Here's what I posted over there:
There are no "magic" frequencies, but there are frequency ranges where certain types of problematic sounds lie. The range where "fizzy"sounds from distortion-generating devices generally lie is ca. 3.5-4.5kHz. This is true regardless of the type of source: tube amp, stompbox, or digital model thereof. It may seem more obvious when you play a modeler, because you can generate much higher levels of distortion at modest volumes than you can with a tube amp, but it's there with tube amps nonetheless.
There is no single frequency you can cut to reliably address this - every device and model will be different - but there is a simple procedure to identify the frequencies you need to attenuate.
1. Place a parametric equalizer in the signal chain after the distortion-generating components. If you're using a tube amp, you'll need to do this to the mic'ed/recorded signal from the amp.
2. Set Q to a modest value, say 4, frequency to something just above 3kHz, and level to substantial boost (ca. 6dB or more).
3. Sweep the frequency range upward until you've maximized the effect and gone past the peak, then set it back to the frequency where it is at a maximum.
4. Increase Q to ca. 6 and fine-tune frequency to narrow down the location of the peak.
5. Bring the level down past zero, and increase cut just enough to clean up the effect. Don't overdo this.
That's it.
