A common problem with the early 944/951 cars is a no-start, or intermittent start situations. Typically this situation is due to the crank & angle sensors (generally known as the Speed & Reference sensors) and/or the wiring.
Though this is a fairly well known issue, what might be less clear is exactly why. The tests that can be done with common mechanics tools will typically only show a complete sensor or wiring failure. What they won't show is an intermittent, or occasional issue. And many times this results in an improperly running vehicle, without the owners knowledge.
The DME, or any engine computer for that matter, needs to know engine crank-angle at all times. To do this, the 944/951 use two sensors. One sensor to identify TDC - Top Dead Center, and the other to read crank-angle change. The sensors that are utilized are known as a Variable Reluctance type. Feel free to google for more information, but what is important is that these are an analog sensor. To utilize analog signals, the computer needs to convert it to digital. The DME does this internally using a custom-made integrated-circuit. Because this transition from analog to digital is the heart of the DME knowing crank-angle, that is where we will focus.
So, on my test-bench, I setup my scope to watch both the VR crank sensor and the post-processed digital signal in the DME.
Here is a proper operation:
The Cyan trace (CH2) is the VR speed sensor input to the DME. The Yellow trace is after the DME hardware reads the sensor and converts it to a proper digital signal. Note that peak-to-peak voltage of the VR sensor is dependent on tooth speed and air gap distance. You can see the voltage peak-to-peak of 7.52 volts, and we get a clean square-wave digital pulse for each converted sine-wave period.
Now look at this one:
In this shot, the VR sensor's peak-to-peak voltage is quite a bit lower. This is due to spinning the trigger wheel much slower. In-fact, at the second sine-wave period we can actually see it has slightly less amplitude than the others, which makes it just low enough for the DME hardware circuitry to not properly recognize this tooth. The result is the DME not converting this pulse to a digital signal, and therefore the DME misses the tooth completely! This would result in improper ignition timing among other problems at minimum.
Now, the scope screenshots show the many teeth per revolution signal (130 teeth for the 944, and 132 teeth for the 951). However, the other crank sensor picks up a single tooth (setscrew actually) per revolution. This is the sensor that the DME uses to get initial crank-position, to "synch-up". If this sensor does not properly get converted from analog to digital, the DME will not identify initial crank-position, and will not try to start/run the engine. The DME only gets one chance per revolution to synch.
It should be clear that if there is an issue with either of these sensors' signal, it can cause significant starting/running issues.