Motec-wide band lambda control

well, if nothing else, I found the invoice for a replacement sensor....and it says it's wide band. hooray! i also remember 5 wires. I also discovered that at high loads using wide band will make a correction in 0.4 of a second. clearly, the map is already very close, so i think i'll try it, do some logging and see how i go. What's the life of a lambda sensor meant to be?

i've made some corrections to the map to bring it inline with .9 lambda at wot using the last lambda command first, though, so we're talking very small corrections.

also, the overrun cut out i'm referring to isn't the red line cut out, but cutting off the fuel when you're coasting with no throttle....every injected car I know of does this.

You've all been very helpful...thank you.

 

the advantage for using a wideband sensor is very simple. The ECU can directly read the lambda with a wideband, whereas, the narrowband requires the ECU modulate the injector pulse widths around stoich to observe toggle events. The narrow band sensor is designed to toggle voltage at stoich (lambda=1). If you look at the narrow band data sheet you will see what I mean, the voltage vs. lambda curve is very steep at lambda =1, it looks like a step function. So in closed loop mode, the ECU verifies AFR by observing the toggling of the sensor's signal as the injectors pulse width modulates around stoich. This is common to ALL narrowband implementations everywhere on any car. You can see now, why it is called narrowband (because of the narrow O2 sensor response), and the slow nature of this algorithm only allow closed loop to be applied at idle and cruise conditions. It is also less efficient because there are periods where the ECU is purposefully burning slightly rich. HTH

 

O2 sensors can last a long time. 30k miles is reasonable, even 50-60k. Life is significantly shortened if leaded fuel used, or exposed to glycol antifreeze (eg. bad head gasket). Extreme heat, or extreme rich mixtures aren't good either. And the response times tend to slow down as they age.

 

Oh, I don't think that’s true.

For any control system, a sensor that is most sensitive at the point of interest is the best solution and the reason nearly all OEM systems narrow band sensors. There is no need to toggle the A/F mixture to read it with a narrow band sensor. The couple aftermarket ECUs I’ve messed with don’t do that and I just can’t imagine why the OEM ECUs would do that either, they use standard PID algorithms. The reading will vary in a cyclic way due to basic system control algorithms in any system trying to control anything. The reason the AF/F appears to vary much more in a narrow band system vs a wide band system is just simple the sensitivity of the sensor. A wide band senor simply can’t measure very small changes, so you don’t see them in the data, but that doesn’t mean they aren’t there. If the mission is to control A/F at or about 14.7, a narrow band sensor is the best choice and should last 100,000 miles.

A wide band sensor is very useful for baseline tuning, but I’ve never seen any reason to leave it in the car after the maps are done unless I plan to mess with stuff again.

 

Definately get a new sensor when you are this far into an issue! The Bosch LSM-11 lambda sensor is the 4 wire sensor for MoTeC 3D,M4,M48,M8)and the LSU sensor is the 5 wire for the M800,M600,M400,PLM. The later MoTeC equipment will also accept the NTK Uego sensor but $450.00 each.

 

I was under the impression that wide band O2 sensors were more delicate and you had to be more careful about
where they're mounted so as not to cook them (I have a turbo car). I also thought they did NOT last as long as a narrow band sensor.

Now the data logging system I installed provides a narrow band electrical output derived from the wide band sensor used when you replace the narrow band one so that your ECU is unaffected. But in my case we just added another bung and left the pair of narrow band ones in place. The wide band is removed when testing is finished.

Am I wasting my time by removing it?

Bob S.

 

Bob S-
yes, wideband sensors are much more complicated devices than narrowband, and its fair to say they won't live as long as a narrowband. Still, you can get many miles out of an LSU4, and many cars use them, volvo, vw, audi etc....they are cheap, about $35-45 internet price. True, sensors should not be located to close to the headers/turbos, or mounted vertically to avoid heat soak.

for the other question, if I'm following you correctly, it seems like you could have left the wideband in for data collection and used the narrowband outputs to drive the ECU, just as intended. Or what you described is perfectly fine too, if you do not want/need to collect wideband data.

 

mke-
I'm certainly not the last word on the subject. Go research it for yourself. There are several good books on the subject. Also, it is easy to make some direct measurements. A VOM meter ont he o2 signal lead will allow you to see the toggle effect.

