I am in no way against a solid and functional knock detection system, although I have concerns (below). A common knock module is not universal, and yes they pick up sounds you can't with the bare ear, and why they often return false positives. It's fine if it can be tuned for the engine's knock frequency, and also tuned to filter noise of that engine design, and detect the frequency in the pattern that is actually knock. That's the big issue. As an analogy, sensing knocking sounds is easy but this is like trying to hear one person, in a sports stadium with thousands of noisy fans, and the guy you're trying to hear is speaking in a foreign language. Noise, frequency, pattern. Just to complicate things, every engine has it's own response signature (different stadium, country and language). So, unless you have a pre-matched filter and interpreter (such as a GM sensor and GM tuned (matched to that engine) module on that GM engine type, tuning the knock detection and filtering system is a considerable task to do reliably and effectively. GM spent millions developing theirs and it's not 100% either. If it's not as reasonably reliable and effective, it's more dangerous or inhibiting than nothing, and that has been the case so far with 'universal' systems for all our engines.
Without that developed OEM-type setup, the alternatives are many and generally use more than one, such as plug reading with EGT monitoring or electronic ears on piezo sensors. Each is a reliable method if used properly, and does not require lots of skill or experience to use effectively, as proven by decades of amazing engines not using KS. Sure, I'd love to have an auto-tuning knock system that works 100% on any engine without destructive testing and I hope one is developed soon, but until then we have to use the best available methods. Why say this? Because you want to create a microchip program that will do what we can do today, but better. What will that take? Well, legitimate questions to define the KS project: Without a dedicated pre-tuned knock system as described, what other effective methods are suggested for Speeduino users today, and how do we translate those to this KS system? Can we teach users how to then tune a knock system without damage, as it (and the user) need to see/hear and identify that engine's knock to know what it's avoiding? How do you prevent the common problem of users abusing it by allowing it to "control" timing, losing power overall on both retard and over-advance, and hitting knock with potential damage every 4 seconds (if
the module is tuned well)? What is the true purpose of a Speeduino knock sensor, and is it potentially more dangerous than safe in these regards? This ad-hoc project needs to know these answers and more in order to have a definition of capability and usefulness for it to succeed. I want to see it really work, rather than hit unexpected hurdles, slowing or stopping the effort.
Actual knock sensor output (resized). It's your job to tune it. Do you see any knock? Where? How do you know?
OK, so through a spectrum analyzer should help make it easy for a newb, or an expert to develop auto-tuning for it:
Ion-sensing ignition timing control is next, making KS irrelevant.