Discussion regarding power limitations of the MS3

[JMEngineer]

Originally Posted by JMEngineer I have someone looking at my ecu this weekend. He probably won't have enough time to find anything useful since I need my ecu back for my car to run but he used to be an ECU programmer for Nissan so he knows a few tricks they use to limit stuff.

^Time for an update on this... He didn't have nearly enough time to reach any solid conclusions but he did say a couple things I thought I should share and also a lot of stuff I didn't understand at all.

Basically he thinks the ecu has some kind of direct hard limit for the power/torque curve or something closely related to it. There are places where multiple calculations are performed and checked against each other, then the ecu takes either the max or the min. Apparently a logging device could somehow be reporting one value for any given output when in reality the true output is a different value, but only when certain conditions are met, i.e. torque is above a certain level. He said we need to actually measure the throttle plate position, fueling, and ignition timing directly at the engine to be sure our logging devices are accurate.

He directed me to a patent, which he said shouldn't be related to our issues but might help understand what the ecu is doing. I thought it was pretty confusing once you get past the abstract but I'll go ahead and attach it for anyone who wants to take a look. I did find it interesting that the patent is owned by Ford - perhaps a more detailed patent search could reveal exactly how our power is limited.

[cld12pk2go]

Originally Posted by JMEngineer ^Time for an update on this... He didn't have nearly enough time to reach any solid conclusions but he did say a couple things I thought I should share and also a lot of stuff I didn't understand at all. Basically he thinks the ecu has some kind of direct hard limit for the power/torque curve or something closely related to it. There are places where multiple calculations are performed and checked against each other, then the ecu takes either the max or the min. Apparently a logging device could somehow be reporting one value for any given output when in reality the true output is a different value, but only when certain conditions are met, i.e. torque is above a certain level. He said we need to actually measure the throttle plate position, fueling, and ignition timing directly at the engine to be sure our logging devices are accurate. He directed me to a patent, which he said shouldn't be related to our issues but might help understand what the ecu is doing. I thought it was pretty confusing once you get past the abstract but I'll go ahead and attach it for anyone who wants to take a look. I did find it interesting that the patent is owned by Ford - perhaps a more detailed patent search could reveal exactly how our power is limited.

I have often noticed that the logged "Calculated Load (%)" sometimes flat lines at a value that seems rather unlikely consistent while my the ECU seems to follow my boost target curve.

Like here from yesterday afternoon:



and here:



Note peak MAF voltage was 4.66 = 279.74 g/s on this run.


Here is from yesterday morning:


Obviously trying to limit load via taking WGDC to 0.


FYI, my TRL tables look like this:



Boost target tables:

Originally Posted by socks i would go too far trying to richen the car up, and i'd have to start all over. kurt (subparpunk) had it dialed in in 2 tries. Get a second set of eyes. what size maf do you have? with my stock maf housing, i ended up around 390 at 5v. dustin, gmail. i needz ur stock terbow

I am at 352.64 g/s at 4.99v. Stock MAF.

[djuosnteisn]

Great info, but I don't see the signature fueling plateau that my logs show.

My maf is 3.25", and i can cal it in 3 maps at most with lower boost levels (18ish or less). It's entirely straightforward.

But when I up the boost beyond a certain point, it just goes leaner and leaner, and that's when i end up in cave-man land.

[Lex]

For cld12, fuel injected amount doesn't get near what you see on your car Dustin. He's not going over 5500 mg/stroke whereas you're almost at 8500. That's a huge fueling difference.

cld12, how many g/s are you seeing calculated from your MAF in those logs?

Edit: As soon as MAF gets above 350-380 g/s it seems to just pin fueling. Dustin, if you were to throw your stock MAF on there, would you be able to run 20psi or no?

