Tuning tips and corrected AFR's

[The Convert]

There are a couple conversations here.

1) The "ADV" sensor as a replacement for the "LSU 4.9" sensor and the required flash changes to make the sensor compatible. I have not been following this discussion much. It's entirely possible fundamental changes are required to the DME tables to adapt the sensor.
2) The effect of back pressure on factory sensor reading. This is what I'm talking about. e.g. does pressure effect the factory sensor reading and if so how?

If we're going to pretend to be o2 sensor engineers lets at least play the part and cite the spec sheets of the sensors we're discussing before patting ourselves on the back at our proposed solutions.

You need to do a little searching then because your above two questions have already been answered ad nauseam in other threads specifically related to the O2 sensors.

[andy_divers]

@3000gt MR . I'll seen if i can find some. Alot of my tests on this were on a laptop that took a $#@! on me. I may have some from shiftsector. All reading are based on datalogs. Im not sure on the dme sample rate, but the haltech logs the afr in 10ms samples currently.
On a side note, dme lambda to dyno lambda variance has a ton of variables to consider. There's no way to point the finger at backpressure for those variances alone.

I don't want to come across as bashing or disagreement. It is a very interesting topic. I was just sharing what I've seen. And I am one of the few people that have put back pressure sensors and standalone widebands in different locations to see this kind of data.

[The Convert]

@3000gt MR . I'll seen if i can find some. Alot of my tests on this were on a laptop that took a $#@! on me. I may have some from shiftsector. All reading are based on datalogs. Im not sure on the dme sample rate, but the haltech logs the afr in 10ms samples currently.
On a side note, dme lambda to dyno lambda variance has a ton of variables to consider. There's no way to point the finger at backpressure for those variances alone.

I don't want to come across as bashing or disagreement. It is a very interesting topic. I was just sharing what I've seen. And I am one of the few people that have put back pressure sensors and standalone widebands in different locations to see this kind of data.

because I am apparently too greedy with spreading my rep and can't rep you again.

[andy_divers]

Someone send Tony a 4channel wideband controller to test this. This is simple enough to test and document on a TT car.

[Terry@BMS]

adv vs 4.9

Yay! Data! Thanks.

Let me know if I'm reading this Bosch 4.9 sensor chart right. There seems little deviation up to say 1.3bar. Jake says the DME is already mapped out below .6bar so let's presume that carries it up to 1.2bar. So what needs to be corrected is from say 1.3bar to 2.5bar, which for lambda < 1 is a fairly linear about 10%. So at say .800 lambda and 1.9 bar we'd have around a 1pt variance?

That seems to be in the ballpark with what I've noticed on the dyno comparing the dyno wideband to the OEM sensor readings. Although maybe 30% higher than what I've seen. So maybe the DME is already compensating for 30% of that difference. I'll poke through the dyno folders some more.

The ADV sensor seems to require more correction for back pressure but that correction is less dependent on the lambda reading?

[Terry@BMS]

Here is some anecdotal data. A comparison of a JB4 log and dyno AFR trace on our E92 with a single turbo at fairly high back pressure, same car with twin turbos, Payam's larger turbo single at likely moderate back pressure, and our MOTIV 135i with fairly high back pressure.

May need to right click and select "view image" for it to size to a readable point in this forum.

Andy, you've got a much better sensor suite setup on your car. Are you not getting a similar .75:1 variance > 20psi when comparing AFR from the DME to an external sensor?

The easiest way to compensate for this is probably just to presume back pressure and adjust the AFR targets per load with that in mind. That's what we've all done to this point. But my gears are turning on a built in JB4 compensation. Seems like a simple thing to throw in. I'll play with it.

[CKI@Motiv]

Yay! Data! Thanks.

Let me know if I'm reading this Bosch 4.9 sensor chart right. There seems little deviation up to say 1.3bar. Jake says the DME is already mapped out below .6bar so let's presume that carries it up to 1.2bar. So what needs to be corrected is from say 1.3bar to 2.5bar, which for lambda < 1 is a fairly linear about 10%. So at say .800 lambda and 1.9 bar we'd have around a 1pt variance?

That seems to be in the ballpark with what I've noticed on the dyno comparing the dyno wideband to the OEM sensor readings. Although maybe 2x higher than what I've seen. So maybe the DME is already compensating for 30% of that difference. I'll poke through the dyno folders some more.

The ADV sensor seems to require more correction for back pressure but that correction is less dependent on the lambda reading?

Not really. The thing is (besides this chart) the 4.9 can have a very large margin which im sure depends on the condition of the sensor. So even if pressure is corrected for you still have that 10-20% uncorrected margin. Theres a very big part of this that your missing besides sensor accuracy too. The V output changes for every lambda so if your telling the DME its a running richer or leaner than it is then it will over correct for it with the offset. The DME doesnt have a standard correction for pressure alone, theres about 5-6 other influences. On top of it all the aging, sensor depreciation, etc etc that we keep erasing every time we start the car is not helping the matter either.

I'm done with the discussion, You can try it but i doubt the results will be much better than what is currently available. Your only looking at 1 variable for 1 condition.

[andy_divers]

@Terry@BMS I will have to look. I don't remember ever seeing a full point variance and I'm WAY over 20psi of backpressure

[Terry@BMS]

@Terry@BMS I will have to look. I don't remember ever seeing a full point variance and I'm WAY over 20psi of backpressure

Thanks the more data the better. If your log to dyno data shows the different increasing as a function of back pressure as my data seems to show. It would at least confirm the theory.

[Terry@BMS]

Not really. The thing is (besides this chart) the 4.9 can have a very large margin which im sure depends on the condition of the sensor. So even if pressure is corrected for you still have that 10-20% uncorrected margin. Theres a very big part of this that your missing besides sensor accuracy too. The V output changes for every lambda so if your telling the DME its a running richer or leaner than it is then it will over correct for it with the offset. The DME doesnt have a standard correction for pressure alone, theres about 5-6 other influences. On top of it all the aging, sensor depreciation, etc etc that we keep erasing every time we start the car is not helping the matter either.

I'm done with the discussion, You can try it but i doubt the results will be much better than what is currently available. Your only looking at 1 variable for 1 condition.

It seems you're getting a bit defensive on the stuff and I don't really understand why. It's not like either of us is going to sell o2 sensors right? I'm just trying to figure out the best way to solve the same problem you're trying to solve. I appreciate the work you've put in to it so far.

From my perspective there are maybe 100 variables that effect the DME reading O2 sensors. But the DME is already aware and well programmed for say 98 of them. The variables that change when we locate the o2 sensor to the manifold are pressure and heat. The way I see it the DME will handle everything else.

So what we need to know is what effect pressure has on the o2 sensor reading, which appears to be a function of not only back pressure but the lambda value. Makes sense if you think about it. And what effect temperature has on the o2 sensor reading. For now we're just worrying about the pressure and those two charts, which I presume are from a datasheet on each sensor, are awesome for understanding that.

Do we know exactly what delta lp / lp(1013hpa) represents? I'm assuming it's the lambda error reading variance. Maybe it's not. But say it is, there is quite a variance from .5 bar/7psi, to 1.3bar/18psi. -25% to +7%. We can sort of back in to how much the DME is accounting for by comparing reported AFR to actual AFR. But it's always nice to see a connection between practice and theory. To work out the theory we'd need to know the DME's compensation constants. I'm not sure why the factory values are top secret lol but I guess I'll dig through the old DAMOS I have to see if I can get a context for it.