We are using a MAX9926 IC to filter 2 VR signals (from a HONDAS2000 Crank and Cam VR sensors) One being 24 tooth, and one being 3 tooth (per cycle of the Engine). At RPMs below approx 5000-6000 the signal comes through the 9926 fine, and triggers the MCU correctly such that it decodes it fine. At around this RPM we start to get extra/missing teeth on the IRQ lines to the MCU. Note we can 'push through' this noise and get to 9k RPM – just seems to be a 5-6K RPM issue…
We also used a scope to capture the signals at the entrance of the ECU to rule out electrical noise. The signals are smooth and clear at both idle, AND when hitting the point the ECU loses sync due to the 'phantom' teeth. The scope doesen't show these phantom teeth so I am assuming it is a 9926 issue, or the parts R1, R2, C1, R3, R4, C26. The only difference between perfect idle signal and higher speed 'issue' signal is the frequency of the wave, and its amplitude, as shown in the plots. We have 50v P-P at high speed and around 20v P-P at low speed.
1) Maybe I should and how could I attenuate both signals say by half? (Adding what, where)
2) Should I reduce the value of C1/C26 as they might be acting as a filter?
With regard to (2) if I keep the throttle down and 'push through' the noise the engine will regain sync and run all the way to 9k RPM without issue, so that is also odd!
Schematic of the input system: