Stock wideband useless? Not quite.

[Boostin1657615274]

^ All I can do is chalk it up to coincidence. Maybe your function and the SAE load function are pretty close for the type of driving you are measuring. I know that's not a very satisfying answer, but I don't work for Subaru.

 

And that's why I came back and said that the difference between the two load formulas is trivial for our purposes. Me getting pedantic about load calculations (saying that the g/rev calculation isn't perfectly accurate and therefore useless) is IMO like saying the stock wideband's readings are useless. I humbly acknowledge the potential hypocrisy so I'm not going to press the issue. In the future though I would just go off whatever the ECU reports and save yourself some trouble.

 

FYI, here is a list of the OBD II PIDs that the 2009 Rx-8 ECU can report. This is from the 2009 Rx-8 Service Highlights document, on board diagnostic section.

 

http://www.legacygt.com/forums/attachment.php?attachmentid=75451&stc=1&d=1257102632

 

I imagine that "load" is SAE calculated load, the measure of how hard the engine is working. And "absolute load" is what I was referring to the most, the measure of pumping efficiency.

 

 

I'd like to come back to a point I made earlier about wideband self diagnosis capability. I have an Innovate LC-1 on my single turbo Rx-7. It gets the job done. But the LC-1 is nowhere near as good as the stock LGT ECU at figuring out whether the sensor is working right or not (the AEM wideband is even worse). Let's look at a list of the error codes that the LC-1 has.

 

http://www.legacygt.com/forums/attachment.php?attachmentid=75447&stc=1&d=1257102632

 

You can see how crude this is. It's not really testing much, just telling you if there's an open or a short. Now let's see what the factory ECU tests to make sure the front O2 sensor is working right. What I did here was cross-reference the LGT service manual diagnosis section, which tells how to troubleshoot O2 sensor codes, with the 2009 Rx-8 Service highlights document, which gives us an idea of how these types of diagnostic tests are performed inside the ECU.

 

The Rx-8 and the LGT both have factory widebands and have very similar O2 sensor codes that mostly share the same numbers, like P0030. The 2005 LGT diagnostic code list can be found starting on service manual page EN(H4DOTC)(diag)-71. Here are the Rx-8 factory wideband DTC's.

 

http://www.legacygt.com/forums/attachment.php?attachmentid=75455&stc=1&d=1257104106

 

http://www.legacygt.com/forums/attachment.php?attachmentid=75454&stc=1&d=1257104106

 

It would be reasonable to assume that, since the LGT and the Rx-8 have similar O2 sensor systems, there is a lot of similarity between the way they do their diagnostic tests. Because the factory O2 sensor control is integrated with the rest of the powertrain management, the ECU can do tests by comparing expected values under certain driving conditions to measured values from the sensor.

 

The AEM's self diagnosis and error code list is simply pitiful. Either you get a reading on the display and cross your fingers, or you get a "---" . Real helpful guys, thanks. I'd take an AEM wideband over no wideband at all, but it's by far the cheapest and crappiest controller on the market.

[NSFW]

Knowing grams per revolution doesn't tell me how full the cylinder is. It's just a measure of mass flow over time. The units don't work. It would have the exact problem that everyone here is pointing out. It doesn't tell us the pumping efficiency (absolute load) because we aren't incorporating the size of the engine into the load calculation. And it can't be telling us how hard the engine is working (calculated load) because calculated load cannot exceed 100% or 1.0 on any engine according to SAE. So grams/rev is a bastardized unit from a theoretical perspective.

 

Load, as defined by the SAE for the purpose of the PID, sounds like another way to express volumetric efficiency. In the Subaru tuning community, "load" means g/rev because that's what the ECU uses internally. LBGT's logs didn't agree by coincidence. They agreed because we're not logging the SAE's load PID. We're logging grams per revolution.

 

There is an SSM parameter called "Engine Load (Relative)" which is probably the same value used for the load PID, but that's not one that I've seen anyone use. When people in the open source Subaru tuning community are interested in volumetric efficiency, we generally load MAF, IAT, and MAP, and do the math from those. If the "Engine Load (Relative)" parameter is in fact the SAE's "load" PID that might simplify logging for some scenarios (AVCS tuning, for example). It's worth looking into.

 

However "bastardized" it may seem, g/rev is very useful in practice because it's easy to derive from MAF and RPM and it's one of the inputs to the algorithms that govern fuel injection and ignition timing. Give g/rev and the desired AFR, simple arithmetic tells how many grams of fuel are needed. From that quantity, and the injector characteristics, the ECU determines how many milliseconds to send power to the injectors. Ignition timing is governed by tables that just have g/rev on one axis and RPM on the other. Since so many the ECU's fuel and timing tables are defined with g/rev as one of the axes, it's a quantity that we find very interesting.

 

But anyway, about that O2 sensor...

[LittleBlueGT]

Now that we know that the load we care about is g/rev, back to OEM O2.

 

I will soon be tunning for alky again, once done (AFRs measured at around 12:1) I will be putting a brand new OEM O2 in the DP. I will be able to compare then. (My stock O2 died yesterday, I have heard of others having problems with them in the Perrin headers, so the new one is going right to the DP)

[Boostin1657615274]

The reduced exhaust backpressure will make the sensor read richer, but the increased distance from the exhaust ports make it read leaner. Having a new sensor will eliminate the effects of degradation though, which is good.

