joe

For lc-1 or lc-2

Just want to confirm, I know the analogue 1 & 2 output volts referenced to Lambda can be programmed using the LM programmer, also the response speed, serial output is unaffected by the programming?

For example the serial AFR range is fixed say 7.35 ~ 22.38 for 14.7 gasoline, and the response speed 1/12, it cant be changed?

[joecar]

Yes. The LC-1 and LC-2 are functionally the same.


Yes, correct, programming the analog channels voltage:lambda output curves does not affect the serial comms;

and, vice-versa, programming the stoich AFR does not affect the analog channels

( the serial comms output is like a console terminal for the microcontroller... the microcontroller controls the LSU sensor, it determines lambda from the LSU sensor, it writes lambda and calculated AFR to the serial comms port, and it drives the analog channel outputs according to their programmed voltage:lambda curves )


LM Programmer lets you change the wideband response filter/speed regardless of the serial comms AFR range.

[RADustin]

I still haven't probed the output on the lm2.

I need to get a serial to USB converter and will reprogram output 1 to match 2. so it will be 0-5V and not emulate a NBO2.

Then I will run this into one of the orange plugs on the V2 and log it next to the input through the ecm. and see which is off.

I did get my truck tuned out fairly well by going all serial- worked much better.

[RADustin]

worked on this a little bit last week.

Still not sure why my analog WB02 input through the AC pressure circuit is wandering so much.

Made a log of a simple around the block drive.

I logged-

CALC.WB02(AFR)- Analog input from innovate to gm pcm through ac circuit. Innovate has 12v power from battery. ground tried to chassis with the ac pressure senser ground for common ground.

EXT.WO2AFR1- digital input from innovate straight to the V2 using the serial cable

GM.ACP- ac pressure voltage.

CALC.WBO2ERROR0- was a subtraction of CALC.WB02 - EXT.WO2AFR1 to see if a simple offset was present

CALC.WBO2ERROR1- was a division of CALC.WB02/EXT.WO2AFR1 to see if a simple scalar was present.


the two calculated errors were useless. the CALC.WB02 is just too all over the place. GM.ACP doesn't natively carry enough precision, so I used my liner equation to scale the voltage to AFR in the reverse to reverse out the ACP voltage with more precision.

so cal'd GM.ACP is (CALC.WBO2-6.92)/3 because my standard equation is CALC.WBO2= (GM.ACP*3)+6.92


Anyways, once I had a good data set for GM.ACP, I graphed it against the serial WBO2 data to come up with a 'better' linear equation to relate the data. Its much different than what I had, and would be CALC.WBO2= (GM.ACP*2.5211)+7.6443 but that equation only carries an R^2 value of 0.75. I was hoping for a better fit but it is obvious the data is just literally all over the place.

For now, I'm just trying to get the gauge on my torque app to be somewhat close- I use the serial WB02 to tune with. So I think this is 'close enough' but in any event, it is irritating the major differences in the two forms of WB02 reporting.

Posting all this up encase someone else is having struggles or may have some insight on why it's off.

20304

[joecar]

Does your wideband allow you to do free air calibration...?

Is there another LSU sensor that you can try...?

[RADustin]

Does your wideband allow you to do free air calibration...?

Is there another LSU sensor that you can try...?

I guess I don't understand how it would matter.

The innovate LC-2 is reporting the same data on its analog and serial lines. I am reading it in two different ways- and getting fairly different values.

From Innovate-
Analog Output #1 (Yellow) is configured as 0 volt = 7.35 AFR and 5 volts = 22.39 AFR

The equation should be AFR=3.008*(volts)+7.35

My y-axis zero cross is off. 7.6443 vs 7.35 which looks like a voltage offset. Then the slope is off as well, where the cal'd slope is much less- 2.5211 vs 3.008. I have no explanation why the slope is off. A simple offset would carry the same slope.

Looking back at my data- if I calc what the GM.ACP volts is supposed to be based on the EXT.WO2AFR1, and I compare this value with the cal'd GM.ACP from the CALC.WB02 PID I can see what possible voltage offsets are going on. I get a min/max offset of -.72/.79 or total swing of 1.51 volts.

I suppose this looks like a floating ground on the ACP volts input. I will investigate that. I really don't know how else it could be this far apart.