[dlbrooks18]

Originally Posted by dzenno

All you would need to do is plug a Cobb AP into your car, unplug your rear o2 sensors or run the stage1 map and be catless and we all know what happens then, you get a CEL for cat inefficiency...ONCE that is triggered, the DME will run a different fuel "mode" on your car...how that reflects on power at this point I can't say for sure but it does impact "fueling"...if you'd like any further details don't hesitate to ask cobb or post in their thread...

another example: when a car misfires, fuel mode changes as well, different from the above...normal running fuel mode is fuel mode 2, when a car misfires DME goes into fuel mode 15 (fuel system shutdown) to prevent damage to your catalytic converters from unburnt fuel...it goes back to 2 once you restart the car if everything is fine and no more misfiring

there are literally massive numbers of monitors in the dme and these are just some of them

Interesting, any one have evidence that supports this? I don't think power and/or fueling is affected.

[Mike@N54Tuning.com]

Originally Posted by shiv@vishnu

The issue has nothing to do with post cat lambda monitoring. But rather front lambda (pre cat) fuel control. The DME employs two closed loop fuel control routines. One for cyl 1-3 and another, running in parallel, for cyl 4-6. So merging all 6 cylinders and feeding the DME the signal from the front lambda sensors (now reading all 6 cylinders) allows for the two closed loop systems to "fight" each other. And opens up the possibility for one bank to run lean while the other to run rich. The average of all 6 cylinders, in such a case, would still read "normal" and the DME wouldn't know anything is wrong.

Some "tooners" haves suggested modifying the DME code to employ only 1 closed loop fuel control routine for all 6 cylinders. Now THAT would be a step in the wrong direction as you are giving up 1/2 of the stock DME's fuel control precision. Something much welcome when making big power. Heck, if I had a genie, I'd wish for a DME that employs 6 closed loop systems running in parallel, each being fed a lambda signals from 6 individual o2 sensors mounted in each exhaust runner (pre turbo). But you'd only see such excess in extreme cases. But for real world street use, dual bank fuel control is great and just adds to the beauty of the n54 control control system.

Shiv

You're talking about this comment? The N55 motor runs a single bank and single o2 sensor right from BMW so it seems like a workable approach. Locating the o2 sensors preturbo seem like a good idea to get some fuel system R&D work done but that can't be the long term plan can it? I wonder why HPF didn't just call AR Design before doing all this work they could have told them about the fuelling problems with o2 sensor placement a year ago.

Yep, that's been the problem with all previous N54 single turbo kits. And when you try to separate the banks your sensors wind up pre turbo and exposed to heat they won't stand up to long term.

Ideas for you:
1) Contact a good flash tuner like Cobb and see if they can switch the fueling to single bank like the N55 motor.
2) Design a device that combines both sensor signals post turbo and outputs only one uniform signal to each bank, keeping the trims identical, in theory.

[dzenno]

Originally Posted by dlbrooks18

Interesting, any one have evidence that supports this? I don't think power and/or fueling is affected.

"Fuel Mode" is one of the monitors in the DME that the Cobb AP exposes and you can easily see it and log it on any car running that tune...i'd happily discuss/show "evidence" but in another thread if anyone is interested but i'm not sure there's much point in showing evidence of 1 of numerous monitors channels available in that tune/DME...not sure what it would "prove" in the end other than that the DME does indeed have different fuel modes and in turn fuel control logic to support them...

[dlbrooks18]

Originally Posted by dzenno

"Fuel Mode" is one of the monitors in the DME that the Cobb AP exposes and you can easily see it and log it on any car running that tune...i'd happily discuss/show "evidence" but in another thread if anyone is interested but i'm not sure there's much point in showing evidence of 1 of numerous monitors channels available in that tune/DME...not sure what it would "prove" in the end other than that the DME does indeed have different fuel modes and in turn fuel control logic to support them...

Cool.. makes sense,

Chris- thanks for the update and I look forward to seeing your new manifold design and the 335 back on the dyno. Gonna be a beast for sure!!

[rader1]

Originally Posted by bryce

this fact/concern was brought up several times in this thread. it's a shame hpf couldn't monitor this thread closer, there were several members (including myself) bringing up concerns of the primary oxygen sensor placement (or lack there of) on the HPF manifold.

LMFAO even we saw this coming a while back and HPF had no clue that this was going to be an issue?
Well, it only took 2 years to get the hardware together the first time so... HPF Single turbo dyno in 2015?

[SIKH335]

Good Luck HPF. I'm sure you will get it working in the long run

[Crazylegs]

downside to such a smart car

[Skydive]

Originally Posted by shiv@vishnu

The issue has nothing to do with post cat lambda monitoring. But rather front lambda (pre cat) fuel control. The DME employs two closed loop fuel control routines. One for cyl 1-3 and another, running in parallel, for cyl 4-6. So merging all 6 cylinders and feeding the DME the signal from the front lambda sensors (now reading all 6 cylinders) allows for the two closed loop systems to "fight" each other. And opens up the possibility for one bank to run lean while the other to run rich. The average of all 6 cylinders, in such a case, would still read "normal" and the DME wouldn't know anything is wrong.