Originally Posted by JMEngineer ^Time for an update on this... He didn't have nearly enough time to reach any solid conclusions but he did say a couple things I thought I should share and also a lot of stuff I didn't understand at all. Basically he thinks the ecu has some kind of direct hard limit for the power/torque curve or something closely related to it. There are places where multiple calculations are performed and checked against each other, then the ecu takes either the max or the min. Apparently a logging device could somehow be reporting one value for any given output when in reality the true output is a different value, but only when certain conditions are met, i.e. torque is above a certain level. He said we need to actually measure the throttle plate position, fueling, and ignition timing directly at the engine to be sure our logging devices are accurate. He directed me to a patent, which he said shouldn't be related to our issues but might help understand what the ecu is doing. I thought it was pretty confusing once you get past the abstract but I'll go ahead and attach it for anyone who wants to take a look. I did find it interesting that the patent is owned by Ford - perhaps a more detailed patent search could reveal exactly how our power is limited.

I do not doubt that there is lots going on and in the back of my mind I always have the thought that throttle, spark, etc may actually not always be as reported. One thing CPE seems to have shown is that the limitation is in the ECU. Which makes a fix much simpler than having to adjust flow on the head for example.

[djuosnteisn]

Not sure bout the stock maf. In a couple weeks i can always put the stock air box on some how, lol. Just to see i guess.

[Lex]

Bump for more logs and data or input.

[GQ_WhiteMS3]

I'm currently back to stock tune ... only Intake installed for the winter ... would anything I can log here in Winnipeg help out?

I had worked my tune in nicely at 21psi tapering to 19 with 70/30 water/meth in the summer.

Winter time here I took all my shit off as that much power is useless when it's -8000

[Fobio]

I will try to do some logs soon/this wknd...4th gear pull to redline is nothing...lol...I do that on the OFF-ramp...

5th gear to redline...well...that's something else entirely...

[Lex]

Originally Posted by GQ_WhiteMS3 I'm currently back to stock tune ... only Intake installed for the winter ... would anything I can log here in Winnipeg help out? I had worked my tune in nicely at 21psi tapering to 19 with 70/30 water/meth in the summer. Winter time here I took all my shit off as that much power is useless when it's -8000

Sure, grab a 4th gear pull to redline using these parameters:

Fuel injection amount
commanded AFR
actual AFR
mass airflow sensor
RPM

Originally Posted by djuosnteisn Not sure bout the stock maf. In a couple weeks i can always put the stock air box on some how, lol. Just to see i guess.

Hey DJ, how wild is it that the car tries to reduce power by throwing all the fuel it can at it instead of limiting it??

Judging by cld12s injection amount (5200mg/st with 320g/s aiflow) and what you're seeing - even if you were flowing 500 g/s you should NOT be seeing 8500 mg/stroke injected amount. You should only be seeing 8000 mg/st.

How much black smoke you seeing out the exhaust?? If you run too rich you can fool the O2 into thinking you're lean.

Unless the parameters we're looking at are different.

[GQ_WhiteMS3]

I'll run that on my drive home ...

[djuosnteisn]

Adam followed me home the day after we were testing the ckt, which was my first blush at this whole thing, and he did say that there was alot of smoke coming out the exhaust. But IIRC, my o2 was only reporting mid 11's. I have logs at home that i can verify.



Hmmmmmmm......

[Lex]

Scratch that, COBB looks at fuel inj amount differently than the DH. The PW between Dustin and cld12pk is similar at around 9-10ms.

Since cld12pk is seeing under 350g/s via the MAF, I tend to believe Dustin shouldn't see much higher - esp at the higher elevation. So I really don't think the car is listening to the MAF anymore and it's likely gone into a speed density state assuming a stock motor and using the MAP sensor.

[superskaterxes]

If its any help I can do a log being that I have a 3bar MAP sensor and it reads above 22 psi but won't display more. I know it reads more because my boost cut works perfect all the way to 30 psi haha. Ill only have the sst on for another week so lmk what logs to take.

[djuosnteisn]

Also super and i were talking about mbc's earlier... so let it be known

I have a mbc. Though i really don't think that makes a lick of difference, but i'm still new to stuff like this.

[SpeedSixxx]

want me to run any thing and log it?

i found out i'm spiking boost at 17 psi and going down to 12 psi super fast and it holds at 12 psi..=(

i been driving my car with this boost leak for over a year now and didnt know..LOL

makin thread on that..

[cld12pk2go]

Originally Posted by Lex For cld12, fuel injected amount doesn't get near what you see on your car Dustin. He's not going over 5500 mg/stroke whereas you're almost at 8500. That's a huge fueling difference. cld12, how many g/s are you seeing calculated from your MAF in those logs?

