robs,
Excuse my hiatus, I was doing some experimentation (which I will post about elsewhere on the MS boards)
Peter is correct.
Consider these 3 wires:
1) From TPS output in engine bay to MS TPS input pin
2) From TPS ground in engine bay to MS Sensor Ground pin which is connected via the PCB layout to the A/D input gnd
3) From engine block/head to MS high current ground pin
If wires #1 and #2 have no appreciable current through them, then there will be no voltage drop across them. Therefore the voltage at the MS from the TPS input pin to the sensor ground pin will be the same as the voltage at the TPS.
If the PCB layout is tight, then the voltage from AD3-1 to the A/D grnd pin will be the same voltage but filtered by the CRC pi filter.
Wire #3 carries large currents. These currents produce a DC voltage drop V = I*R.
Additionally said currents have di/dt. Wire #3 will have some inductance. This produces an additional voltage drop V = L * di/dt. This voltage tends to look spikey. If the di/dt lasts for say, 10 us, then the spike is about 10 us wide. This is the kind of noise easily filtered out by an RC filter. The sum of these 2 voltages is the noise voltage on wire #3. Therefore if you measure the voltage at the MS pins, from the TPS input,
to the high current ground pin, it will be equal to TPS voltage plus said noise voltage.
The assumption that there is no appreciable current going through wire #2 will NOT be valid if, due to the layout, there is a conductor, such as a PCB trace (or a polyfuse!), through which the currents for wire #2 and wire #3
both flow! It gets worse as the resistance or impedance of the shared conductor is increased. This is called "shared impedance" or "shared conductor" noise coupling. It has the very bad misnomer, "ground loop". This is what I suspect is happening with the polyfuse. (i.e. shorting it out reduces the noise)
On the MS PCB layout the sensor gnd and the high current ground pins are connected together via the polyfuse. The polyfuse separates the grounds of two groups of circuits.
The problem is that the polyfuse has resistance and any currents flowing through it will produce a voltage drop. This voltage drop appears as a voltage between the grounds of the 2 systems. (This is in addition to the shared impedance noise coupling above) The current through the polyfuse is the sum total of ground return currents of any currents of the signals that go from one system to the other. In addition to the IR drop, there is an additional drop due to the increased inductance from enlarging the loop area of the di/dt of said currents, producing spikey voltage drops due to the inductance. This is what caused the noise in the AEM that I fixed in this post:
http://www.msextra.com/forums/viewtopic ... tt#p315544
Clear as mud?