Two Philosophies of Gain Reduction: Vari-Mu and Optical Compression

Vari-mu and optical compressors are often grouped together as "vintage" compressors, but electrically they have very little in common.

In a vari-mu design, the tube itself controls the gain. In an optical compressor, gain reduction is handled by a light-sensitive element controlled by the sidechain.

That difference affects the way the compressor responds to level, how it recovers after a loud passage, and how much of its character comes from the gain-reduction element itself.

It also explains why these two types of compressor tend to find different jobs in a studio.

What happens inside a vari-mu compressor?

The name comes from variable μ, or variable amplification factor.

A suitable tube is operated so that changing its control-grid voltage changes its gain. The sidechain detects the signal level and generates the control voltage. As the signal becomes stronger, the operating point of the tube shifts and its gain decreases.

The important part is that this does not behave like a simple electronic switch between "no compression" and a fixed compression ratio.

In traditional vari-mu circuits the transition into gain reduction is gradual. The effective ratio and knee change as the tube moves through its operating range. This is one reason these compressors can reduce several decibels without making the transition into compression particularly obvious.

The Fairchild 660/670 and Manley Variable Mu are well-known examples of this approach.

The details differ considerably between designs. Tube type, sidechain circuit, amount of feedback, transformer loading and the attack and release networks all affect the result. "Vari-mu" describes the method of gain control, not one specific sound.

What happens inside an optical compressor?

An optical compressor separates the detector electrically from the gain-control element.

The sidechain drives a light source. A light-dependent resistor, or another photosensitive device, responds to that light and changes the attenuation of the audio signal.

In the classic arrangement, more light causes the photoresistor's resistance to fall. The audio attenuator is arranged so that this produces more gain reduction.

It sounds simple, but the interesting part is the behaviour of the optical element over time.

A photocell does not necessarily return to its original resistance immediately when the light disappears. Its recovery can depend on how brightly it was illuminated and for how long.

A short peak and a sustained note can therefore produce different release behaviour even if the peak gain reduction shown on the meter is similar.

This is particularly noticeable in designs derived from the LA-2A approach. The initial part of the release can be relatively quick, followed by a much slower tail.

That behaviour is difficult to describe with one release-time number.

Program-dependent does not mean the same thing in both designs

Both vari-mu and optical compressors are often described as program-dependent, but for different reasons.

In a vari-mu compressor, part of that behaviour comes from the nonlinear transfer characteristic of the variable-gain tube and from the sidechain circuit.

In an optical compressor, much of it can come directly from the physical response of the optical element.

Take a vocal phrase as an example.

A short loud consonant may cause relatively little reduction. A sustained vowel keeps the detector active longer. When the singer becomes quieter again, the photocell may still be recovering from the previous part of the phrase.

Instead of treating every peak independently, the compressor carries a certain amount of recent history with it.

That is one reason optical compressors work so well as levelers.

Attack time: topology alone does not tell you the answer

It is common to read that vari-mu compressors are slow and optical compressors are faster, or the other way around.

Neither is a useful rule.

Attack time depends on the actual design.

A traditional vari-mu compressor may deliberately use attack times in the tens of milliseconds, allowing the initial transient through before reducing the body of the signal. But other variable-mu circuits can operate faster.

Optical compressors vary just as much. The light source might be an electroluminescent panel, LED or another device, while different photoresistors can have very different response times.

The classic LA-2A, for example, has an attack around 10 ms. A modern opto compressor with a different cell can behave quite differently.

When comparing compressors, the actual attack behaviour is more useful than assumptions based on the topology.

Release is where optical compressors become particularly interesting

The recovery characteristic of a photocell is one of the defining parts of many optical compressors.

In an LA-2A-type circuit, release happens in stages. After gain reduction, part of the recovery occurs fairly quickly, while the remainder takes considerably longer. Heavy or sustained compression can lengthen that tail.

Imagine a bass line with long notes.

The compressor reduces the louder part of a note, then begins recovering as it decays. If the next note arrives before the cell has completely recovered, the compressor starts from a partially reduced state.

The result is different from a compressor that simply follows a fixed 500 ms release curve after every event.

Whether that behaviour is desirable depends on the source, but on vocals and bass it can be very useful.

Vari-mu compression and transients

Traditional vari-mu compressors are often used with slower attacks.

With the attack set this way, a kick, snare or other transient can pass through before the gain reduction reaches its full amount. The compressor then acts more strongly on the body of the sound.

This can produce a useful combination: the peak remains defined while the average level is controlled.

On a stereo mix the same behaviour can reduce overall movement without flattening every transient.

It is one of the reasons vari-mu designs became associated with bus and mastering compression.

But the result depends heavily on the attack setting. Set the compressor faster and drive it harder, and the behaviour changes considerably.

Where does the colour come from?

This is another area where the usual descriptions can be misleading.

A vari-mu compressor does use a tube as its gain-control element, so the compression process itself can contribute nonlinear distortion. Many classic designs also use additional tube stages and transformers.

But a properly balanced vari-mu stage can be surprisingly clean.

It is therefore too simplistic to say that vari-mu compression automatically produces large amounts of "warm tube harmonics".

The same applies to optical compressors in the opposite direction.

The optical attenuator itself can introduce very little distortion, but the complete compressor may contain tube gain stages, input and output transformers and other circuitry with a very obvious sonic contribution.

