Load Glitch
 

I have e-mailed @Cobb Tuning; and am 99% sure some of the other tuners have as well concerning a fueling glitch that occurs when certain calculated loads are hit.

This graphical example is the best visual illustration I have of this happening, and would be happy to provide 1000's of logs, that span all he mazdaspeed generations, and all of the models that exhibit this behavior.

This issue has been known for a long time, and as more and more people utilize BT's, this is becoming a time consuming hassle that really needs attention imo.

I can tell you that I feel that all of our tuners are having to revise maps and spend additional and unnecessary time on proper fueling for each and every tune that exceeds 2.5 load.

Whatever table exists, or whatever load scaling we need access to, etc. This problem needs addressing, imo.

Again, I would be happy to provide countless logs across all generation Mazdaspeeds and Mazdaspeed models but this is getting really old.

Just for emphasis, this is the same graph with the addition of the resulting afr included.

As they say on FB, "pic for attn".

http://i.imgur.com/5pKkox1.png

Sounds about right to me. I've noticed this behavior ever since I started tuning the BNR.

@Rob@Hypnotictuning; and myself have experienced this everytime, sometimes multiple times when I have had a mechanical issue with the vehicle that was fixed. This is a LOAD of shit (see what I did there lol) and needs to be resolved by the tuning software people. Cannot agree more.

Rob, I really, *really* hope yours is the straw that finally breaks the back of whatever camel is stopping them from resolving this issue.

Not gonna hold my breath that they do anything about it tho. You know as well as anybody that Cobb has dropped us into maintenance mode and if they considered this symptom a bug it would have been given proper attention already.

Still, fingers crossed and all that!

Didn't ever really have to deal with it on the k04, only once or twice it happened. But I've definitely noticed it more with the BNR.

For now, because I don't really feel like dealing with it, I've just made sure to keep load below 2.5. Quite an annoyance honestly, and will only become more so as time goes on.

Cobb asked that I provide logs. I want to have guys who are self tuning and other tuners provide a few as well as to give them examples from self tuners, other tuners etc.

I've attached a classic example (as seen graphically above).
@TiGraySpeed6; @Vansquish; and @g00s3y; when you get a chance, can you attach some logs as well?

I thought the consensus was that after 2.5 load we are truly in OL and the AFR's will show how off the MAF cal actually is. Before that the ECU simply has hidden fuel trims?
I'll have to look when I get home, but I had some logs where I played with it and got the step to be basically non-existent while still hitting AFR targets below 2.5

Good stuff Rob.
Is this issue present in VT?

This is @kashbringer;s log from just the other night, trimmed to just the WOT portion.

Car is supposed to run an 11.6 AFR up to 4500 (cover knock in the spool region) then 11.8 to redline.

5th port kicks on at 3.8V, and as you can see, the MAF g/s flat line.

I will go back through some of his more recent logs and see if i have more, i just happened to have this one open when reading this thread.

Not as far as I know. Good to see u posting. :13:

Here is another example. Pump gas car I purposely avoided hitting 2.5 load on. One map (avoidance, the afr is nice and tight, the next map, targeting over 2.5 load, afr goes haywire).

(logs attached)

http://i.imgur.com/TYhrthN.png

I'll post some logs if I can dig one of my 2.5+ load ones up. I only ran the relevant map for a little while, as it was annoying to have the herky-jerkiness that this caused.

Edit: Clearly I also need a bigger intake.

3" just isn't enough (that's what she said!)

I'll have to make a new map to show it. My current one I hit a max of 2.45 so I don't have to deal. Should be able to get a log with it tonight or tomorrow.

Here we go, both logs, right around 42-4300, hits 2.5 load, and along come the 12.35 AFR, i'm targeting 11.76.

Have you tried adjusting your MAF table to correct in that region?
I.E. Changing MAF voltage @ 3.9 from ~272 to ~285.
That would be a VE of ~107% which lines up close to what I see on other people's cars. I've seen as high as 120% in the spool region.
Now at 4.18 volts, your curve is closer to what I'm used to seeing with VE dropping off in the higher rpms closer to 100%.

Again, I think 2.5 is just the point where we go to true OL without any hidden fuel trims and your MAF cal is off by a bit there.

@chaser27; it can be tuned around, but it is a bandaid, and honestly, is very time consuming, @ least imo.

I'm concerned about it for more reasons than the extra time and revisions it takes us to tune the car properly.. I think that fueling the car properly without any bandaids needs addressing.

Versatune has had it done now for months.

please look @ the above graphs a bit more closely.

There are times, given that calculated load is rpm dependent that a given maf voltage hit fueling dead on, and at another rpm, is off by a ton.

