Intermittent no start
 

Sometimes the car E320 1994 have no start, it always start but sometimes you have long crank time and let go of key and crank one more time and the engine start right up. When the engine won't start there is no spark. Is the only input for spark crank sensor? No trouble code.

Bad crankshaft sensors seem to be common, also, these years had wiring insulation that crumbled. ( Search Mercedes bio degradable wiring )

I'm pretty sure that anti theft on these years only disables the starter from turning.

The car have new wiring harness, next step is crank signal to bad not to have a scope.

You can check the crank sensor with a AC volt meter, these sensors are a coil of wire wrapped around a core / magnet. I "think" that 2 VAC at cranking speed is good.

Can also check with a Ohm meter, not sure what it should read but I'd expect ~ 800 Ohm?

If you have a wiring diagram, you can check for signal at the ECM connector.

Also, if you lose crank signal, you won't have injector pulse either.

Problem solved
 

Bad crank sensor:)

Good news. How did you determine this to be the cause? Did you swap it, or measure it? I was just going to post that some meters don't measure AC well when it isn't 60hz...

The problem was no spark so I removed the crank sensor and clamped the sensor in a vise and a portable electric drill rotating at the sensor then I got AC voltage and when I tapping on the vise the signal got lost, and when tapping vise again the signal came back. The OVP relay is also a culprit and the symptom is the same but i had a brand new one so I ruled out the OVP relay.

I'd hope that most meters can measure AC volts over a larger scale than 60 Hz. I'll dig around some manuals and see what the freq range is.

Haven't heard of a restriction in low frequency on an AC voltmeter, but one always can learn more.
There is a restriction in frequency when it gets beyond 30 kHz (if I remember right), this is considered to be RF (Radio Frequency) but below that I don't see why they couldn't.
Some DMM's have a frequency counter (not very accurate) built in and usually can measure up to 1 MHz. The problem is the waveform, if it deviates from a perfect sine or a 50% duty square wave, it will be inaccurate.
Actually I was saying "Thank you" for posting the findings and the remedy to the problem.

Very creative test, I must say, yet effective!