The LA-2A is a good example. It is an optical compressor, but it also has a tube signal path and transformers. Calling it a transparent compressor simply because the gain-control element is optical misses a large part of the circuit.

The Tube-Tech CL 1B makes the distinction even clearer: its gain reduction is optical, while the following amplifier is tube based.

The topology of the compressor tells you how gain reduction is controlled. It does not describe the entire audio path.

Ratio and knee

Vari-mu compressors naturally lend themselves to a progressive transfer curve.

Near the onset of compression, the ratio can be relatively gentle. With increasing level and control voltage, gain reduction becomes stronger.

This is the origin of the characteristic soft transition associated with many vari-mu designs.

Optical compression does not require one particular ratio.

The LA-2A has its own program-dependent compression and limiting behaviour, while a compressor such as the CL 1B provides an adjustable ratio from 2:1 to 10:1 despite using an optical gain-reduction element.

So statements such as "opto compressors have a fixed gentle ratio" are only true for particular circuits, not optical compression in general.

On vocals

This is one of the strongest applications for optical compression.

A voice contains fast consonants, sustained vowels, pauses and large changes in average level. A photocell with a nonlinear recovery characteristic tends to follow the longer-term shape of the phrase rather than responding identically to every individual peak.

With moderate gain reduction, the effect can feel more like automatic level control than obvious compression.

That does not mean that every vocal should go through an opto compressor. A very dynamic singer may need faster peak control first. In that case a faster compressor can catch the peaks and an optical stage can handle the slower level changes.

For straightforward vocal tracking, though, opto remains a very practical choice.

On bass

Bass is another natural application.

The relatively long duration of bass notes gives the optical element time to react and settle. The release then follows the decay of the note rather than immediately returning to zero gain reduction.

This can make the level more stable while leaving the initial attack of the note intact.

Sidechain filtering becomes useful here. Very low frequencies contain a great deal of energy and can dominate the detector even when they do not sound excessively loud. A high-pass filter in the sidechain reduces this effect without removing bass from the audio itself.

On drums

Neither topology has exclusive rights to drums.

A vari-mu compressor with a slower attack can preserve the front of the kick and snare while increasing density behind the transient.

An opto compressor with a suitable cell can also work well on room microphones, drum buses or parallel paths, particularly when the goal is movement rather than strict peak control.

For close-mic peak limiting, however, a fast FET or VCA design is often easier to control.

This is a useful reminder that compressor topology should follow the job rather than the other way around.

On the mix bus

Vari-mu compression is particularly well suited to small amounts of stereo bus gain reduction.

One or two decibels can be enough.

At these levels, the compressor changes the relationship between peaks and the body of the mix without obviously pulling individual elements forward and backward.

The transformers and line amplifiers also remain in the signal path, so the audible result is not necessarily caused by gain reduction alone.

It is quite common for engineers to prefer the sound of a vari-mu compressor with the meter barely moving.

Optical compression can also work on a bus, but its memory effect needs to suit the material. A slow photocell recovering from a loud bass event may continue affecting everything else in the mix.

A sidechain high-pass filter can make a substantial difference.

The same meter reading can sound very different

Five decibels of gain reduction is not a complete description of what a compressor is doing.

Consider two compressors both showing 5 dB.

One reaches that reduction slowly, leaves the transient almost untouched and begins recovering immediately.

The other reaches the same 5 dB but remains partially compressed for another second because the optical cell has not fully recovered.

The meters may show similar maximum reduction. The envelope of the audio will not be similar at all.

When comparing analogue compressors, listening to the shape of the gain reduction is generally more useful than comparing the number of decibels.

Vari-mu or optical?

For stereo buses, stems and mastering work, a vari-mu compressor is often a sensible starting point when the goal is moderate gain reduction while retaining transient movement.

For vocals, bass and other sources where the average level needs controlling without constantly reacting to every short peak, an optical compressor is often easier to use.

There is plenty of overlap.

A vari-mu can work beautifully on a vocal. An optical compressor can work very well on a stereo bus. The circuit, settings and source matter more than the category written on the front panel.

The useful distinction is in the way each design produces gain reduction.

A vari-mu circuit changes the gain of an active tube stage.

An optical compressor uses the sidechain to control a light-sensitive attenuator.

Everything else follows from the implementation.

Two examples from RSE Audio

We use both approaches in our own compressors because they solve different problems.

The RSE VMU-1 is a stereo vari-mu compressor built around a fully balanced tube circuit. It uses six dual-triode tubes and six transformers, with selectable attack and release times and a switchable sidechain high-pass filter. It was designed primarily for stereo bus, stem and mastering work where small amounts of gain reduction and repeatable settings are important.

The RSE RA-2A takes the optical route. Its audio path is tube based, but gain reduction is controlled by a custom optical cell. The cell has a two-stage recovery: a relatively quick initial release followed by a slower tail. Compressor and Limit modes change the amount and character of gain reduction, while a sidechain high-pass filter reduces the influence of low-frequency energy on the detector.

The fact that both units contain tubes does not make their compression mechanisms similar.

In the VMU-1, the tube is part of the variable-gain mechanism.

In the RA-2A, the optical cell controls attenuation while the tubes perform the amplification and other functions in the signal path.

That distinction is a useful way to understand vari-mu and optical compressors in general.