Rob, I looked at the haywire log and did the same thing to yours. Assuming you are looking for an 11.8ish AFR at 4500 rpm, your MAF cal needs to go from ~290 -> 264 g/s @ 4.27 volts. Doing a quick calculation (assuming sea level, because I don't know where the log was taken) will put you at 106% for VE, which is basically the same as ^^ in the same rpm range.

It's just odd to me that the fix for two different cars, one lean and one rich would be moving them both to a similar volumetric effiency assumption in the MAF cal.

@Justin@Freektune; @ms3blackmica; and I can't remember PD's handle, you guys have any logs or experience in this phenomenal? I remember seeing a thread or two about it with Stratified mentioning it, and I think the "solution" back then was to keep massaging the MAF curve. If we have more tuners post up, we'll have a stronger case to present.

A proper fuel curve does not drop down 10%. move up 10%, back down 10% etc.

The blue line is a corrected fuel curve in the graph below. The curve is smooth and predictable, as should be, up until the load calculation is 2.5 and above, then drops out a good % and returns to the same smooth, predictable pitch.

http://i.imgur.com/5pKkox1.png
Do you have some sort of access to a load scalable VE table to make VE adjustments that I'm not seeing?

I don't think it's jumping around 10%, I think the ECU is turning the hidden fuel trims on and off.
The only thing I'm using is the lazy man's maf builder thing I built a while ago

http://www.mazdaspeedforum.org/forum/foru...f-tool-186955/
It has a tool in there to estimate airflow in higher RPM's where we don't have LTFT or STFT readings. I'm plugging in the data from the logs and using the (target AFR/actual AFR) ratio to adjust the MAF reading and figure out what VE that would be... if that made sense.

There is a few threads that discuss the work around and working with the MAF to get around this "glitch". Not sure if this is the right place to discuss it again.

Well is it really a "glitch" or is it just where it goes to a true OL schema? That's all I'm asking.

The black line, which is fitted to the blue line (slight errors), below, is a 3rd order polynomial (and corrected, much like what your ez man's calculator does, but manually). That is what the stock curve is as well.

Changing the intake maf i.d. may change the scale and curve, but @ a smooth and predictable rate.

I feel as if there is a table in the ecu that is fucking our couch. LOL.

http://i.imgur.com/Zklu9NR.png

Yes, when you look at the "corrected" MAF curve, you see a section that doesn't line up with the rest of it. I think you were one of the people that first realized that the ECU uses hidden fuel trims in OL. If those go away above 2.5 load, you will be able to see a true "corrected" MAF curve.

I'd be willing to bet that the corrected MAF curve lines up with your MAF table pretty well because of the hidden fuel trims. If you adjust the MAF curve for the 2.5+ load section, I bet it will still hit targets below 2.5

Rob did you ever take a look at what I posted in the Versatuner thread on this subject? totally different platforms, but still mazda and possibly works the same way....

I saw you mentioned that the baro comp tables had an effect on the fueling.

I did make an attempt to change what available tables we have in terms of baro correction without much luck.

(i.e. I ganked your idea already and unfortunately, no dice. Thank you though, we can use all the input we can get).

Chaser, I appreciate what you are saying, but the blue dots are corrected values both before the drop out and after.. the drop out is the only section of 2.5 load and above.

I understand that they are all corrected in the blue line, but I think the rest of the line already has hidden fuel trims applied to them in the background, so when you correct them, it doesn't change them. The only part that changes is the part with no hidden fuel trims. I'm not very good at explaining myself...

I changed my MAF curve by something like 15% when I was testing this a year ago or so and my AFR's were still spot on. There are hidden fuel trims affecting the AFRs in the log, so there isn't anything left to correct when you go back and do it from the log; leaving only the data above 2.5 needing correcting.

What we may need to do is find an acceptable VE curve for different rpm's at max load to get close for everyone. Using the data from your logs that were running rich above 2.5 had you at ~114% at 4500 rpm. Using the target/actual ratio for AFR brought that number to ~106. The other guy was running lean and correcting his also brought his VE to ~106. When I was doing more tuning (still not as much as you guys) I found everyone's car had a very similar VE curve when it was all said and done.

I actually haven't tried doing anything with this maf cal yet to be honest. When I had the CS 3", I used the calibrations from the "bigger intake calibration" thread in Stratified's sub-forum, and it was spot on without having to change anything. I realize that this one may not be perfect and needs to be worked on, but I just wasn't in the mood to mess with that recently. I just went to the 3.5" a week or two ago.

I'm going to check out your tool (giggity), because I'm lazy, and it's seems appropriate. Make some adjustments and see what happens.