 

Without proprietary manufacturer data (current and AFR functions) and good instruments, it would be hard to quantify how much the reading changes from the backpressure effects (less backpressure = richer reading), the distance effects (greater distance = leaner reading), and the temperature effects (lower temperature = richer reading to some extent). Hence my point that you can't tune by widebands alone and must take their readings with a proverbial grain of salt.

 

According to the Bosch document (which is all we have about widebands really), changes from standard lab test conditions result in the following:

 

More absolute pressure = more current = leaner reading

 

More temperature relative to controlled lab conditions = more current = leaner reading

 

Distance effects are not specifically mentioned, but anecdotally we know that locating the sensor farther from the head results in a leaner reading.

 

Without being able to quantify though it can be a little frustrating. It's kind of like how we all know what basic factors go into someone's FICO credit score (debt vs credit available, paying bills on time, number of recent credit checks etc) but we don't know the EXACT mathematical weights.

[NSFW]

On a related note, last night I finally did some logging to compare with WBO2 with my post-cat stock O2 sensor. Unfortunately the only computer I own that has Excel installed, is toast. I'll post scatter plots when I'm able to.

 

Computer Win7'd, Excel reloaded, and...

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/DualO2.png

 

Again, this is with the stock O2 sensor relocated to the downpipe.

 

Not off by much. The sensor scaling can be adjusted, so I'm sure it's possible to bring it closer.

[LittleBlueGT]

Computer Win7'd, Excel reloaded, and...

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/DualO2.png

 

Again, this is with the stock O2 sensor relocated to the downpipe.

 

Not off by much. The sensor scaling can be adjusted, so I'm sure it's possible to bring it closer.

 

 

At 11.5:1 and at 12:1 AFR how much do the sensors disagree? (%)

[NSFW]

At 11.5:1 and at 12:1 AFR how much do the sensors disagree? (%)

 

Looks like about 2.5%, which works out to a bit less than 3 tenths.

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/DualO2-Error.png

 

I'm pretty sure that plot could be a lot straighter if I make the corresponding adjustments to the O2 sensor scaling.

[bugblatterbeast]

 

You can see here that the higher the exhaust pressure, the more it increases the current signal. The more the current signal increases the leaner the sensor is going to read relative to what it would read with no pressure.

 

Except that you've got it backwards. Yes, the current increases in magnitude with increased operating pressure, but no it doesn't cause a false lean reading if lamda is less than 1. As p increases, so does the correction term p/(k+p) * (k+p0)/p0. if lambda is less than one (as it should be under boost) the sensor pump current is negative. the correction term is a multiplier so the sensor current becomes more negative which makes the ecu think that the mixture is actually richer than it really is. the only time your statement is correct (more pressure causing leaner readings) is when lambda is greater than 1 and p >p0, which would have nuked the motor.

 

looking at it intuitively, it makes sense. if you up the exhaust pressure, for the same mixture, the pump cell has to work harder to keep the reference cell happy as there's a larger amount of excess fuel to be burnt at the sensor surface.

[bugblatterbeast]

The reduced exhaust backpressure will make the sensor read richer,

More absolute pressure = more current = leaner reading

 

More temperature relative to controlled lab conditions = more current = leaner reading

 

 

not under all conditions. more pressure means more current, yes. but more current only means leaner if you are at lambda > 1. at lambda < 1 more current means richer. the only point that should remain static as a function of pressure is lambda 1. when you are lean, more pressure makes the reading look leaner. when you are running rich, more pressure makes the reading richer.

[bugblatterbeast]

^ All I can do is chalk it up to coincidence. Maybe your function and the SAE load function are pretty close for the type of driving you are measuring. I know that's not a very satisfying answer, but I don't work for Subaru.

 

g/rev is probably used because it makes sense for fuel injection calculations. the ecu needs to know how much fuel mass to inject in every cycle. the injector open time calculation is more direct from a g/rev perspective.

[LittleBlueGT]

Computer Win7'd, Excel reloaded, and...

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/DualO2.png

 

Again, this is with the stock O2 sensor relocated to the downpipe.

 

Not off by much. The sensor scaling can be adjusted, so I'm sure it's possible to bring it closer.

 

Was this done at WOT?

[NSFW]

Some of it is, but a lot of it is part-throttle. If I just did WOT stuff there would be a lot of 11.X and 14.7ish, and not much in between.

[LittleBlueGT]

Computer Win7'd, Excel reloaded, and...

 

http://i193.photobucket.com/albums/z151/Legacy_NSFW/Tuning/DualO2.png

 

Again, this is with the stock O2 sensor relocated to the downpipe.

 

Not off by much. The sensor scaling can be adjusted, so I'm sure it's possible to bring it closer.

 

 

My findings of the O2 in the DP is supporting your findings. FWIW the lowest current I logged on my O2 was -1.38 Ma.

 

Here is what I started with and ended up with for a new front O2 scalar (top is new bottom is OEM):

 

http://i248.photobucket.com/albums/gg194/littlebluegt/O2scalaroldandnew.jpg

 

 

Here is the OEM scaling graphed out:

 

http://i248.photobucket.com/albums/gg194/littlebluegt/O2scalar.gif

 

Here is mine now scales out:

 

http://i248.photobucket.com/albums/gg194/littlebluegt/NewO2scaling.gif