On the second pull above: 4.66v = 279.74 g/s. My 291whp dyno was at 4.51 ~ 253 g/s. If you were to scale the dyno via the g/s ratio that it would be ~321whp.

Unfortunately, the stock snail can't deliver much more that that...

[superskaterxes]

on the MBC,

i was thinking that if you take the ECU's ability to control boost away, this has to affect other systems. load is based upon boost and if your load target is say 2.5 and your only seeing 2.0 max load because your limiting boost to X psi then other systems have to take over.

i think we should find a definitive way to peg the MAF and see if we can control this speed density calc somehow. anyone wanna dig up that old info from ztuner about when he switched the whole car to speed density?

[GQ_WhiteMS3]

Here is my log

STOCK map ... AP un-installed ... only hard part on the car right now are Cobb SF SRI and test pipe.

Temp -5 C

[djuosnteisn]

180 g/s corresponds to 7750 mg/cyl

DH obviously calculates this differently. And maybe it just pegs at a certain level?

[superskaterxes]

i think we need to take a log of a stock MAF dia that is pegged from a 22-23psi BT pull and compare it to a log of the same car but with say a 3.25 MAF with scaled values except for all voltages above 3.5, there should be 500's in all the cells.

this should give us some insight into how and what other systems are taking over in each situation.

i also remember on this log lex posted above, that dustin was doing some MAF testing and you can see that the inj amt is pegged till about 5750 rpm and then suddenly sky rockets and the AFR begins to drop at almost the same rate beginning at the exact same spot.

dustin can you shed some more light on this?

[djuosnteisn]

Not sure what kind of light to shed? That's the log i took a few days ago, on the way home after all this mess started.


Maybe VVT plays a role? I know near redline my VVT drops to like 4 or so on that map. So essentially the intake valve would be staying open longer in the cycle.

I'm not sure how this would help though. If the injection start is referenced to the int valve opening, i would actually think this would hurt, cause it would essentially be closing the window slightly.

[superskaterxes]

i thought i remember you posting in a thread a while ago about you upping the maf in higher rpm and thats when u saw this increase in inj amt. this was like months ago haha.

[djuosnteisn]

I started that thread like 3 days ago hahaha. My dyno thread, "cars acting a fool".

Don't think i did one besides that ever. First time i ever cal'd my maf it took like 3 maps total, and was a breeze.

Only other testing i've done is the fuel injectors and fuel rail pressure stuff.

[Lex]

Originally Posted by GQ_WhiteMS3 Here is my log STOCK map ... AP un-installed ... only hard part on the car right now are Cobb SF SRI and test pipe. Temp -5 C

See, in this log the MAF is used and correlates to injected amount and AFR directly.

In Dustin's log, the MAF is no longer used since it's way out of wack. However, the car is still able to calculate an injected amount and command an AFR and still MEET that AFR.

It must be using a speed density type of algorithm likely using the MAP sensor and using another fuel table ALONG with the one we see in ATR to calculate commanded AFR and injected amount. This is why the one we see in ATR does affect the fueling but it no longer matches it. Some form of blending is used and COBB did not give us access to all the tables that work in this type of region.

So the question is - WHAT makes the car stop using the MAF?
How far does it have to scaled to be ignored? I
s the MAF only used at low loads and not a very high loads?
Are the MAF and MAP sensors compared and if they are too far off from each other MAF is no longer used?

Super, I look forward to seeing your logs.

[djuosnteisn]

I can definitely say, tricking the ecu to stop using the maf is NOT as easy as filling the table with 500's. Doesn't work.

I think Lex is accurate with the idea that it's ignored above a certain air flow point.


Our ecu's are full of redundant tables, and we know the algorithms will check multiple talbes and use the lower or higher option based on the table values.

There is probably something similar going on here, and at a certain level, the ecu prefers an alternate load calc over the traditionaly maf one.


I can also say that in this alternative mode, it seems that temperature isn't used, cause my afr's were scary lean in the mid 20's on way to work, and mid 11's later that day in high 40's.

