Concerning another thread about using the WBO2 output to the NB input, I’ll share my experience. (Plus, I didn’t want to hijack the other thread.)

This graph shows the oscillations I experienced in a 10 second time period cruising at about 50mph. I have a pretty good tune for driveability with the NB sensor with BLM’s pretty much AVERAGING 128 in most conditions with a fully warm engine. With the NBO2 I never noticed any abnormal surging or oscillation while cruising. I unhooked the NBO2 and hooked the WBO2 into the ECM and at a very steady throttle I couldn’t get rid of the oscillations.

I tried changing the PROM settings for BLM update cell amount, rates, positive and negative integrator error. I tried setting the LM1 to 1/12th of a second output. Results were maybe slightly better, but the same. The AFR continued to swing a full point or so during steady cruising with steady throttle. Any other type driving and it wasn’t even noticeable that I was using the WBO2 instead of the NBO2.

I’m not looking to fix this, I just wanted to point out my experience. I have several theories on why. Large cam (CC306), high stall converter (3200), WBO2 in the X-pipe about 3’ down stream from the header collectors and possibly just giving up too soon on tuning it that way.

It’s actually is tough (for me) to get the LM1 to read a smooth AFR with the NBO2 providing the signal, IN CLOSED LOOP. The AFR is all over the place with the ECM perfectly satisfied everything is fine and BLM’s are in order. Now, in OPEN loop it’s easy to tune and the AFR is fairly smooth and to heck with the BLM’s.

So far, my best option is, (as GRUMPY mentioned several months ago), I tune it in open loop. As long as I know my AFR is where I want it, I really don’t need closed loop. Believe me, I really wanted this to work out in closed loop, (for some hard headed reason) but having the LM1 is starting to pay off since I started tuning in open loop.

Now understand, this thing runs really nice in closed loop with the NBO2 sensor. I’ve put 16,000 miles on it running it that way. Most of the guys I know think I’m nuts for spending this much time trying to get it to run better, smoother, whatever, and a lot of you guys would wonder how I got it to drive so well. (BTW, I really believe the torque converter is what makes this camshaft so EFI and street friendly.)

So, the experts can chime is as to why, and the curious can try to figure it out, but I’m happy with my WBO2, in open loop.

SBC 383 w/TFS 23° aluminum heads, full roller, 11:1 SRP's.
CC306 camshaft .544/.576 w/1.6 RR
730SD w/30# injectors, 58mm Holley TB, Holley Stealth Ram Intake
700r4, 3200 Vigilante converter, 3.73 gears w/BFG drag radials.
4040# Step Side Chevy pickup truck
8.34/83.80mph w/1.88 60' w/stock suspension (except it’s lowered 4”)

 

Wait wait wait.. did you hook the 0-5v output up to the 0-1v ECM input???

 

No, I used the LM1's 0-1v output. I actually adjusted it from the preprogrammed .450 at 14.7 from .430 to .500 without much change at all in the overall AFR. The ECM just seemed to compensate with the INT.

The problem appeared to be from looking at the scans from Datamaster that the voltage would swing from .100 to .900 causing the ECM to over compensate so I tried narrowing this range in the LM1 output.

In other words, the LM1 seemed to overshoot whatever value the ECM was wanting it to use as stoich. This value used to be considered a *known* value but lately threads have questioned it's accuracy.

 

Did narrowing it help any? Very interesting, for sure. Definitely something we'll have to take into consideration if we do the NB simulated output.

 

I adjusted the LM1 from .280 for 15.01 and .780 for 14.08, not knowing wether or not those values were accurate but keeping the cross over point at .450. I was just trying to get rid of the oscillations. It helped slightly but not enough to leave it hooked up that way. I did try programming it for .600 14.7 but didn't notice much different in the way it ran, if any.

The only reason I brought this up was because the DIY WBO2 thread is getting cluttered with questions about using the .0 to 1v output and I thought maybe this may provide some information from my experience. I've heard others have no problem doing it but it just didn't work for me. Of course, like I've already said, maybe I gave up too soon.

It's simple to hook back up if someone has any ideas I'd try them.

I've read threads on Innovate and the distance from the headers, (time wise for the exhaust to travel to the sensor), is so small as to be practically discounted.

 

Good info, thanks alot for sharing.
From what you describe the reaction is what I would have imagined it to be.
The ECM is not able to "Control" the swing of the WB due to one of a few reasons.
a.) the feedback is so fast that you are getting out of phase in the control loop. (need ECM to react faster to the error)
b.) The ECM is not seeing the correctly "Filtered" feedback and trying to correct for fluctuating readings. (small margin changes that should have been dampened out)
c.) The short term integrator is just not fast enough to correctly keep the error in check.

I don't know how deep the code goes for these issues but speeding up the reaction (either "I" or "P" in the PID control) of the "correction" control should enable stability of the loop.
Going the other way and only sampling the feedback slower would not be desirable, just induces a lag in the system and will make control that much more difficult if not impossible.

I'm still curious if the code can support a faster loop
The processor SHOULD be capable of doing this if not overloaded with other functioning code.
I work with electro-hydraulic control loops that respond in mS and they can be controlled via analog stuff easily, the digital conversions are usually where the problem with timing and feedback begin.