I agree with this.

Originally Posted by shiv@vishnu

...have suggested modifying the DME code to employ only 1 closed loop fuel control routine for all 6 cylinders. Now THAT would be a step in the wrong direction as you are giving up 1/2 of the stock DME's fuel control precision. Something much welcome when making big power. Heck, if I had a genie, I'd wish for a DME that employs 6 closed loop systems running in parallel, each being fed a lambda signals from 6 individual o2 sensors mounted in each exhaust runner (pre turbo). But you'd only see such excess in extreme cases. But for real world street use, dual bank fuel control is great and just adds to the beauty of the n54 control control system.

With respect Shiv, I believe you may be incorrect here.

The DME has per specific cylinder mixture data because each cylinder fires one at a time. The time it takes that gas sample to flow to the O2 sensor, and the sensors' hysteresis is modeled as a function of load, temperature, and so forth.

The reason for twin oxygen sensors on this car is the twin exhaust paths of two turbochargers, not the reverse. The N55 has a single turbo and a single pre-catalyst oxygen sensor and is every bit as capable of specific cylinder readings, assuming that BMW has a system at least capable as that of ten year old Audis. The N55 is controlled as a two banks of three cylinders as well... is it not?

I can understand that solving this problem without modification of the underlying software may be difficult, but it I'm fairly certain that it can be properly solved in software, given the right information.

Thanks for clarifying...

-Michael

[OpenFlash]

Originally Posted by Skydive

With respect Shiv, I believe you may be incorrect here.

The DME has per specific cylinder mixture data because each cylinder fires one at a time. The time it takes that gas sample to flow to the O2 sensor, and the sensors' hysteresis is modeled as a function of load, temperature, and so forth.

The reason for twin oxygen sensors on this car is the twin exhaust paths of two turbochargers, not the reverse. The N55 has a single turbo and a single pre-catalyst oxygen sensor and is every bit as capable of specific cylinder readings, assuming that BMW has a system at least capable as that of ten year old Audis. The N55 is controlled as a two banks of three cylinders as well... is it not?

When you mount the primary lambda sensor downstream of a turbocharger, the exhaust pulses are literally chopped up. Under moderate to high load, a DME would be incapable of identifying individual pulses of exhaust (and which cylinder they came from) because the pulses are no longer in tact. And under high engine load where there is gobs of exhaust pressure stock between the exhaust port and the turbine inlet, the post-turbine exhaust is nearly homogenized. Which is why even narrow band o2 sensors can provide a stable ~.65v reading when the engine is operating at stoich.

Also, I don't believe the N55 controls fuel in dual-bank mode. In fact, there is no diagnostic code nomenclature that differentials between bank1 and bank2.

I can understand that solving this problem without modification of the underlying software may be difficult, but it I'm fairly certain that it can be properly solved in software, given the right information.

Anything is possible, sure. But if the DME was able to accurately monitor/control lambda on a per cylinder basis, there wouldn't be any need for injector coding. Or even the existence of many lambda-related fault codes.

Shiv

[JamesM3M5]

Originally Posted by Skydive

With respect Shiv, I believe you may be incorrect here.

The DME has per specific cylinder mixture data because each cylinder fires one at a time. The time it takes that gas sample to flow to the O2 sensor, and the sensors' hysteresis is modeled as a function of load, temperature, and so forth.

The reason for twin oxygen sensors on this car is the twin exhaust paths of two turbochargers, not the reverse. The N55 has a single turbo and a single pre-catalyst oxygen sensor and is every bit as capable of specific cylinder readings, assuming that BMW has a system at least capable as that of ten year old Audis. The N55 is controlled as a two banks of three cylinders as well... is it not?

I can understand that solving this problem without modification of the underlying software may be difficult, but it I'm fairly certain that it can be properly solved in software, given the right information.

Thanks for clarifying...

-Michael

Are you sure that is possible with a common "log" manifold for 3 cylinders with the lambda sensor after the turbo? At 3000 rpm that's one pulse in a bank of 3 cylinders every 13 milliseconds. Add to that the exact time delay needed to figure out which cylinder pulse the lambda sensor just read, I don't see how the ECU can accomplish individual cylinder trims for 2 banks of 3, let alone one bank of 6 on the N55.

[vasillalov]

I've been working on translating the MSD80 file for INPA, you know, the software that BMW dealers use to diagnose the car... For the past couple of days I've been working on the section that deals with Injector adaptation and I can tell you for a fact that the DME maintains all kinds of information for each individual cylinder.

The DME also creates short term and long term injector adaptations based on Lambda sensor voltage, lambda sensor heating, coolant temperature, knock, rough idle, whether it was a cold start or a warm start. Furthermore, at least half of the individual injector adaptations are based on various parameters the DME sees from the Lambda sensors.

So yes, the DME does indeed make individual cylinder adjustments based on all kinds of parameters received from the lambda sensors.

Heck, the DME even performs test on VANOS camshaft stiffness!

