[s-cars] No Start ABY/AAN engine - Problem Found?

[John Cody Forbes]

On Oct 2, 2010, at 6:22 PM, Huw Powell <audi at humanspeakers.com> wrote:

> 
> 
> 
>> Second, you do not want the pins on the cam. The timing belt has slop
>> and the valve springs cause the cam to bounce back and forth in this
>> slop. This will cause an ignition timing variance. Because it's on
>> the cam that variance will be doubled. You do not want your ignition
>> timing bouncing around +\- 10 degrees ;-).
> 
> So how come cam-driven distributor spark control systems don't seem to do that?
> 
> Not saying they are rock-solid, but they sure don't bounce around that much.

Well I'd be purely speculating on this, but I've got some guesses. 

One, they are designed, and the ECU programed, with the 5-window dizzy in mind. The turbo car setup is expecting an extremely accurate signal and would be programed and designed as such. 

Two, hall sensors output a fixed voltage (+5vdc falling to ground) square wave where the ECU triggers on the vertical falling edge. VR sensors, as the flywheel pins, output a slightly less precise sine wave (variable voltage +vdc to the inverse -vdc then resting at 0vdc) that triggers on the zero crossing point only after reaching a peak of at least 2.5 or 3volts depending on the ECU. In the context of the 5 window ignition you get much more accuracy versus the VR system Ben was considering. VR works on the flywheel due in part to the larger diameter giving better resolution, and partly because the flywheel spinning twice as fast gives a stronger signal (more voltage over the same time period therefore a more vertical falling edge) from the VR. A VR on the cam would possibly not deliver a strong enough signal to trigger the ECU during cranking.

Three, possibly the variance is fairly low, but on a boosted engine one degree can make a big difference between ok and oh no. Certainly Audi spent a great deal extra money to produce the one window + flywheel system. Extra cost for adding the VR bracket to the block, the two VR sensors, the cost of machining flywheels with the dowel, making new hall rings, a different ecu, etc adds up to likely millions of dollars of cost if you add up the development and production costs over the couple hundred thousand type 44 turbo cars made. They would have spent that money on better lawyers to sue 60 Minutes if they didn't feel it necessary to achieve the greater accuracy ;-).

Again this is purely speculation based on experience. The only real world facts in the above are those pertaining to sensor waveforms. I have logged hundreds of hours on a scope deciphering different automotive sensor waveforms and have a large volume of experience with custom ECU systems using sensors of various types for various applications. On my own 5ktq with coil-on-plug ignition I personally failed at using a VR sensor for the cam true-TDC reference signal initially until I amplified the sensor output.

Following this paragraph is a link to a scope output showing 5 flywheel VR triggers on top in blue, and the cam reference on bottom is red. You'll note the varying aptitude of the blue trace. My machinist made a mistake centering the flywheel on his rotary table and the 5 dowel pins were installed offset by about 0.25mm (the closest was half a mm less sensor gap than the farthest). HALF OF ONE MILLIMETER makes a huge difference in signal strength. The flywheel was ruined and I started over, this time doing the mill work myself. Also notice how comparatively weak and lazy the red cam signal is to even the worst (but faster moving) VR signals.

http://jcforbes.com/jcfpics/6-4-10-ms-cop/audi_vr2


-Cody