 

BTW, I've understood from reading here that closed loop is supposed to swing rich-lean-rich-lean and know from looking at scans that it does, but I've never felt it driving with the NBO2.

I assumed, when I wired the WB in, if I had any problem it would be from trying to stay to close to 14.7 and not get cross counts. Believe it or not, cross counts were similar. Sometime Datamaster showing them slow, most the time ok, always staying in closed loop, and never setting any code.

Here's a LM1 log under similar conditions with the NBO2 sensor.

 

This is a very interesting post, and some good reading, so I had to bust in with my 2 cents. I really don't have any advice on fixing the oscillations, just that I would say that the LM1 is a very fast and precise sensor, and like someone said above, maybe the ECM cannot compensate for such a fast signal. Sounds like you played with the adjustablilty of the analog out of the LM1.

My advice would be to stick with the NB for closed loop operation, and tune away with your LM1 in open loop. Sounds like you have been there/done that, and are happy with the results.

I have the LM1, and starting to get used to it. Smasterson, do you have the analog convertor module, to get those other parameters in your graphs? How hard was to set that up? I see that you have the %kPa, was that easy to set up? I am running a MAF car so I would need LV8 or just the MAF value. So far cannot get the analog out of the LM1 to read in my "other" scan tool.

 

Ok, well maybe I just "can't see the trees thru the forest".

Let's see if I can explain my thinking, maybe someone can save me from hitting that tree (or brick wall!)

We have the ECM seeing a slightly lean condition, so it bumps the INT to get it to go slightly rich. When it sees the result of the correction, it will bump it a bit more, or go the other way to get what it wants to see. The ECM can only perform this loop at a certain speed, therefore, if the sensor reads the change faster, it's just waiting for the ECM to make the change. Now all of this happens many times a second. So , the O2 doesn't oscillate, it's just reading what the ECM is doing. The ECM is what is doing making the oscillations.

So, basically, it's like email. The message is there, but you don't read it until the next day. It's good that it's there early, but not a big deal, it's not used until needed.

Am I totally lost here?

 

Just from what I've seen is looking a various calibrations.

The ECM will try to shoot for around 450mv, and it then uses gain tables for various error values of the 02 reading. And from what I've seen on C3 stuff the gain lookup tables only go to 100mv error. This works for the narrow band sensor. but when the wide band is brought in there can be big 02 swings, like +/- 300mv easily. The ecm will think the motor is going very lean/rich, when infact the motor is fairly close to stoich.

So something tells me that the ecm is compensating too much with the wb connected. And the error tables would need to be expanded to say like 300mv of error, with smaller gain values for the 02 error.

SO I guess, the code would have to be modded to make the WB work.

 

Does the WB output the NB simulation as a linear 0-1V signal? If it does then I see how the ECM has trouble with it. The WB should translate the NB simulation as a non linear signal. It is easy to do, not an engineering problem at all.

No need to change the ECM calibration in that case.


JP84Z430HP, Yes I agree with your thinking.

 

Me thinks, it's a bit more complex then folks are making out what the ecm does. It's several routines that the ecm uses for finding rich and lean, and toggling back and forth across it.

My opinion, would be, to dial the car in to a known tune, and then in back to back data logs look at what the NB, and WB converter look like. BLs will tell part of the pic, but looking at the BPWs would be better, IMO. Cause you want to see how lean and rich you go with the new setup.

 

If the WB sensor is reacting more quickly than the NB does, then I don't see any reason that this can't be done on the signal processing (ie. firmware).

I'm with Bruce, in that I think that this issue could benefit from some further analysis to better understand what the dynamics are. Looks like Steve is well on the way here.

It'd be good to fully understand what the NB simulated output transfer function is. My thinking is that will be the big part of the equation. The relative dynamic sensor responses and the expectations of the ECM will be the other.

 

Just as a FWIW.
On the GN I've been running C/L with a GM 3 wire O2 sensor about 4' further back from the stock location. 3" away from it is where my WB is mounted. With BLs of 116-132, the plugs burn the prettiest light beige. Normally GNs run with a bone white colored plug. The WB is toggling across stoich., just as one would imagine.

Now, with the truck with the NB in the stock location, and the WB 3' further downstream, with 128s, the WB is again toggling across stoich.. But, the plugs in the truck are still the bone white.

While not direct comparisons, there just might be some things going on other then what we think.

Next, project is mounting a heated O2 in the truck right up by the WB, and seeing if that duplicates the plug coloring, of the GN.

 

I'd be very interested in the results of this experiment! Further more, it would be more interesting to figure out the "WHY" of it. After some of the discussions I've read, I'm not thinking that the distance is slowing down the reading, but I'm thinking it's more of a turbulence thing? Or, maybe further down the pipe, the gases are cooler (Slightly, hence the need for the O2 to be heated), and causing the reading to be different. From the standpoint of what the OE's do, I've seen them pretty much everywhere, from within a few inches of the cylinder head, to a few feet away from the manifold.... Interesting to ponder though......