[Lex]

maybe it assumes the entire MAF is bad at that point which includes the IAT sensor. You'd think BATs would come into play.

[superskaterxes]

i noticed as i was doing some tuning that the fuel OL/WOT commanded EQ (no knock A) table was different then the (no knock B) table. i was trying to adjust my AFR and after changing the A table several times nothing was changing. but as soon as i changed the B table my AFR was DEAD on to my target. this B table also used to be labeled as "unused" in a previous version of ATR but was updated at some point.

dustin i realize you cant just fill the MAF with 500's but that seems to only be valid for the ENTIRE maf curve. i suggested just filling cells from say 3.5-5v with 500 and see what happens. we can slowly expand that range till driveability becomes an issue then you should know the threshold.

and the entire concept of speed density is based upon pressure and temperature. pressure can be converted to air mass but only with the help of a temperature reference for the incoming air. this would make sense as to why mazda used a MAP sensor with an integrated temperature probe known as the BAT sensor.


someone also needs to test the DBW throttle A and C tables and see what happens when the duty cycle values are raised above 80. whos to say 80 is the MAX? i think this follows the same lines of you dont want to run the WGDC at 100 for very long because you can burn out the solenoid or it might get stuck.

dont get me wrong i normally would have no problems testing this stuff but its like 25 here and ive been out of meth for like a month cause my local shop hasent ordered any because of winter. i just havent been WOT since haha. im also gona be in the process of going BT int he next few weeks so after im all dialed in i can do some big boy testing.

[QuickSpool]

I was talking w/ a developer that has done work on this ecu and then doing research on the new 2nd gen ECU. He said the logic is completely different. I know it hasnt been cracked yet but perhaps we could plug their ECU into our car? They are the same chassis and motor, how close are the electronics? Are they pinned the same? Perhaps the reason they changed the logic is because they realized that what we are running into is some issue w/ the data set not being wide or large enough.

Just spit ball'n.. comments?

[Lex]

Originally Posted by QuickSpool I was talking w/ a developer that has done work on this ecu and then doing research on the new 2nd gen ECU. He said the logic is completely different. I know it hasnt been cracked yet but perhaps we could plug their ECU into our car? They are the same chassis and motor, how close are the electronics? Are they pinned the same? Perhaps the reason they changed the logic is because they realized that what we are running into is some issue w/ the data set not being wide or large enough. Just spit ball'n.. comments?

Did the developer mention HOW they are different in any way?

[superskaterxes]

Originally Posted by QuickSpool I was talking w/ a developer that has done work on this ecu and then doing research on the new 2nd gen ECU. He said the logic is completely different. I know it hasnt been cracked yet but perhaps we could plug their ECU into our car? They are the same chassis and motor, how close are the electronics? Are they pinned the same? Perhaps the reason they changed the logic is because they realized that what we are running into is some issue w/ the data set not being wide or large enough. Just spit ball'n.. comments?

lol they prob added MORE BS logic and walls and safeguards to prevent even 300whp. FML

[Fobio]

it almost seems like the original DI system (by Hitachi?) has redundant and overly complicated parameters, that are somewhat narrow relative to percentage of power gains we're subjecting our cars to...I FEEL that it's almost Japanese, culturally: a lot of attn to details for minute adjustments over what is perceived too be an adequate power range...and that's it.

to expand that "adequate" range, likely requires a logic overhaul rather than just changing a couple of tables...however, I wouldn't put it past them, that some of those redundant tables are either catch-all's or safety nets, but can also be used to affect the overall logic, ie. slight room for change.

[FreeFlyFreak]

Originally Posted by QuickSpool I was talking w/ a developer that has done work on this ecu and then doing research on the new 2nd gen ECU. He said the logic is completely different. I know it hasnt been cracked yet but perhaps we could plug their ECU into our car? They are the same chassis and motor, how close are the electronics? Are they pinned the same? Perhaps the reason they changed the logic is because they realized that what we are running into is some issue w/ the data set not being wide or large enough. Just spit ball'n.. comments?

