Through-Zero FM
When FM modulation goes deep, cheap implementations bounce the frequency at zero instead of passing through. That bounce makes it sound harsh. Through-zero FM goes through cleanly.
Hear the Difference
Start with the depth low. Both modes sound the same: clean FM. The modulator wiggles the carrier frequency just a little, and it never gets close to zero. No problems.
Now crank it up. Hear the buzz appear in Bounce mode? That's the oscillator hitting zero and reversing instead of passing through. Every time the instantaneous frequency would go negative, the cheap implementation takes the absolute value. The phase snaps direction, creating a sharp discontinuity in the waveform.
Switch to Through-Zero. The buzz is gone. Same depth, cleaner sound. The oscillator lets the frequency go negative, which just means the phase runs backward for a moment. No snap, no kink, no extra harmonics.
What's Happening Inside
The diagram below shows three signals over two modulator cycles. The modulator (top) swings the carrier's frequency up and down. When the depth is high enough, the instantaneous frequency dips below zero. That's where the two modes differ.
Why It Matters
Analog synths that do FM often bounce at zero because negative voltage-controlled oscillator frequencies don't exist physically. Digital synths have no such limitation. A phase accumulator can decrement just as easily as it increments. That's why digital through-zero FM (as in the Yamaha DX7 and modern Eurorack digital oscillators) sounds so much cleaner at extreme depths.
Here's what the oscillator is doing on every single audio sample. It advances its phase (its position in the wave cycle) by a tiny step. How big that step is depends on the frequency at that instant.
In bounce mode, the step size is always positive. If the modulator pushes the frequency below zero, we take the absolute value and pretend it didn't:
\[\underbrace{\phi}_{\text{phase}} \mathrel{+}= \frac{\overbrace{\lvert f_c + m(t) \cdot d \rvert}^{\text{always positive}}}{\underbrace{f_s}_{\text{sample rate}}}\]Where \(f_c\) is the carrier frequency (the note you're playing), \(m(t)\) is the modulator's current value (swinging between -1 and +1), \(d\) is the depth (how far the modulator pushes the frequency), and \(f_s\) is the sample rate (48,000 samples per second).
In through-zero mode, we drop the absolute value bars. The frequency is allowed to go negative, which means the phase runs backward for a moment:
\[\phi \mathrel{+}= \frac{f_c + m(t) \cdot d}{f_s}\]That's the entire difference. One character of math. At low depth, \(f_c + m(t) \cdot d\) never goes negative, so both modes are identical. At high depth, the bounce creates a sharp discontinuity (the phase jerks forward when it should be going backward), and that jerk is what you hear as buzz.
Try it. Use the A/B toggle to switch between Bounce and Through-Zero. Sweep the depth slider and listen for the point where Bounce starts to sound harsh. Try different C:M ratios for harmonic and inharmonic timbres.
Experimenting
With harmonic ratios (1:1, 1:2, 1:3), both modes produce harmonic timbres at low depth. Bounce mode adds inharmonic splatter as you push the depth up. Through-zero keeps the spectrum clean and symmetric even at extreme settings.
The inharmonic ratio (1:√2) is interesting because neither mode produces a purely harmonic spectrum. But bounce mode still sounds noticeably harsher and more chaotic. Through-zero is dense but controlled.