Based on my understaning of how ECU piggy-backs (e.g. Emanage / Canzoomer) tune AFR, I wonder if they are going about solving the problem BACKWARDs for an ECU like the RX-8's.

From what I gather, these piggy-backs work by intercepting and "tweaking" the MAF input signal to the ECU. The idea is that by telling the ECU that more or less air is comming in than actually is, you can trick it into running richer or leaner respectivly.

This seems all well and good untill you try to run on an advanced ECU like the RX-8's. It seems that the ECU monitors the wide-band O2 sensor not only in closed-loop, but also in open-loop mode. The ECU will then proceed to "correct" the maps back towards target stock values by adjusting its internal long term fuel trim table (LTFT).

Basicly the issue is that the ECU has a kind of slow feedback loop even when running in "open-loop" mode. While many may see this a "problem", I find myself thinking this feature would be GREAT if it were only "auto learing" the settings needed to meet OUR target AFRs (rather than the stock ones). This would be easy for an ECU "reflash", but sadly that isn't available. But could it be done with a piggy-back type modification?

It seems to me that it can: by flipping the standard piggy-back "backwards". Instead of intercepting and modifing the MAF input, souldn't we instead intercept and modify the wide-band O2 sensor ECU input? Thus switching from modifing the combustion system "input" variable to modifing the combustion system "output" variable. Here is an example of how I immaging this to work. If the ECU keeps wanting to tune a point in the map to 11.5 AFR, but we want to run at 13.0 at that point; we'd "adjust" the O2 sensor so that it "read" 11.5 when the actual AFR was at 13.0. I can think of several advatages and disadvanages of this type of reverse piggy-back setup.

Advantages:
- Immune to ECU "unlearing".
- Should also allow tunning of AFR under closed-loop conditions as well if desired.
- Easy to understand maps (+-AFR from stock is probably easier to think about/set than percent changes in airflow).

Disadvantages:
- Potentialy slow tunning process (must wait for ECU to do it's LTFT learning before re-adjusting).
- More complex senor intercept electronics? (Perhaps the O2 sensor is more difficult to read / impersonate than the MAF?)
- This scheme only applies to AFR, timing/advance must still be controled using other means.

Is there something fundamental I'm missing? Obviously this type of piggy-back would be useless on any car that doesn't "learn" in open-loop, so I can understand why this approach hasn't been taken in the past (requires a very modern ECU). I'd appreciate input from knowlegable tunners to know if this idea is all screwed up or actually interesting.

- David

 

That is very sneaky....I love it!

Thinking "backwards" can pay off bigtime!

...and I see no reason why it won't work.

S

 

so when does the prototype come out

 

So when the 02 sensor reads 11.5:1 it is actually 12.5:1 and the computer doesnt know the difference? wouldnt even need to spend a bizallion dollars on a computer. just maybe a hundered or so on the sensor. Call it the "Performance Sensor" cheapest mod the RX-8 has ever seen, and wouldnt really be detectable by Mazda. Great idea! Someone get moving on it. You should get a cut of the money! On top of it all we would get better gas mileage.

The trick is to get a major manufacturing company to produce a faulty o2 sensor!

They could make a whole range of um from 11.6:1 to 13.5:1, to tune simpy buy the sensors in your estimaed range.

Im sure a comparable device could be created for timing advance/retard

 

What would happen in the areas that the ECU wants 14.7 ish....would you get 15.7ish+ and a big bang? I suppose with a controller it can be active only in areas that you want to adjust...like the current MAF/MAP adjustments......

I'm sure that Mazda has routines that test the sensor that would have to be worried about as well. Can it be this simple??

 

great idea and creative thinking (we need more than we have) but I dont think that will work. As I understand it the a/f has to be variable to get everyday performance. IE cold starts are differant, idle is differant, wot is differant than cruise. Timing issues. The way I understand it also is with the factory set up its a coordiated readings between the maf and the widebands. Not just one or the other. even in openloop. Now a stand alone complety bypasses that system so no problems,but a piggyback well then you have experience. i am not the smartest on this types of systems --ijmo. just my 2 cents worth----keep thinking. I see a reflash in the future(at some point)
olddragger

 

I was looking into doing something like this to make my IT RX7 run richer. If I remember correctly, reducing the voltage comming from the O2 sensor make it look like a lean condition and forces the car to run richer. If this is true, to make the 8 lean out you would have to add voltage to the O2 sensor and it would have to vary with conditions. Also, I believe, the O2 sensor is a high impedance device, therefore it would be sensitive to external current and voltage variations.
But, then again, I could be wrong.