From the Acessport for 2010 thread

This is from the Cobbtuning forums... Quote Originally Posted by Trey View Post I'll try to answer as best I can. For the 2010 MS3, Mazda has changed the internal hardware for the ECU. This involves a new processor type that operates using a different "operating system". An analogy for this is that the 2007-2009 ECU speaks Japanese, the new one speaks Icelandic. So in order to learn how to have a conversation with the ECU, our engineers need to study an entirely new language. Secondly, with the hardware "language" change, they've altered the process used to get the ECU into the reprogramming mode. This involves a process of knowing how to calculate the proper password, knowing all the right commands to get in, and then how to handle sending the ECU the new tuning changes which is typically encrypted in a unique manner. All those things changed from 2007-2009 to 2010, so this ECU might as well be running a new direct injected Porsche GT3 for all we care from the engineering end. Once we get those items figured out, we next need to figure out what all the data means, how the ECU logic functions, what is the same as the old car, what's new. Where are the fuel tables, load target tables, etc. The best way I can answer the question "how long" is to make an analogy to you going to Best Buy to buy the latest sequel to your factory Xbox game and the guy at the register asks, "So, how long will it take you to finish this game?" How accurately can you answer that? So when you respond, "Don't know, haven't even started playing it yet". To which he retorts, "Well, you played the last one right? Should take you the same time, right?!?" And you think to yourself, 'geez man, I have no idea. This thing has a bunch of new levels, I still have a to-dos around the house that may prevent me from playing it 12 hours a day, and... crap, the entire game is in Icelandic. I don't even know Icelandic!' As for if the current AP hardware is compatible, we certainly hope so but we can never make a guarantee until we have a working product. It's very expensive for us if it is not, so you can be sure we're in the same boat as you hoping no changes are required. Cheers, Trey @ COBB

[802MS3]

fucking Mazda. why is everything so top-secret with them? you'd think it would be less work to just drop the same ECU in, instead of creating a new OS, new hardware, new logic, etc. wtf?

[Lex]

It's very interesting that these changes were made. The engine is the same - so why make all these change? The engine AFAIK is not being carried over into any new platforms either.

So in other words, if you were Mazda, why would you spend the money in developing something completely new?

I will give you a hint. Mazda did not spend any more money. Mazda did not develop the new or old ECUs. They contracted a company that specializes in control systems. Those guys develop the ECU and sell it to Mazda along with with software that allows the people at Mazda to access all the tables. This type of engineering software is common in the industry but good luck getting your hands on it. It would literally be the ultimate flash tool and very illegal to obtain.

So, for 2010, it's likely Mazda contracted out someone else or had the same ECU company develop a new one. Why? Usually it's emission controls. The new ECU is likely faster and better adapted to meet emissions targets.

All that Mazda cares about is that they can calibrate the engine and that it meets those emissions requirements.

[stealthspeed6]

delete...

[bova]

so you guys said you aren't seeing over 2.0 load? i hit 2.2 at 18 psi in 4th gear just today with afr at 11.62.

[djuosnteisn]

Post up log bova! With all the parameters lex mentioned above.

[Haltech]

Originally Posted by Lex It's very interesting that these changes were made. The engine is the same - so why make all these change? The engine AFAIK is not being carried over into any new platforms either. So in other words, if you were Mazda, why would you spend the money in developing something completely new? I will give you a hint. Mazda did not spend any more money. Mazda did not develop the new or old ECUs. They contracted a company that specializes in control systems. Those guys develop the ECU and sell it to Mazda along with with software that allows the people at Mazda to access all the tables. This type of engineering software is common in the industry but good luck getting your hands on it. It would literally be the ultimate flash tool and very illegal to obtain. So, for 2010, it's likely Mazda contracted out someone else or had the same ECU company develop a new one. Why? Usually it's emission controls. The new ECU is likely faster and better adapted to meet emissions targets. All that Mazda cares about is that they can calibrate the engine and that it meets those emissions requirements.

All i need is the name and i can get it. There isnt anything, i cant get my grubby hands on. If its software, i can obtain it, esp through my Japanese and Chinese contacts.

[Lex]

^ We need to find out who the developer was and where it was made. Didn't someone have one of these ECUs open?

[djuosnteisn]

Bioevolve did

Inside of ECU, ever wonder?