Blog / Mixing

The Sidechain HPF Filters What a Compressor Hears, Not What It Compresses

May 22, 2026 7 min read Mixing
sidechaincompressionmix-busde-essingdsp
The Sidechain HPF Filters What a Compressor Hears, Not What It Compresses

Most engineers use the threshold, ratio, attack, and release knobs on a compressor. Some go further with knee shape and parallel mix. Very few touch the sidechain filter. That is unfortunate, because the sidechain filter changes what the compressor reacts to without changing what the compressor processes, and that distinction solves several mixing problems that the standard controls cannot.

Two paths inside every compressor

A feedforward compressor splits its input into two parallel branches the moment the signal enters the plugin. Branch one, the signal path, goes to a VCA, FET, or gain-cell that will eventually apply attenuation. Branch two, the detector path, goes through a rectifier or RMS-detector, then through the gain computation stage where threshold, ratio, and knee live, then through the attack and release smoothing. The output of the detector path is a control-voltage style gain-reduction signal that modulates the gain element in branch one.

This is documented across FabFilter, iZotope, and Sonible learning material in nearly identical language because the architecture is universal. The audio you hear is the unfiltered input multiplied by the gain envelope that the detector decided to apply.

Anything you do to the detector path, EQ it, high-pass it, swap in an external source, does not touch the signal path. The audio still contains the full spectrum. Only the trigger logic changes.

What a sidechain HPF actually does

When you engage an internal sidechain HPF on a bus compressor and set it to 100 Hz, the kick drum and low bass content are removed from the detector signal. The compressor no longer 'sees' those low-frequency peaks, so it stops slamming gain reduction every time the kick hits. The bass content in the signal path is unchanged. It still passes through the compressor. It still receives the gain reduction value the detector computed from the mids and highs.

The result on a mix bus is glue without pumping. The kick stays punchy, the bass stays present, and the modulation of the wider spectrum that everyone calls 'breathing' stops being driven by sub-bass events. This is the technique Nail The Mix and others call 'the sidechain filter hack' for stopping mix bus pumping.

Frequency settings that tend to work

There is no industry-standard frequency, but practical ranges have emerged:

  • EDM, hip-hop, and bass-heavy genres: 100 to 120 Hz, sometimes higher
  • Rock, indie, and pop: 60 to 80 Hz
  • Mastering bus: 60 to 80 Hz, biased low to preserve sub presence
  • Vocal compression: 100 to 150 Hz to stop plosives and proximity rumble from triggering gain reduction

The SSL G-Series 500 Series module documents five fixed positions at 30, 60, 105, 125, and 185 Hz, which gives a reasonable sense of where the designers thought the useful range lived. The exact filter slope on that module is not consistently specified across third-party sources, so the safest claim is that it is a Butterworth-style filter in the 12 dB/oct neighborhood.

The SSL hardware that never had one

The most commonly cited example of sidechain HPF usage is 'the SSL trick' on the 4000 G Series bus compressor. The catch is that the original SSL 4000 G console bus compressor did not have a sidechain high-pass filter at all. Universal Audio's documentation for their UAD SSL 4000 G Bus Compressor plugin states explicitly that the SC Filter knob is a plugin addition with a continuously variable 12 dB/oct filter sweeping between 20 Hz and 500 Hz. The original hardware did not provide that control.

The mix bus glue heard on hit records from the late 1980s and 1990s, the era when the SSL G console became canonical, was done without sidechain filtering. Engineers either lived with the pumping, used very low gain reduction values, or worked around it with serial processing. The sidechain HPF entered the SSL world later, through the 500 Series G-Comp module and software emulations.

This historical detail matters because the technique is often described as a recreation of how the classics were made. It is not. It is a control that modern engineers gained access to decades after the SSL G's hardware design was frozen.

De-essers are the same architecture

A de-esser is structurally identical to a sidechain-filtered compressor, but with the filter pointed at the wrong end of the spectrum on purpose. Instead of removing bass from the detector, a de-esser boosts a bandpass region between roughly 3 and 10 kHz so the compressor 'hears' sibilance louder than the rest of the signal. When an S, T, or SH sound crosses the threshold, the compressor ducks the full-range output for the duration of the event.

The mechanism is the same. The filter target is different. Wikipedia, LedgerNote, and FabFilter all describe it this way.

This is why a compressor with full detector EQ, like the FabFilter Pro-C 2 with its low cut, high cut, and bell band on the sidechain path, can do double duty as a de-esser. Boost a bell at 6 to 8 kHz on the detector, set a fast attack and release, and dial the threshold until only sibilants trigger gain reduction. The result is a working de-esser using the same plugin you bought for bus compression.

Detector EQ as a frequency-dependent threshold

Boosting a band on the detector path is mathematically equivalent to lowering the threshold for that band. The compressor sees that frequency range as louder than it actually is, so it crosses the threshold sooner. Cutting a band on the detector raises the effective threshold for that range.

This means a sidechain EQ is a way to set a different threshold per frequency band without using a multiband compressor. The signal path stays single-band, with all the phase and transient behavior that implies, but the trigger logic becomes frequency dependent. For mastering or careful bus work, that is often a better compromise than splitting the signal into bands.

What older opto compressors did automatically

The LA-2A and LA-3A do not have a sidechain EQ control, but their control element imposes one. The T4B electro-optical attenuator in the LA-2A consists of an EL panel driving a pair of light-dependent resistors. The EL panel has a non-flat frequency response in its drive current, and the LDR has its own time and frequency behavior. The net result is that these compressors respond more strongly to mids than to lows, by accident of their electronics rather than by design.

Universal Audio's product material describes the photo-electric cell's frequency response without using the words 'detector EQ', and that is probably the right framing. It is not a sidechain filter. It is intrinsic frequency weighting that ends up behaving similarly.

The A/B test that makes it real

The fastest way to understand what a sidechain HPF is doing on a mix bus is to bypass it. Pull up a mix with a busy kick, set the bus compressor to 2 to 3 dB of gain reduction with a slow attack and auto release, engage the sidechain HPF at 100 Hz, and watch the gain reduction meter on a kick hit. Then bypass the HPF. The gain reduction needle will start slamming on every kick event, and the wider mix will modulate visibly in time with the low end.

That modulation is the pumping the filter was hiding. Whether you want it gone or not is a creative decision. Knowing that the filter changes what the compressor hears, not what the compressor processes, is what lets you make that decision deliberately.

Sources