 

under a load or WOT conditions your computer will not be going for 14.7:1 AF, under deacceleration or crusing i could see a AFR of 14 and higher but you wont detonate or blow a motor under those conditions.

Another thing to think about is that in essence lower fuel will be burnt than the computer thinks, and it may be cleaner with less unburnt fuel. I wonder if that would figure in at all to being able to not ran a cat converter and the sensor beleiving that the cleaner exhaust burn is due to the cat working properly? Just throwing it out there but wouldnt it be nice to save fuel, better A/F ratio, remove the cat, avoid CE lights, and get an extra 20hp? Okay so its a long shot, but it doesnt hurt to dream.

 

You are waaayy overestimating the capabilities of the ecu.

It absolutely does not monitor the afr in open-loop conditions and adjust the ltft from there. What it does is monitor the afr during closed loop conditions, adjusts the afr based on the feedback in the ltft at that area (the ltft is really a map of load vs rpm), and then has the cells in the ltft corresponding to open-loop conditions linked to the cells in the closed-loop ltft.

If you avoid tuning in the closed-loop mode you will absolutely have no issues with the ecu compensating. Its only when you get gready and tune in the closed loop areas that you will have a problem - as demonstrated by countless people here....

 

exactly.

 

Remember what CZ said:

"Some of you who are tuning have posted that the car ECU "learns" a long term fuel trim and adjusts against the tuning. You are right!

BUT: The car checks AFR when it is in closed loop mode. That is below 70% throttle, and below the closed loop/open loop transition rpm. So, what IS the transition rpm?

By measurement:
1st gear: 6,200rpm
2nd gear: 5,500rpm
3rd gear: 4,800rpm
4th gear and up - same as 3rd.

So, some lessons to derive from this:
1) Tune above transition point ( 6,200) and do NOT touch the fuel/air below 6,200 unless it is only 5% or less.
2) If you want to tune down to 5,500 avoid full throttle in 1st gear below 6,200
3) If you want it tune more, and want to get around the "learned" long term trims, simply disconnect the battery ground if the ECU "learns" a fuel trim. This returns the car to default fuel/air settings. Of course it will "learn" again, so this is not a long term solution. I installed a replacement battery ground post that has a 1/4 turn **** to disconnect. This allows for quick and easy resets.


Also, regarding weather and temperatures:
If you tune in cold weather the air is denser. When you run at a higher temp the air is less dense. That means the mixture will be a bit different. Humidity will also affect it. The car has a barometric pressure sensor as well, so it does compensate for altitude and air pressure change due to weather, however it does not have a means for detecting humidity. My advice, if doing fairly aggressive maps is to do a set for both cold/dry conditions and for warm/wet ones. Save them and use as needed."

 

The routines the ECU uses to ensure the Wideband O2 ( current based) and narrow band sensors (voltage based) are working is a lot more complex that in previous cars. The ECU sends signals to illicit changes in these sensors...and expects certain responses in return. The protocols are not as easy to fool as might be thought. You would have to be able to fool both circuits in the same manner to give the ECU the feedback it expects not to throw a CEL

 

I was going to post this info, but thanks to r0tor

From what I have observed, it isn't really that simple. The ECU monitors closed-loop AFR and adds/subtracts LTFT over time. Even after tricking my MAF signals to revert LTFT back to zeros, my open-loop LTFT retained the same. Monitoring some more, it seemed like the ECU takes the mean LTFT within closed-loop and copies that number to the open-loop area. For example, the mean LTFT in closed-loop is 6, I see that all cells in open-loop (WOT or gradual throttle) received 6 LTFT.

The only way I reset the open-loop LTFT back to zero was to remove the ECU and plug it back in. No need to shut off battery. This idea of intercepting the WBO2 signal might work, but you would need to set RPM/TPS/Load controls to its behavior. You can't just plug it in and expect it to work fine. If this intercepting device can be shut on and off via a signal, then an RPM switch from Summit Racing can do this. It costs $45 - $80 I think.

