RomRaider

It's my understanding that the stock O2 sensor on 32-bit single-scroll Subarus is capable of being accurate over a pretty wide range, but its location in the up-pipe causes it to get inaccurate when under boost. I'd like to try to characterize that inaccuracy, to see if some Excel magic can produce somewhat-consistent and somewhat-accurate AFR values. Basically I want to see if the stock sensor's deviation from a wideband sensor is actually a predictable function of UP pressure.

So, does anyone have, or is anyone willing to create, logs that show the following columns:

* AFR from the stock O2 sensor
* AFR from a wideband O2 sensor
* Manifold pressure (preferably absolute, but relative is OK if you know your ambient pressure)
* EGT (wasn't in my initial post but it's a good idea, thanks JSarv)
* RPM
* Time

(Anyone have other parameters in mind that might be relevant?)

For practicality I'll assume that UP pressure is (roughly) a function of manifold pressure, at least when manifold pressure isn't changing too much.

I am sure there are plenty of people getting ready to type "just get a wideband" and I certainly will. But it's not going to happen for a couple months (as part of an UP/turbo/DP install) and I think this data might be interesting to experiment with in the meantime. I do not expect this to produce a formula that would render a proper wideband un-necessary, but I am curious as to whether the stock O2 sensor readings can be corrected with well enough to be interesting.

So, if you've got logs, I'd love to se them.

Thanks!

I think the biggest problem is, is the variation in up-pipe pressure. Because there will be more pressure prior to spool than during spool.

I think the biggest thing needed would be a pressure gauge hooked up to the up-pipe to provide a constant basis information to base differences of readings. For instance if its off .4-1 during spool but off .9-1 while in full boost but off 1.0-1 on boost taper, there has to be a variable to base it off of. Manifold pressure will probably not be a good variable, because back pressure will be different at different turbine speeds....


Another one that would be very important, egt... If egt's were high there is a chance of the sensor being off a bit.

I'm not saying its not possible, but egt and up-pipe (relative) pressure would be needed to make a good solid calculated variance.

Yeah, I agree:



I'm guessing that I won't be able to get repeatable results during spoolup, when the relelvant factors are all changing rapidly. If this works at all, it will probably only work when the relevant factors are all stable or changing slowly.

If UP pressure gauges were installed from the factory I'd certainly ask for that instead. But my goal is to see how accurate of an AFR reading we can get (or can't get - this produce nothing useful), using only the factory sensors.

EGT certainly couldn't hurt. Seems to me that EGT downstream of the turbo is going to vary pretty much proportionally with EGT upstream, at least after the spoolup, so I'm guessing that it's not going to be a big factor... but it would be good to have logs with that just in case I'm wrong about that. I'll update the first post.

Thanks!

Well, my 02 (16bit) the front o2 follows the wb02 until about 1.3 load (in that area) and then drops straight to 11.35 while my wideband drops a bit slower and reads the true readings while my narrow band stays right at 11.35 the whole time...

So under boost really messes with it, I assume so does heat (makes a slightly rich reading)


Good luck keep us informed if you get anything figured out!!

Jerod

I don't understand why we can't just relocate the stock front O2 sensor to a downpipe bung, then get some Merchgod lovin' in order to broaden the scaling column for the sensor so we can scale it more precisely and farther down than just 11.3:1. I think this would make a great project for someone willing to experiment! Hint Hint!

I think it would, but there would be a slight delay over stock values. There has to be a reason for its location...


I think its a good experiment.

This has already been discussed. You should just add to the original thread.

topic1398-60.html

Here's root's 04 STi with the front O2 sensor moved to the JDM position.

I remember that thread, but I want to go in a different direction now. The "Front O2 sensor limit" thread established that the stock O2 sensor could be reasonably accurate down to about 11:1, if relocated downstream of the turbo, which is great. After 8 pages I figure that thread has reached a fine conclusion.

That diagram is about the correlation between a relocated stock O2 sensor readings and wideband readings. My hypothesis (and the subject of this thread) is that the signal from the stock sensor in the stock location might be correctable, by a function of manifold pressure, to yield useful information under boost.

To figure that out, I'd like to study some logs containing AFR from the stock sensor, AFR from a wideband, and MAP, and preferably with EGT and RPM as well. I'm going to move my stock O2 sensor into my downpipe when I swap my DP, so my car will never be able to collect this information. I was hoping that some folks here might be interested enough to provide some logs, though.

With luck, it might be possible to sanity-check AFRs in stock vehicles, within a still-somewhat-useful margin of error. At worst, we'll find that, no, we just can't get useful information from the stock sensor under boost. I'm hoping for the former, of course.

But keep in mind this is because its only scaled down to 11:1. Check out the O2 sensor scaling in romraider/ecuflash.
Merchgod could possibly add more scaling to the map for us so we could get it down to say... 10.0:1

You can rescale it yourself. And most have a duplicate axis value anyway.

I've done that. There's a graph floating around. When the stock front O2 reads 11.03:1 then the wideband is showing around 12.3. At least, that's what I thought I saw. If you can't find the graph, I'll search through my photobucket later tonight.

I'm not sure if there is a mA limit to the front O2 sensor. Changing the lowest limit may damage the sensor.

But, that is not a true test of the sensor compared to an aftermarket wideband sensor/controller.

I want to see someone with two bungs right next to each other in their downpipe logging the two sensors. I bet the results will be strikingly similar.

And Merch:

Yes, we can rescale it, but we need a larger axis for more precision imho. Hack the table for more cells in the axis! We need double the amount to make it pretty precise so that we can get some scaling down in the 10:1 area.

If you go back to the original thread, you will see that the graph that I posted six above was taken from root who has two bungs... one right after the other in their downpipe.

That graph, by itself, won't help... Do you have the log that produced the graph? Does the log include manifold pressure? If so, that would be great.

What would manifold pressure tell you? You would need to know exhaust gas back pressure which is impossible since it rises rapidly with rpm even as MRP decreases.

In any case, it's not worthwhile to use the front O2 sensor in the stock position unless you are tuning E85 or something that is tuneable with lambda > 0.8.