Front O2 Denso Planar sensor scaling
 

Rescaling the front Denso Planar O2 sensor:
Our stock ROM table looks similar in shape to the Bosch 4.9 mA vs lambda curve with a scaling factor of 0.61.
I've applied an extended scale and tested down to AFR = 10.5 against a PLX running a Bosch 4.9 sensor that I installed last week (lambda range 0.68-1.38).
I think the Denso planar sensor is performing quite well with the extended scaling, will test down to AFR = 10 next. Info cross-posted in mad_sb's thread in another section.
Rescaled curve and correlation graphs:


***Updated sensor data in post #15 (Kodename47) and post #16 (my revision)

Interesting results, I'd planned on doing something similar with the Innovate LC-1 I have but won't bother based these results and other feedback.

Which A01G RR defs are you using? I only have a single table defined for Front Oxygen Sensor Scaling while you have A & B?

Cheers. Which def are you using?

I'm using Kodename47's last update which I thought was the most complete one for A01G:
http://www.ft86club.com/forums/showp...&postcount=336

Bastardized version of Kodename47's with some other tables that I found along the way derived from A00C and A01C.

PM me if you want a copy.

I'm not sure why you need the 2nd sensor scaling? Only changing the one in the definition I have works fine. @ztan?

Something to note is that our rom and front sensor has a number of climate based compensations that you will not have on a traditional wideband which might make correlation difficult outside of a "close enough" range.

could that be related to failsafe or limp home mode ?

No idea whether it is needed or used, but it is there. The trouble with this ROM is that there is a lot of indirect referencing and there may be a part of the code which uses the second table (?limp home) which you don't find unless you step through the disassembly or find at the inopportune time when your car goes into limp home and slews all your modified sensor scales.

No idea why the MAP scaling values are in 3 places or RPM/load limiters tables are in 2 places, but they are and I've changed them all.

Agree - there is a pressure comp table which I understand affects the planar sensor significantly and may skew things under boost.

How can we verify the accuracy of a traditional, non-compensated wideband like a LSU 4.2 or 4.9, esp under boost?

I'm just going to point out that yes accurate wideband info is ideal but it's not the end all of tuning. I typically use a tailpipe wideband as a suggestion when tuning cars. However my personal car has never ran a wideband. And I run 18psi on pump currently.... Previously I was running up to 24psi on e and never really cared about the slight inaccuracy of the oem O2...

Although in regards to op I appreciate the post! It's good to see more data like this. :) I had at one point plotted my lm2 vs a stock O2 and in the 11-12afr range it was fairly close and "good enough" for me at the time.

Rememeber perfect afr isn't everything. Perfect reading simply helps when calibrating maf / ve tables and such.

After running my car on a dyno recently I have had a chance to compare my O2 scaling, which was provided ECUtek. Below is a graph showing what my O2 "AFR" output was, the dyno is input in 500RPM increments so isn't 100% accurate, I have already removed 0.15 from it to take into account the tailpipe location and the cat. Note that at the top end my O2 limit is 10.9, which is why it flatlines.

I would also like to note that my O2 readings are always consistent, but this proves they have been consistently reading richer and therefore I running leaner than it shows. Because of this consistency I have managed to emulate the dyno output for my sensor.

The top graph shows the basic dyno output, my original scaled O2 AFR output, the same output using @ztan's scale from above and how it would look using my new derived scale. The lower graphs show the current vs lambda output curves of the various different scales.

http://res.cloudinary.com/lagbv048w/...s/Front_O2.jpg

Now that's not to say that @ztan's scaling is wrong, however it definitely wouldn't suit my setup. I also can't currently re-do the rich end of my scale accurately due to the reading hitting the limit, although it should be rather close I hope. I'll test it over time to see how well it does.

So this is for the stock O2 sensor, correct?

Also, how do you suppose the ECU utilises the "Rich Limit" value?

Yes, it is. It seems to work on some cars, similar to the ECUtek provided scale, which will read a little leaner than this. That was also done by correlating to 2 wideband sensors.

The rich limit is literally just the value that the ECU stops reading any richer.

Thought I'd update this with my "finished" scale:
http://res.cloudinary.com/lagbv048w/...t_O2_Scale.jpg

Thanks, that has prompted me to do a bit more tweaking. I found, as you did, that my last scale at the top of the post read a bit rich. I tried your finished scale against my PLX Bosch 4.9 and it made the Denso sensor read a bit lean compared to that reference. I also had significant STFT oscillation at closed loop cruise operation, possibly related to the sharper elbow on your last curve under AFR 12.5.

Latest scale attached with sensor correlations and summary curves of different scalings; this one is working pretty well for me with good closed loop and WOT tracking of my Bosch 4.9 wideband, there are some hysteresis issues as the Bosch wideband responds much more quickly to transients.

The MY17 seems to have improved scaling. It's going to be hard to confirm at this point of the sensors are the same however the scaling is quite similar to @ztan's. I've added the graph below

http://res.cloudinary.com/lagbv048w/...es/MY17_O2.png

Sorry to bring up such an old thread, but I was wondering if non-FI folks like me would benefit from using @ztan's sensor scales specifically for more accurate logging in OL situations?

Is this a bad idea if I don't have a wideband sensor to correlate against? It wouldn't directly affect the operation of the engine much (0 lambda is still 14.7 AFR), but would affect any future changes I make.

Been using that scaling in my car for years, it just allows you to see and log afr llwer than 12.17, its handy when scaling maf or logging tunes that run richer targett afr than 12.17. oft tunes for example target 11.8, but thier tunes has rescaled o2 to read that but not as low as 10.

Basicly their is no harm in running this scaling

Exactly the information I was looking for. Thanks!

So I went to go do this in my tune using romraider and Kodname's OFT definitions and there is only one front O2 sensor scaling. It doesn't indicate A or B like the screenshots above. Is this expected? It's the same for the regular OFT definitions as well.

Im also working under the assumption that I can just match my scale to the screenshot, so let me know if I am wrong on that.

yes only one occurrence listed in kodes defs.


yes copy the entire table