Fuel levels toggle back and forth between just
"a little rich" and just "a little lean"

Idle
Partial Open Throttle
The toggling speed of the meters are a good approximation of my 84 Porsche
944. You will notice that the toggling is much slower when the car is idling,
faster at part throttle. At wide open throttle (WOT), the O2 sensor is not used
by the DME, there will be no toggling and the meter will show rich.

Description - How the DME monitors
oxygen levels in exhaust gas
This Motronic
Lambda system takes the raw Lambda (l) sensor input and
converts this
signal so that a computer can understand it. Some computer programming and flow charting
will be discussed. Oxygen sensor and O2 sensor are other commonly used names for the
Lambda sensor.
The Lambda is a
minor system in the Motronic. It is an option, an add-on. The first of the European
& ROW Motronic equipped cars had no Lambda capability. It is not a critical
system. It is a smog device and does not enhance engines power. It does just the
opposite. What the Lambda system does is fine tune the fuel to maintain an ideal
air to fuel ratio of 14.7:1. The Motronic ECU meters out fuel according to preprogrammed
fuel maps. The Lambda system then makes minor fuel level changes with feedback
from in the exhaust gases (oxygen levels).
<--- Voltage
Divider Circuitry
BOSCH,
the designers of the Motronic fuel injection systems, defines the following
levels:
VLean
= +0.45V
VReference = +0.475V
VRich = +0.50V
A simplified
Lambda circuit for an early Motronic ML1.2 is shown below. At left is the voltage divider
section made up of four precision resistors between +5
Volts and ground. This (Vl) increases and passes though each of the threshold voltages, the output of
one of the comparators will changes.
Strategically between VRich
and VLean, is a midway point called VReference.
This leaves a ±25mV wide window
between lean and rich. If the Lambda sensor is not connected, the input to the comparators
will remain at VReference. If the Lambda sensor is present,
the comparator inputs will toggle back and forth between 0V and 1V.

The Lambda System Schematic

This schematic has been simplified. The Lambda
sensor connects to R801, and with C801, make up a low pass filter. Noise frequencies above
1Hz are filtered out. Through R808, this point is connected to VReference. R802, R803, R804, and R805 make up
the voltage divider as previously described. Each comparator is wired to trip at a
different threshold voltage. The comparator outputs go to the 1802 processor.
EF3 and EF4 are hardware interrupts. They set
a flag that the processor routinely monitors. Once alerted to the flag, the processor
investigates the cause. Opposite to what is commonly thought, the processor does not
constantly check the Lambda's status. The Lambda circuitry tells the processor when there
is a change in status and then the processor gets involved.
The outputs of the two comparators change as
the inputs transition from rich to lean and back again. The
following table shows the binary equivalent of the Lambda output voltage. This is how an
analog input Vl is converted to a
two bit binary number.
Name |
Lambda Output Voltage (Vl) |
Voltage |
Output |
Binary |
| EF3 |
EF4 |
EF3 |
EF4 |
| Rich |
> 0.50V |
0 |
0 |
0 |
0 |
00 |
| In-between |
0.45V to 0.50V |
5V |
0 |
1 |
0 |
10 |
| Lean |
< 0.45V |
5V |
5V |
1 |
1 |
11 |

Lambda Voltage vs. Time

This is an ideal representation of the Lambda out voltage vs. time. This is a engine at
idle. At part throttle, the repetition rate is about double. TD
is dead time. During dead time, all fuel level adjusting activities stop. Dead time is
about double for an idling engine than at partial open throttle. It is less demanding at
idle.

This is a less ideal diagram but more realistic. The Vl
oscillations, shown here between the two threshold voltages, is from "cylinder
scatter". Differences in each of the four cylinders can cause the Lambda voltage to
jump about and even go in the reverse direction that the Motronic is trying to correct to.
Starting with a slightly lean condition, the
output is 11binary. The fuel is
increased while changes in the output is monitored. As the output changes to 10binary,
we enter the 50mV neutral area and continue incrementing the fuel level. When the output
changes too 00binary, the fuel incrementing stops. There is a
wait period. After the wait period is passed, the fuel level is decreased incrementally
from 00binary, passing through 10binary,
and 11binary and again it stops for a wait period. This cycle
continues endlessly.

Lambda Two Level Control Process

The basis ML1.1 system, the early Euro 944, has no Lambda system. The air flow sensor
and speed sensor are used to determine fuel flow from a map that is stored in system
memory.
Next to it is the ML1.2 Lambda system flow chart, the early USA 944, which has a Lambda
sensor (O2 sensor). The fuel level is a two level control process, map and Lambda control.
The Lambda system also uses the air flow sensor and speed sensor to determine fuel flow
from a map. The Lambda control makes positive or negative changes to the fuel level in
order to fine tune the fuel output. The range of the Lambda control is quite limited which
requires the fuel map to contain values very close to Lambda=1 (air/fuel ratio 14.7).
Comparing the maps of a non-Lambda (Euro 944) and Lambda system, the Lambda maps will
appear flat and unimpressive. The power is in the non-Lambda map.
The fuel map is always used with a Lambda system. One common but false saying is that
the fuel map is no longer used when the car is equipped with an Lambda sensor. This is far
from the truth. Where does the original fuel value come from? Is it a guess? No. Any
change in load or engine rpm, the DME must go to the fuel map to get its initial value.
Then the fine tuning begins. The Lambda sensor subroutines take charge and the fuel is
continuously toggled back and forth. If the Lambda sensor is disconnected, the Motronic
gets it fuel levels from the maps. It never just "goes rich to protect the
engine" which is another famous and false saying. When the Lambda is
disconnected, it uses the fuel map.
For simplicity sake, the contribution of temperature has been omitted from the
discussion. Fuel atomizes poorly at low temperatures requiring some mixture enrichment.
