Problems with crankshaft frequency waveform

Hi, everyone.

I've lately been learning how to incorporate exhaust pulse and crankshaft frequency testing in order to identify misfires. 

Getting more and more American imports here in Poland, so it's become useful with all those V6s and V8s with coils under the intakes.

Anyhow, as much as I think I have figured out using exhaust pulses, I cannot, for the life of me, manage obtaining a usable crankshaft frequency!

I was wondering am I doing somehting wrong or is my scope my limiting factor? I'm using a Picoscope 2204a with 100 MS/s.

If I use a longer timebase, the frequency waveform comes out... spikey.

What gives? No amount of playing with the frequency base changes anything, it's still unusable. Changing the sample rate achieves nothing either.

Wrong scope? I know the 2204a isn't listed as an automotive scope, but it's still used by many with successful results.

The second attached picture shows the "best" I could get at 10ms/div but it's still too little data. You can see a slight suggestion of a downward slope. Any more timebase and it's spikey.

P.S. what do you guys like using to identify "fishbite" style misfires? The kind sthat appear only once in a while, but a very sensitive driver might pick up? I've heard that the crankshaft method isn;t as accurate then. Apparently, an exhaust pulse would be better.