Here is a log with my previous setup to see how versatune reacts with load above 2.5 ....
I target 12.00 and at this setup i have 4 direct meth nozzles (d03 from devilsown)

With all due respect, it matters little what semantics you want to use, or how Cobb goes about working with us in terms of allowing us access to the tables we need, but, this thread is a request for Cobb to either EXPOSE HIDDEN FUEL TRIM TABLES THAT OBVIOUSLY STOP @ 2.5, or fix a glitch...

There is a table that is affecting the fueling, that we don't have access to or scalable load for it and that is the point of this thread.

Thank you for your responses, and I have been tuning around this issue for a long time and am growing tired of it.

What VE table are you adjusting @chaser27;
Because, as far as I can see, the VE table that we have available to us is only scaled (read only labeled no edit possible) to 2.0 load

Please note the date. I'm well aware that there are active trims in open loop.

http://i.imgur.com/PNERYui.png

@chaser27; @rfinkle2;

So I did adjust the maf, it did seem to get things to fall in line. It isn't perfect, but I still need to adjust it a bit in places.

http://i62.tinypic.com/5zqfi9.jpg


EDIT: Forgot to upload the log.

Can you post a screenshot of your maf curve?

http://i61.tinypic.com/15q7nm1.jpg

I agree, we have enough people going BT/built motor now and hitting 2.5+ loads, that we should be able to access those tables. I couldn't remember if you were still active when we tried to figure it out last time. My apologies if you were.

As far as the VE numbers I'm coming up with. I don't have access to a different table that goes beyond 2.0 load. I'm using the data from the logs to calculate an estimated VE. I can show you guys an example of the calculations if you wish, but they are built into that tool I posted earlier if you want to use them.

Did you use the tool to build that? When looking at the curve, there are a bunch of cells that aren't really in line with the others... I'm posting a pic of the delta between each cell to show what I mean.

Each cell will be increasing by 11 or something and then one will only increase by 5, then back to changing by 11. This will cause some oscillations in your AFR. The ECU does better with a smooth curve in my experience/opinion.

Also, you may want to keep an eye on your fuel pressure, it's dropping to 1600 psi range every now and again and seems to possibly correlate to when your AFR goes to 11.9-12.0.

I made a smoother version of your MAF I'm pasting below... It is a little higher down near idle, but for the most part it is withing +- 2% of your current one. Just without the bumps. I'm pasting a pic that shows the percentage difference vs. yours as well.

I used the now defunct MAFiNATOR to get my curve. I know it isn't perfect, so I usually fix it up after. But I have been a bit lazy lately (1 year old son has his top 4 teeth coming in, apparently in a growth spurt, and just recently decided he doesn't like to sleep much, burnt out from the BNR install fiasco, in process of looking for a house...) and did not do anything extra to the final curve I got from it. I agree completely that the ECU would do better with a smooth curve. I need to start using excel more and graph logs more to really look at whats happening, that's something I should have caught, or at least realized when I was doing it.

Fuel pressure I have always seen a minor dip here and there. I have noticed it happen more in the lower/midrange, as opposed to up top. I'm also only targeting 1700psi at 2.0+ load. I was actually just sticking with the stock 1669 for a while testing to see if it made a difference, another member brought up the idea, I decided to try. Things seemed to be fine, but I figured just targeting 1700 would be fine, as I didn't notice much of a difference in actual HPFP press. when changing from 1749-1700.

Thank you for taking the time to actually make another curve, you did not have to at all. But I will copy that one in now, flash it and I'll see if I have some time to cruise around for a little today, let that set in, and get a log tonight hopefully.

Back on topic, are there any other main problems with ATR that "hinder" the ability to tune correctly? I know that some tables are mislabeled (would be nice to get fixed), some don't seem to work correctly (boost comp by gear seems to randomly decide when it wants to kick in), we don't currently have access to all the tables we could be using (not at fault of Cobb, I could only imagine the trouble people would encounter if they had access to everything, but I do think there are more that could help the majority if they were "released"). Aside from that, any other real problems or glitches with ATR?

We really need some sort of boost by gear function tbh, one that controls the solenoid differently in individual gears.

Limiting tq by cutting down the APP can heavily bog a BT car, and I don't like the idea of having the throttle plate closing and opening when aux fuel is being sprayed pre throttle body.

There are multiple strategies that we could use, but all rely on the APP vs. DBW translation, which is pretty much a mystery as well.

Between us (lol) the boost by gear function does work (but only in 1st and 2nd gear) i.e, the boost comp 1st and 2nd gear is active, but the remainder are not.

Lol, speaking of the boost comp tables, the y-axis, I'm guessing that should be APP not RPM? Considering the x-axis is RPM, the y-axis being APP would make a lot more sense.