325 dies under way,..

Michael Yeager > wrote:
>On 22 Dec 2007 10:53:59 -0500, (Scott Dorsey) wrote:
>
>>Okay, if this is what I think it is, the ECM is just a little Intel 8051
>>with 4K of memory running in a little loop that doesn't do a whole lot
>>and has no facility for producing error messages. So you can't look at
>>the sensors with an analyzer, you can't pull codes off the ECM, because
>>the hardware just isn't in the ECM for that.
>
> This is most definitely a Motronic unit. Don't know where you got
>the above description but in the grand scheme of things it's partially
>correct.

I got it from taking a bunch of ECUs apart.

>Any ECU that's running a fuel injected engine operates in a
>fairly simple fuel mapped mode that uses the oxygen sensor to slightly
>modify the fuel injector timing to keep the proper mixture. They use
>an array of sensors to determine engine temperature as well as the
>position of the crank or cam as a timing reference and to measure the
>amount of air the motor is pulling in. If everything is working
>properly, the injectors spray the correct amount of fuel at the proper
>time and the coil fires the spark plug at the right moment and the
>motor runs.

Yup.

>Different versions of Motronic have different capabilities
>for self diagnostics and use different sensors but they all function
>basically the same.

What self-diagnostics are in there? According to my dealer mechanic
and one local independant mechanic, there is no error reporting at all
from the earlier Motronic boxes. The later ones have a 3.5mm connector
on the back of the ECU connected to the UART on the 8051, but nobody
seems to have anything that can be plugged into it.

Also there is the "diagnostic connector" under the hood, which appears
only to be useful to provide a spark sensor, TDC sensor, and to reset
the oil change and inspection lights. I don't see any other signals
besides those, at least on my E28. The dealer tech said that there was an
analogue diagnostic machine that plugged into it at one time but that
"nobody has any of those any more."

> In this case, the motor appears to run well until this occurs. The
>original poster has said nothing or the motor searching for idle so I
>believe we can remove the suspect of vacuum leaks. Either the Motronic
>relay or the fuel pump relay would instantly cause the car to shut
>down and not restart. If there is spark, the Motronic relay is not at
>fault. The actual ECU would not be at the top of my list. There is a
>connector underneath the intake manifold that has a history of
>collecting moisture and corroding, check the pins in there to make
>sure it's clean. Not sure what signals go through there on your model.

I agree that the ECU is not high on the list, but the majority of ECU
failures seem to be due to cold solder joints and are easy to diagnose
with a sharp rap.

The multipin connector under the intake is a really good call, I like
that one.

>Crank reference sensor on my 89 had a broken wire right at the sensor
>itself and until I replaced it and bent the wire back I never knew it.
>This caused the car to run fine for a while and then do exactly what
>yours is doing. Biggest pain in the ass I ever tried to find. Got one
>from a junkyard and changed it (figured at $100 for a new one, what
>the hell) and the problem went away. The compound where the wires go
>into the back of the sensor was all missing on my original sensor.

Makes perfect sense. Can you explain why a bad oxygen sensor can cause
the same problem? I have seen a 328 where the sensor wiring was damaged
in such a way that the vehicle would run fine with the sensor removed
(and the ECU presumably ignoring it), but would die intermittently with
misfires with the sensor in place. That one still makes me scratch my
head.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

Michael Yeager > wrote:
>On 23 Dec 2007 10:13:32 -0500, (Scott Dorsey) wrote:
>>What self-diagnostics are in there? According to my dealer mechanic
>>and one local independant mechanic, there is no error reporting at all
>>from the earlier Motronic boxes. The later ones have a 3.5mm connector
>>on the back of the ECU connected to the UART on the 8051, but nobody
>>seems to have anything that can be plugged into it.
>
> Your mechanic may be correct on the early boxes. However, they test
>the sensors and light the check engine light but you can't get the
>code for what sensor is reading bad without a code reader. I have seen
>these on EBay as well as a few other sites but don't know what their
>capable of. Later versions allow you to read the code on the check
>engine light by pressing the accellerator pedal all the way to the
>floor five times an a certain amount of time. The ECU then flashes the
>check engine light in pulses that can be counted and the code looked
>up from there. I see from a later paragraph that you have an E28,
>these definitley had an early motronic version. At least it isn't
>J-Tronic (what a difference!!!).

Yes, the E28 has no check engine light. Mine _does_, however, have the
useless 3.5mm jack on the back of the ECU but apparently not all do.

The later units that allow use of a code reader.... where does the code
reader plug in?

