The SD-3A is a VMR module that recreates the Teletronix LA-3A leveling amplifier, a feedback optical compressor developed by Urei in the late 1960s.

We used the UA re-issue as a hardware reference.

The LA-3A is simple to use with only 2 knobs: a peak reduction knob  adjusting the amount of compression and one make up gain knob. There are two compression modes that can be selected on the back panel (limit and compress) as well as high frequency(hf) contour screw to adjust how the compressor responds to high frequency content.

Like its predecessor, the LA-2A, the LA-3A uses a T4 electro-optical cell to control gain reduction, offering a smooth and natural compression sound. While sharing components with the famous UA 1176 level amplifier developed a few years earlier, it’s the LA-3A’s use of solid-state amplifiers instead of tube based amps which contributes to its punchier and more transparent sound.

While the LA-2A is famous for its use on vocals, the LA-3A is reactive enough to be used on any type of audio content, including drums.

Modeling

LA-3A Block Diagram (from the original instruction manual)

SD-3A block diagram

Input/Output

Although designed to be as transparent as possible in the audible range, the LA-3A is made of electrical components that are not perfect and have an influence on the sound. We can measure these imperfections by turning the Peak Reduction knob to its minimum value, ensuring that the compression does not disturb the pass-through measurement.

Transformers

Transformers are used to transfer energy from one electrical circuit to another and step voltages up or down, match impedance, eliminate ground loops, and block DC signals. That’s why they are widely used in audio equipment.

We won’t dive into the complex world of transformer modeling here, but we can simply say that they consist of two long wires wrapped around a magnetic core. This magnetic component leads to a characteristic low frequency saturation as shown below.

Equalization and harmonic distortion

The subtle behavior of all the other analog components can be modeled using standard equalization and non-linear functions. The LA-3A has a clean, flat response with subtle harmonics. At low frequency we can see the effect of the transformer saturation.

Secondary signal path

To model the optical attenuation of the original unit, we derive a secondary signal path. This part of the algorithm has the job to accurately compute the amount of compression to apply to the audio to obtain the compressed output. We call this signal Gain Reduction or GR.

As in, we use a feedback topology where the input to the secondary path is taken after the gain has been applied.

Static Compression Curve

The LA-3A has a fixed-threshold, fixed-ratio, and soft-knee. The user can modify the gain of the signal entering the secondary path with the Peak Reduction knob to trigger more or less compression.

Frequency dependence

The amount of gain reduction of the LA-3A is frequency dependent.

Even with the HF Contour screw in the flat position: the hardware compresses the high frequencies less than the low frequencies in a level dependent manner and SD-3A will recreate this accurately.

Turning the HF Contour control lets you tune the amount of compression around 5kHz.

Timings

At the heart of the leveling amplifier is the electro-optical attenuator that consists of photo-conductive cells optically coupled to an electro-luminescent light source. The intensity of the light is proportional to the input audio voltage. This video by analogvibes is a good introduction to the inner workings of such cell: What the heck is a T4 cell? Teletronix LA2A – BYL-Series I, Part III – analogvibes

We carefully identified and modelled all the specific behaviors of the compression timings which are key in the character of the compressor. Since these timings depend on many factors, we measured them across a wide range of conditions to ensure accuracy.

1. Attack and release in compress mode PR=5

• On this example, you can see the plugin takes about 60ms to attack to 90% of the final gain reduction target.
• It takes about 150 ms to release 50% of the target but will take about 10 seconds to go back to a fully released state.

Listen In

Now for the fun part, let’s dig into how it actually sounds. Make sure you’re listening with some monitors or headphones to get the full experience.

Drums

Here’s an example on a drum buss, with around 5 to 7 dB of Gain Reduction using the Limit mode to get that powerful drum sound:

Dry

With SD-3A

Guitar

Here’s another example on an acoustic guitar, with the HF Contour engaged, compressing more the HF and allowing us to control the strong picking sound while maintaining the original tone:

Dry

With SD-3A