It isn't a question of sensitivity, both narrowband and wideband are sensitive. The wideband simply is useful over a larger range of AFRs. The narrowband, practically speaking, only indicates which side of stoich the AFR is at. If the mixture is within a very narrow lambda band, yes, the ecu can measure lambda, but accuracy falls off very quickly. This dictates use of the toggle algorithm.


***edit**
let me resuggest, a o-scope is best for observing the o2 signal. An analog meter won't work, but any digital meter should capture fast enough to observe changes. And, as I alluded, the toggling is an effect of the algorithm. You are correct, there is a very complicated layer of logic/calculations around closed loop updates. The point is, the final injector PW on narrowband is always changing.

 

That was kind of my point. I have used and O-scope on the signal and the couple set-ups I've really worked with and there is no "toggling". There was noise and oscillation, the signal averaging that is part of the control algorithm damps it to the point that the injector pulse time are steady within less than 1%. I don’t see how you could ask for anything more.

And it is definitely a question of sensitivity just by simple definition. The signal from the sensor is analog and must be digitized to by used. To do that, the full range of the signal is divided into the # of bits the ECUs processor will accept. When that is done, with a narrow band sensor, about ½ of the bits are in the control range, giving very good resolution in the region of interest. With a wide band senor something less 1/20th of the bits are in the control range, so the ECU simply doesn’t have much data to work with. Low power closed loop using a wide band sensor just doesn’t seem like a good use of the sensor. It works and more and more aftermarket ECUs are allowing it, but I honestly believe that is more about marketing than engine control. People like to read A/F in there data logs and to do that, they need a wide band sensor and since they already have it they figure they might as well use it I guess….but that doesn’t make it better or even as good as a narrow band sensor. I think the key is that most people doing engine mods like know A/F at WOT and don’t real much care if they lose say ½ mpg (maybe it’s less) with a less accurate sensor as long as they can pass emissions.

 

I think understand your point now. To paraphrase, the voltage swings ~600mV over a small range of lambda values (0.9-1.1), therefore the AFR resolution is increased within this band because the A-to-D convertion occurs over wide voltage range.

I hope that captures well enough. It simply isn't true though for a couple of reasons. First, the narrow band sensor is not accurate, except for the switching point. The voltage response drifts with temperature changes, and the drift amount varies from one sensor to the next. (unless it is an lm-11. In that case, special circuitry is necessary to backdrive the sensor to measure the resistance, then the reisitance infers the temperature. The temperature allows a correction to be made. It is really a 3-d transfer curve) Secondly, digitizing the signal over the 0.9-1.1 lambda does indeed provide a large number of "digital buckets" (i.e. resolution), but what are you resolving...lambda of 14.61 vs lambda of 14.62. Even if it were accurate, it doesn't translate into anything meaningful.

I tend to agree. Narrowband is mature and effective. Manufactures finely tune the ECU software and engine for excellent results. Emissions requirements are getting more strict though, and seems to me that wideband may be necessary. Wideband sensor has come down drastically in price though and is gaining popularity. Now for an aftermarket install, I would not consider a narrowband system because wideband has the huge advantage that any target AFR can be set for any efficiency/load point. This is a great convenience because, more often than not, the installer did not develop an accurate fuel map with a dyno.

It's an immensly interesting subject to me. And there's alot more to be said. I certainly continue to learn all the time about EFI.

 

The Bosch 5.2 on the 355 and later DEFINATELY adapts. After making changes that will modify the AFR e.g change to higher specific gravity premium fuel, you will see a much richer mix initially on WOT. After 3 or so dyno runs, you will see the AFR re-establish itself at the ECU's target AFR number. This could not happen unless the ECU software was adaptive. When I refer to MAP systems, I am talking modern Motronic 5.2 onwards, which can run MAP or MAF depending on programming, not the early EFI systems.
On the issue of AFR's having used both factory systems - Porsche and MB, as well as chip tuning software, you'd be surpirsed the amount of data available to the ECU.
Aside from the obvious eg. manifold pressure, rpm, load, pre and post cat lambda and voltage, MB stardiagnostics even calcs torque readings!! The info I have seen demonstrates, to me at least, that the assumptions you are making are not correct. This doesn't neccessarily make me right though!
Whe doing dyno tuning, I used to video the outputs of the factory diagnostics systems, whilst in capture mode, and combine that with the dyno readings for accurate development work. I'll see if I can dig up any of the screen shots if anyone is interested.

 

Not true. Whilst thet are different, they are no more complex and certainly no shorter in life span.