[vasillalov]

Originally Posted by bryce

do you have a link or source for that? not trying to be argumentative, but everything i've learned on this platform and previous platforms has indicated that secondary oxygen sensors merely detect catalytic converter efficiency; not fuel control or anything more than that. on previous cars, i turned off rear o2 sensors to avoid having cat codes

Yes. It is in INPA. The DME actively adjusts injector adaptations based on condition of both pre-cat and post-cat lambda sensor parameters.

[Skydive]

Originally Posted by JamesM3M5

Are you sure that is possible with a common "log" manifold for 3 cylinders with the lambda sensor after the turbo? At 3000 rpm that's one pulse in a bank of 3 cylinders every 13 milliseconds. Add to that the exact time delay needed to figure out which cylinder pulse the lambda sensor just read, I don't see how the ECU can accomplish individual cylinder trims for 2 banks of 3, let alone one bank of 6 on the N55.

In point fact at least two of the factory turbocharged Porsches run pretty much exactly this post turbocharger configuration at yet higher engine and turbine speeds with per cylinder lamda.

[JamesM3M5]

I wouldn't be so sure about that. The ECU definitely makes individual cylinder adjustments. Most of it, I believe, comes from smooth running measurements at idle and lower RPMs and is not possible with lambda sensors after the turbos. Smooth running measurements and adjustments are made by watching the crank trigger signal accelerate and decelerate whenever individual cylinder variances cause differences in combustion efficiency. You can watch the smooth running adjustments live via diagnostic equipment. It's much more powerful than simply determining cylinder misfire like much older systems. Even the Siemens MSS50 back from 1995 did away with individual throttle synchronization required on all previous M cars with smooth running adjustments, making tiny fuel corrections to individual injectors by the microsecond in real time.

I just don't see how post-turbo lambda sensors are capable of telling the DME anything other than homogeneous exhaust oxygen/fuel content. Again, that's 13ms between pulses at only 3000 rpm, not to mention the delay from exhaust port to sensor. I'm not saying it can't be done, I just see how it's even feasible.

[vasillalov]

Well, the fact is that you can actually see, save to a file and delete those adaptation values for the injectors based on the lambda parameters. I am almost done translating the file and very soon we'll be able to look into this in English.

[Skydive]

Originally Posted by GeorgiaTech335coupe

Plus I'm sure rewriting the control system that works as well as the DME would be a very tough task, even for those who are educated in control systems. A lot of people don't have a full understanding and under-appreciation of the complexities of control systems, how they work, why they are needed, and how to go about testing before implementing them. Controls is always one of the tougher courses in the mechanical engineering discipline - lots of diffEQ involved.

It can be complex, but I can think of several platforms on which it's all there from the factory, and simply requires control bit changes to "turn it on". On the other hand, the cars with which I am most familiar are based on Bosch systems rather than Siemens, and I haven't yet decided to jump in to the Siemens stuff... which could be totally different. (Interestingly BMW went back to Bosch for the N55, and it apparently runs as a single cylinder bank, though I am reading conflicting information in that regard.)

I do know that it is absolutely possible to derive per cylinder lambda information from post turbo sensors. Porsche does it this way, but that's Bosch code. It would seem likely that competing systems would have similar capabilities... but I don't know.

Interesting stuff though! Big thanks to Shiv, BMS, Chris, Dzenno, and Vasillalov, among others. It's information sharing like this that facilitates evolution and sometimes even paradigm shifts in performance envelopes. It took 7 or 8 years to "crack" and revise some of the software barriers found in the now elderly Bosch ME7 for example, so I wouldn't be at all surprised if folks make similar discoveries and significant revisions of MSD 80/81 over the next several years.

[JamesM3M5]

Originally Posted by vasillalov

Well, the fact is that you can actually see, save to a file and delete those adaptation values for the injectors based on the lambda parameters. I am almost done translating the file and very soon we'll be able to look into this in English.

I applaud your efforts, definitely some hard work you're doing and sharing to the community. It is looking good for the N54 community when people take on these projects and share the info.

[rooringhusky]

Originally Posted by JamesM3M5

I applaud your efforts, definitely some hard work you're doing and sharing to the community. It is looking good for the N54 community when people take on these projects and share the info.

+1 Great effort and dedication vasillalov...your efforts regarding this and previous help regarding injector coding are under appreciated in the community.. Thank you very much

[jtdigital]

Price listed on their website $5950, says coming soon
http://www.horsepowerfreaks.com/part...Kits/N54/18813

[Q4P]

Originally Posted by jtdigital

Price listed on their website $5950, says coming soon
http://www.horsepowerfreaks.com/part...Kits/N54/18813

You know it doesn't work atm, right? Although, if that is a true price, it is an awesome deal indeed.

[gian335]

now correct me if im wrong. but is that an automatic transmission i see in the video? can the auto's withstand 500+ whp before sumthing goes wrong?

[AYF80]

Originally Posted by gian335

now correct me if im wrong. but is that an automatic transmission i see in the video? can the auto's withstand 500+ whp before sumthing goes wrong?

Yes its an AT, but they are also testing and installing on MT's as well. TQ handling for the AT is not 100% sure yet.