 

Let's just get in the system itself and change the target AFR's there. I know that's how the flashes are done. The R flash runs pulses of lean conditions from mid to high RPM range. None of the previous flashes did this. Italy and Japan are already doing this, could someone with contacts there send them distress signals? We really need their help in this.

hfalex gave us tips on how they reprogrammed the ECU already, but none of us here have figured out the hardware aspect of it. I read that free DOS software they used to reprogram, it should take some getting used to to learn the capabilities of that software.

How much does an RX-8 ECU cost? Just in case something gets messed up during the process...

 

italy and japan are altering the PCM physically in order to change the programming. this is not the best solution since it make sit difficult for the guy who is far from the person who knows how to get it done. also its not neccesary with the proper software and hard ware. just give it a little more time. there are efforts underway to make this a reality. i know ive been saying that for a few months already but really just give it a little more time.

 

^ Right zoom44. But the problem here is that nobody likes to work with our car. It's been two years since the release and we're still hoping for someone to take interest in our cars. Is anyone following the new Mustang's development? I heard they have almost a similar ECU.

 

the ecu are similar in that they have to "speak" a common WDS(or whatever they want to call the new tool) "language" and they operate in the same fashion but they are not similar. they are made by different manufacturers using different chips- and yes there are people that want to work with our cars.

 

show me power levels comign out of italy that are NA. the only ones ive seen have been FI

 

dude people have been burning clutches and trannies on this car since about 2 months after delivery. ito did at least 2 before he even started porting.

i still havent seen alex show any gains on NA cars. im not saying astra hasnt done it i just havent seen them talk about it. ive seen the results of their SCs.

 

Thanks for the responses and the enlightening info. So it seems that I was taken in by the common misconception that the ECU "tunes" under open-loop conditions. Given that, it makes perfect sense for the ECU to use some function of the close-loop trim tables for the open-loop region. Also now it makes sense to me why Greddy was using some kind of dongle to try and get the 8 to run in open-loop mode much earlier (so that they could apply tuning over a greater region).

I've been assuming that the ECU in closed-loop mode opperates as a classic feedback control system. Because it takes some time from input airflow change till measured AFR responds, there is time delay in the feedback loop. As a result of this time delay, a closed-loop feedback control system looses stablity once the rate of change in the system gets too fast. So it the system starts changing too quickly, then it makes sence to swich into open-loop mode (i.e. stop using feedback information). Given this, these transitions from closed-loop to open-loop seems a bit strange to me:
While the transition with throttle makes sense (less throttle = slower changes), the different rpm transitions with respect to gears seem backwards to me. At WOT, conditions are changing MUCH faster in 1st gear than in 6th. So why is the rpm transition point the highest in 1st? Shouldn't it be the lowest? I.E. because conditions change faster in 1st gear so we are forced into unstable conditions earlier.

I'm probably suffering from another misconception. I would apreciate being set straight.

- David

 

Those were the words of the imortal Maurice AKA "Canzoomer"
Maybe the resident rotary gurus can shed more light for you. I only drive what my tuner gives me. At my age it is more fun that way

 

You have half of the picture correct in that the closed-loop system does act like a traditional feedback loop - and there are issues of stability with rapid transitions and a good reason to have open-loop control.

The other half of the picture is that a stoich 14.7:1 ratio does not actually give you max power. Max power in the renesis is somewhere right around 13.5:1. So it really makes sense to have an rpm to go into an open-loop mode for more power and better throttle response even if you have low throttle input (think of a road course where you are on/off the gas). It also assures you under aggressive driving conditions like this your not going to get the fuel system confused.

There is also an emissions part of the equation since you really need a 14.7:1 ratio to acheive good emissions control - so normal driving conditions should be closed-loop.


As far as why its different in each gear - I don't have a good answer for that as I've also have been pretty skeptical of it. It seems to me its either real or maybe a slight delay in reporting time of when it switches.

 

When you say "it realy makes sense to have an rpm to go into an open-loop mode for more power and better throttle response", are you impling that in closed-loop the target AFR is always stochimetric? Can't closed-loop mode target any desired AFR (14.7, 13.5, etc.) as long as it's stable, or is it limited to only 14.7? I can see why this would be the case if we had only a narrow-band O2 sensor, but shouldn't the wide-band provide more flexebility?

- David

 

traget in closed loop is always stoich