A noise gate that chatters on a high-gain patch is not broken, and it is usually not set wrong either. It is reading the wrong signal. An inline gate detects the same distorted signal it is trying to mute, and distortion is exactly the thing that destroys a gate's ability to find a threshold. The fix is to give the detector its own signal — a sidechain. Here is what that means, the hardware that does it, and how to rebuild it on a modeler with no cables at all.
The Short Answer
| Inline gate | Sidechain (key-input) gate | |
|---|---|---|
| What it detects | The signal passing through it | A separate signal you choose — your dry guitar |
| What it gates | That same signal | The distorted signal downstream |
| Detection on high gain | Poor — distortion flattens the envelope | Accurate — dry signal keeps full dynamics |
| Chatter / early cutoff | Common past the edge of saturation | Largely solved |
| Hardware example | Standard ISP Decimator II, most pedal gates | ISP Decimator II G String System |
The whole game is decoupling. An inline gate has to detect and mute on one signal. A sidechain gate detects on one signal and mutes on another. Past a certain gain level, that separation is the difference between a gate that works and a gate that fights you.
Why an Inline Gate Loses the Thread on High Gain
Distortion is compression with attitude. Run a clean note through a 5150-class preamp and the gain stage pulls the quiet parts up and holds the loud parts down until a decaying note and the background hiss are sitting maybe 2-3 dB apart. That is the envelope your inline gate has to read.
There is no stable threshold in that signal. Set it high enough to silence the hiss and it eats the tail of every sustained note; set it low enough to keep the sustain and the hiss walks right under it and the gate flutters. The threshold knob feels broken because the signal it is measuring has had its dynamic range squeezed out of it. Pick attack is everything here, and on a compressed signal the gate can barely see the transient that tells it you just hit a note.
Putting the gate before the distortion does not save you — the noise is generated by the gain stage, so a gate up front has nothing to catch and then you amplify the hiss behind it. The noise lives downstream of where the inline tool wants to sit. That placement problem is its own subject, covered in where to put a noise gate.
What the Sidechain Input Actually Does
A sidechain — also called a key input or detection input — is a second jack that feeds the gate's detector circuit independently of the signal path. You send it your dry guitar. The gate watches that clean signal, which still has a wide gap between a struck note and silence, and uses it to decide when to open. The gate itself acts on the distorted signal later in the chain.
Now the two jobs are on two signals. Detection happens where the dynamics are intact; muting happens where the noise is. The gate closes the instant your dry note decays past the threshold, not whenever the compressed distorted tail finally gives up — which is why the chatter disappears and the note tails survive. This is the same logic as the 4-cable method, which uses the amp's effects loop to achieve the same split; the sidechain gate builds the detection jack into the unit so you do not need the loop to do it.
Decimator II vs. Decimator II G String System
This is the buyer question, so here is the clean line. The standard ISP Decimator II is one input, one output, inline. It detects the signal that runs through it. It is a fast, quiet gate and it is enough for a lot of rigs.
The Decimator II G String System adds the dry-detection jack. You plug your guitar into the detector, run the dry signal on to your amp's front end, and place the actual gate after your distortion or in the effects loop. The "G String" is shorthand for guitar dry signal detection. That dry tap is the entire reason it costs more and the entire reason it holds up on a modern metal rig where the standard unit would stutter. Other gates take the same approach by other names — the Boss NS-2's send/return loop and the MXR Smart Gate's detection behave as a related idea — but the G String is the one most players mean when they say sidechain gate.
The Surprise: Release Time Wasn't the Problem
When my first high-gain gate chattered, I did what the forums told me: speed up the release so the gate snaps shut faster and the flutter stops. It did not stop. A faster release just made the flutter faster — a stuttering buzz instead of a slow one. I tried the opposite, a long release to ride over the noise, and that smeared the gate into the next note. I had assumed chatter was a timing problem.
It was not a timing problem. It was a detection problem. The gate's timing controls were all reacting correctly to a signal that had no usable threshold in it. The moment I fed the detector the dry guitar instead, the release time I had already set behaved — there was finally a real boundary for it to act on. The lesson generalizes past gates: when a control feels impossible to dial in, check whether the tool is even looking at the right signal before you keep turning knobs. A gate is downstream of gain staging, and so is most of what people blame on the gate.
The Modeler Version
On a modeler you get the sidechain for free, because routing is just block placement. Two ways to do it:
- Key input, if your gate block has one. Route the gate's key or sidechain input from immediately after the input block — your cleanest tap — and place the gate block itself after the amp or drive block. That is the G String System rebuilt in software.
- Two gate blocks, if it doesn't. Put one gate right at the input so it reads your clean dynamics, set its threshold just over the noise floor, then add a second gate after the amp block to clamp the noise the modeled gain stage adds. This two-gate approach is the standard modeler move and it gets you most of the way to true sidechain behavior.
Either way, the threshold finally sits in a stable place because at least one gate is detecting a signal that still has an envelope. And keep the expectation honest: a gate mutes the gaps, it does not remove the noise while you play — the gate vs. suppressor breakdown covers what each tool actually does to the noise itself.
What to Do Next
If your gate chatters and no threshold setting fixes it, stop adjusting the threshold. The signal it is reading is the problem, not the number. On a pedal rig, either run the 4-cable method with a send/return-capable gate or buy a sidechain unit like the Decimator II G String System and feed it your dry guitar. On a modeler, gate at the input, add a second after the amp, and use a key input if you have one. The detector needs a signal it can actually read — give it the clean one, and the gate you already own usually starts behaving.