In fact, I'd almost make the argument that wideband lambda's (certainly the NGK) is made to a higher standard than you're average OE-standard narrow band.

 

Well now what do I do?????? Dueling technicians, worse than attorneys!

Just kidding. The thread is great. For me the longevity issue isn't that much of an issue since it's a race car and I like to datalog every session anyway. But that is what I was told by the techs at PLXDEVICES. Their wideband (I forget the brand I think it was a Bosch piece) was $79. MY stock O2 sensors are NGK so just maybe I'll just be blessed with zero failures.

Ok, keep dueling, it's fun!

Bob S.

 

Ehh, even Commodes do that these days. But you are right, the M5.2 (and a lot of other OE systems) have these sort of "self-learning" or "long term trim" functions.

But you're all looking at the wrong end of the equation for the answer........

 

LOL.

I;ve got customers who have been using those NGK's in race cars for 3 seasons without a hiccup. So long as you stick to unleaded, they're fine.

A very similar sensor is used in late model Honda's with no reliability issues either.

 

Meaning... run good quality fuel of known ingredients, and check the spark plug colour/condition to see your mixtures/engine health...

good, I like it... back to basics!!

The adaptive technology is good in theory, but how does it "undo" what it's learnt...

lets say, you've run 95 or even (god forbid) 91 RON fuel in your M5.2 car for a fair while, and the car had adapted to it.... now you put 98RON or better in... how long does it take to return to full potential....??

I know that, in a 360 with M7.3, changing from 98RON to 102RON gives instant results, AND as soon as you stop using the 102RON, it goes straight back to whatever it did with 98 RON.

I've done this many times, and analysed data from the track. Zero doubt.

 

To be felt instantly, it can only be a result of too much ignition advance for the 98 RON fuel.

 

Except reading plugs with leaded fuel was a breeze. Good ACCURATE readings with unleaded are not as easy IMO. Racing a TZ250GP bike for two seasons in the ARRC, gave me plenty of exposure to reading plugs. With the added complication of pressurised airboxes, two stroke tuning is an art, more than a science.

 

I want to get a new wide band lambda sensor. Where in Melbourne can I go to get it, and how will I know it's compatible with the settings in my Motec?

incidentally, it's mounted in a section of exhaust where the by-pass valve used to be (end of the "y" piece). Can anyone confirm that there'd be NO problems getting an accurate reading from there?

 

The most accurate place for the sensor is at the collector or front section of the cat... where the std one is on 2.7, or the FRONT one on 5.2.

Sampling at the Y piece cannot be accurate, as flow around the system varies with rpm and throttle position.

 

fanbloodytastic.

all the "experts" including Nizpro, btw, have been stuffing around (on and off) with my car for all these years using a sensor that can't get an accurate reading. grrrrrrrrrrrrrrrrrrrrrrrrrrrrr

that might help explain why pulling 5% out of the fuel gave it so much extra oomph on the road, and why it's doing it with a "reading" of .94

 

Peter
Any decent exhuast shop should be able to relocate the sensors for a few hundred dollars. You are running two I assume? Why the change from the original location, are your headers non standard?
As an aside I saw very little difference between the lambda readings coming out of the OBDII port (from the front O2 sensors) and from the tail pipe sniffer. When an O2 sensor starts dying it does read lean, which with an adaptive ECU's give more fuel, which causes an over rich mixture, which makes the problem worse.....
That's another issue with dyno operators, O2 sensors are considered consumable, but many tight arse operators don't change them anywhere near as frequently as they should. If I see a WOT reading approaching anywhere near a lambda of 1, I insist on the sensor being changed.

 

i wouldn't even know what to ask for...and who's going to fit them??

just one.

no idea, and they are standard. the holes where the lived originally are blocked on one side and there's a non used sensor in the other.

 

Peter
Sorry, edited my post after you responded.
First thing I would do is have two sensors put back into the original locations, which I assume have either been welded over or plugged. I assume the Motec has no problem dealing with the two separate inputs. Mayvbe one of the experts can chime in...
Mr Muffler in Prahran is cheap, quick and knows what they are doing with the exhuast mods.

EDIT: O/T how come my posts don't come up as edited when I edit them e.g. "Last edited by Aircon : Today at 03:08 PM.
", but everyone elses does?
EDIT 2: See if this works
EDIT 3: Mike it must have something to do with being a moderator

 

I think it's only noted if it's edited after more than 2 minutes or so (vague recollection), otherwise maybe it's because you're a moderator?