The dash wheelspeed has been junk since I have the car, mostly ~0 (showing some activity but showing usually much lower than real speed), later speedo's been staying at 0 permanently.

Now I have the VEMS v3 ECU logging the wheelspeed to SD-card, and [Display-only Round] showing the ECU wheelspeed in digital format

* among other runtime data, the wheelspeed is sent over the RS232 in [AIM format], besides wheelspeed: lambda, MAP and EGT are also displayed in my setup (using 2 display-only rounds).

Wheelspeed Sensors overview - some more info below

* transmission front-left shaft (3-pin AMP connector, but only 2 pins used with a reed-relay between pin2,3).

* 4 high-pulsecount ABS sensors (VR ?)

I plugged in [PS2 wheelspeed cable] to the PS2 connector of my old 3B ECU with (using 2 wires: pin2,3 of 3pin injector connector for the audi transmission side)

* it worked the first time

** calibration: 112 Hz (at 100 kp/h) with only rising edge: 27.7 m/s / (112 1/s) = 0.248 m (sounds like 8 pulses: 0.63m*pi / 8). Verification needed.

* After the winter the ECU-wheelspeed showed mostly 0 or 255.

** We suspected firmware problem first (fw was upgraded a few times), but on bench all was perfect.

** We installed a 1uF SMD (0805 size) capacitor in the DSUB9 of the wheelspeed-cable between the Base and Emitter of the BC817 (sot323 SMD) NPN transistor.

** Soldering was easy: there was provision for a 0805 component. The 1uF * 2k7 input pullup = 2.7msec which limits the max freq to appr 350 Hz ~315km/h (the top-speed is below 290km/h anyway). 220 nF was almost enough, but there was still some noise. The 1uF seemed to fix it perfectly.

----

Calibration

Anyone knows the number of pulses per wheelrot ? 8 or 9 ?