>>Also there is the "diagnostic connector" under the hood, which appears
>>only to be useful to provide a spark sensor, TDC sensor, and to reset
>>the oil change and inspection lights. I don't see any other signals
>>besides those, at least on my E28. The dealer tech said that there was an
>>analogue diagnostic machine that plugged into it at one time but that
>>"nobody has any of those any more."
>
> I've wondered what information could be gathered from there but
>since I haven't seen much talk about it from the tech community, I
>didn't figure it was all that important.

I don't think anything much useful can be, other than TDC. Which
admittedly can be a very useful thing to know in conjunction with looking
at other signals.

>>>Crank reference sensor on my 89 had a broken wire right at the sensor
>>>itself and until I replaced it and bent the wire back I never knew it.
>>>This caused the car to run fine for a while and then do exactly what
>>>yours is doing. Biggest pain in the ass I ever tried to find. Got one
>>>from a junkyard and changed it (figured at $100 for a new one, what
>>>the hell) and the problem went away. The compound where the wires go
>>>into the back of the sensor was all missing on my original sensor.
>>
>>Makes perfect sense. Can you explain why a bad oxygen sensor can cause
>>the same problem? I have seen a 328 where the sensor wiring was damaged
>>in such a way that the vehicle would run fine with the sensor removed
>>(and the ECU presumably ignoring it), but would die intermittently with
>>misfires with the sensor in place. That one still makes me scratch my
>>head.
>
> Only thing I can think is that the sensor shorted to ground or
>became open (whichever causes the ECU to richen the mixture) and the
>mixture becomes so rich it won't fire. That's the ONLY thing I can
>think of for that. I did see an oxygen sensor so screwed up that as
>soon as it got up to operating temp and the ECU tried to go closed
>loop and run off the oxygen sensor input the car would flood out and
>die. Unplugged it and the check engine light came on but the car ran.
>A new sensor (for a 1990 5.0 mustang) solved the problem. $35 for the
>sensor and five minutes rewiring the plug beats $90 for the same
>sensor with the BMW plug...

I can see that. That is _very_ rich....
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."

On Sat, 22 Dec 2007 11:24:41 -0600, "Bill" > wrote:

>Oh, yeah, stalls at stops. I bet it idles a little high when started cold
>and the RPMs hunt. The Idle Control Valve may need a cleaning. On top next
>to intake manifold, it's a silver cylinder about 5 inches long with a big
>hose in and out, maybe like a "T" shape with a 3-pin connector on the other
>end.. If it sticks, clean before replacing, it is very expensive.
>
>Bill in Omaha
>'86 535i
>
It seems to idle fine, even when cold. No hunting and quite smooth for
a high miles engine. When it dies, there's no sputtering or trying to
stay alive. It all just stops in mid stroke and you coast along in
silence.

>>'86 535i
>>
> It seems to idle fine, even when cold. No hunting and quite smooth for
> a high miles engine. When it dies, there's no sputtering or trying to
> stay alive. It all just stops in mid stroke and you coast along in
> silence.

It does sound like an intermittent connection.
I would suggest that you may find it by trying to induce the failure while
it's running normally.
With the engine idling nomally, just start wiggling any and all electrical
connections and wiring runs.
Don't forget under the dash and around the ignition switch.
I wouldn't suggest beating on the computer but you could flex it a little
and flex the harness going through the firewall.
Crank and cam postion sensors should be rapped with the back end of a
screwdriver.

Jack wrote:
>>> '86 535i
>>>
>> It seems to idle fine, even when cold. No hunting and quite smooth for
>> a high miles engine. When it dies, there's no sputtering or trying to
>> stay alive. It all just stops in mid stroke and you coast along in
>> silence.
>

>

my 88 325is did that for about 2 weeks before/as the fuel pump died

Scott Dorsey wrote:

>Okay, so it's getting fuel, and it's misfiring. I agree this sounds
>like an ECM or sensor issue.

Was there ever designed an engine-management system that would drop
dead if a single sensor went bad? I thought they all have "limp home"
modes, should that happen.

Michael Yeager > wrote:
> There are always going to be certain things that will kill it dead.
>Only way around that would be completely redundant systems. The
>"hall-effect" sensor (known comonly as a reference sensor) will kill
>the car dead as the computer has no idea if the motor is turning over
>or not. Also, even if it tried to assume that the motor was running,
>it would have no timing signal to base anything on. Some sensors will
>make the car run rough or sluggish, others kill it dead...

With the old Motronic system, though, a lot of things will kill the
engine dead that on more modern and robust engine controllers will
enable a limp-home mode. For example, the oxygen sensor failures or
the failure of the air mass sensor.

What do you want from 4k of code? It's not a Cray in that box.
--scott
--
"C'est un Nagra. C'est suisse, et tres, tres precis."