O2 Sensor and Open Loop

[Som]

Good feedback. I'll update the image later today regarding the O2 sensor. From I can tell, "long term" basically refers to the DME's adaptation, which affects all the maps. The chart would be more accurate if it calls out the "short term" trims then.

As for the 4500RPM, I'd love to get a definitive reference/answer for this. If the AFM isn't used above 4500RPM, it means the TPS WOT switch wouldn't really have any impact on performance above 4500RPM. My only data point here is that during my last race weekend I noticed the engine bog at 4500+ at WOT on straights. I've since found that my TPS was not signaling WOT correctly, so my working hypothesis has been that the lack of WOT signal prevents the DME from going into open loop -- even above 4500RPM.

That said, it's hardly a bulletproof "scientific method" there -- I haven't even reassembled the car with a good TPS to see if the bogging is gone. It's just a working theory. But if there's some sort of documentation that indicates that ALL >4500RPM operation ignores the AFM, it would be good to see.

I guess the other nuance here is -- does ignoring the AFM at >4500RPM mean the same thing as "open loop". If they're 2 different things, then per the chart above, I guess that means >4500 RPM + no WOT would mean the DME has a map that factors in O2 sensor data without AFM data -- and that only seems like it would ever be applicable in neutral or MAYBE 1st gear. Can you get to 4500+ RPM without triggering a WOT on the TPS? Seems like kind of a pointless map to have, but *shrug*, who knows.

I'll try and do some digging and see what I can find, too.

Som

[jlevie]

From data logging I could see that the signal from the AFM became fixed above about 4500rpm. So at that point the DME can't be in fully closed loop mode as one of the inputs is invalid. As to what the DME uses as map data then I have no idea. But I do know that if the DME doesn't see the TPS go to WOT performance will suffer and it appears not to use the WOT maps.

Can you reach 4500rpm w/o tripping the WOT switch? Yes you can in the lower gears. I never tried 4th or 5th, but I could do it easily in 3rd.

There are a number of mysteries buried in the DME and it is hard to figure out exactly what it does.

[TobyB]

...but darn, it's kinda fun tryin'!

t

[Som]

Updated the table.

I realized another unknown -- if you're at high RPM but no WOT, will the DME factor in the O2 short term trims? Maybe that would explain the dip in power?

If not, then the "High RPM w/o WOT" and "WOT + High RPM" columns are the same in terms of sensor usage, but different in terms of maps, which I guess gets confusing. The table was meant to implicitly indicate the maps being used given a set of conditions, but if there's a 1-to-many relationship of sensors inputs to map outputs, then it might be worth making the table a bit more explicit in this regard.

Speaking of maps, and my limited knowledge of how the DME works (I figure I know how I would program one, but not sure how they're actually made), are the maps basically static multiple input/output look up tables? Like "given inputs a,b,c,d,e what are the values for x,y,z"?

Do we know how many look up tables / maps the DME has? I would think one would be able to make an educated guess based on how much memory the DME has and how much memory the maps would take up?

Som

[jlevie]

To the best of my knowledge the maps are static tables, but the DME uses learned adaption to modify data from the map, which can also be modified by short term fuel trim.

I would presume that high rpm w/o the WOT switch uses engine speed, temperatures, and O2 data. But I don't really know for sure and can't think of a way to check. The WOT signal results in AFM, engine speed and temperature data used up to 4500rpm. After that only engine speed and temperature are used as inputs.

[nando]

full load maps are 2 dimensional. part load maps are 3 dimensional. I don't think there are too many "overlays". basically there's a part throttle target and a full throttle target. O2 and coolant corrections are added to that.

actually, most of this is already known. Motronic 1.x has been widely hacked. I'm surprised nobody has done a full disassembly yet - the binary for it has to be tiny. I'm used to looking at 4mb files from newer cars with ten thousand parameters, lol.

[Som]

I figured as much (that these old DMEs would probably have already been fully understood at some point). Kinda feels like the internet accelerates time by making things only 10 or 20 years old feel like centuries.

Kinda feels like "digital archeology" would be a thing -- digging up obscure informational artifacts buried in old posts and websites.

Digging into this kinda feels like I'm in this post-apocalyptic world where "I know mankind has invented this before, but now I gotta do it again because most of the modern world was buried under volcanic ash."

Heh...

Som

[nando]

Yes. In the modern age, most computers/components have datasheets that explain how they work and how to find stuff. These things were made in the 80's before internet use was common - so such resources don't exist. The only way is hardcore binary hacking, hex editing and ASM disassembly..

I've had binaries for the 173 for ages but never really looked at them. Looks like there's about ~220 parameters. It wouldn't be too bad to map it all out. The whole parameter space looks like maybe 4kb.

I've looked at some of the public TunerPro XDFs, there are probably better ones but the ones I've looked at are fairly poor quality. There are clues but not complete information.

Perhaps someday I'll come back around to it. Meanwhile, we have people on the forum who are good resources. I mentioned ForcedFirebird before, he knows Motronic 1.x like the back of his hand IIRC.

There are some documents published by Bosch that describe in detail how Motronic works. I can't remember where I put them it was so long ago that I found them..

[95BMWIC]

Just because it's in open loop, doesn't mean O2 trims do not apply. There are short term trims (closed loop) and long term trims (open or closed loop). Long term fuel trims always apply. So even though it ignores short term O2 feedback at WOT, the O2 feedback mechanism still affects WOT fuel delivery.

I believe the SpecE30 guys have confirmed this. I seem to remember Jim talking about having to reset the DME every once in a while because it would go lean (it's always trying to reach the internal lambda targets).

long term fuel trims are stored in volatile memory. short term trims are not, but short term trims are used to determine what gets stored in memory. When you remove power from the DME, this will reset any adaptations (Motronic 1.x doesn't have flash memory like a newer DME).

As far as I know, there aren't any adaptations for ignition, since no version of Motronic 1.x has knock sensors.

ForcedFirebird knows some more about the AFMs function at WOT. I remember him posting that it actually was still used, not completely ignored. but maybe that was only up to 4500rpm..

This (at least temporarily) solved my issue. 91 325i automatic, poor idle, changed O2, TPS, and coolant sensor - still ran rough, Swapped to a known good ICV, and a known good AFM, switched boots, made sure no vacuum leaks - still ran rough..

Read through this posting - reset DME (Motronic 1.3) and now it idles and no check engine light.. driveable

I thought I read somewhere I need to run a series of trips for the motronic to learn and adjust to a more smoother idle operation - is that true? (i.e all the discussion on "learned adaptation" suggests this..)

[jlevie]

The DME will adapt with time and it will help to mix urban and rural driving cycles. A few hundred miles should do it. If the idle is still rough after that, something else is wrong and that will have to be identified and fixed. Other engine management problems may also explain why a DME reset worked.