Plugin Alliance SPL IRON Mastering Compressor Plug-in Analysis


The SPL IRON Mastering Compressor is not a copy of a classic unit, but an original concept in itself. SPL’s goal was to conceive a compressor that provided smooth, musical and transparent compression inspired by the sonic and technical operations of Fairchild, Collins and Gates compressors, which used remote cut-off tube biasing to achieve a well-balanced, well-compensated and musical compression. But equally, to be versatile enough to adapt perfectly to the requirements of modern mastering studios. Consequently, the IRON combines not only the sonic virtues of these legendary tube-based compressors, with the advantages of the high dynamic 120V operating voltage; It also sets a new benchmark in terms of tube compressor technology, with the innovative implementation of a parallel dual-tube circuit.

The supply voltage is key for the overall dynamic response of a processor. The higher the operating level, the higher the maximum level a circuit can handle. As the majority of essential acoustic and musical equipment depend on this relation; a higher operating voltage also has a positive impact on dynamic range, distortion levels, and signal-to-noise-ratio.

This behavior is demonstrated in the diagram below:

The input signal is “monitored” by a resistive opto-isolator, feed-forward peak limiter, in the control path of the variable-bias tube circuit, and as a feedback compressor in the audio path. When input signal levels exceed +15dB, its function is to limit signal peaks; and therefore, reducing THD within the variable-bias tube section. The result is a silkier, more homogeneous sound in the higher frequencies.

The signal of each stereo channel is then split across two different twin-triode tubes, by a custom-made Lundahl transformer. The second tube is connected in parallel to the remote cut-off tube, it has a considerably steeper characteristic curve. The combination of the response curves from both tubes results in a more well-balanced and musical sound.

The overall response of the compressor is heavily influenced by the position of Input. This behaviour is partly due to how hard the tubes are being driven; but also, because there is a resistive opto-isolator, feed-forward peak limiter (in the control path) at the input of the circuit. Consequently, a lower input signal will either not trigger the opto-isolator cell, or the circuit will have less impact on the audio signal before hitting the tubes. This approach produces a more consistent transient response. Conversely, increasing the input will drive the opto-isolator cell harder, rounding off transients before hitting the compression stage: producing a smoother, leveling style of compression.
To contextualize, a transient rich input signal will benefit from a lower Input setting; this approach will provide better transient control. Conversely, on a “well-balanced” input signal, Input could remain at unity or even increased, if more saturation is desired.

This behavior is demonstrated in the images below:

The behavior has a “similar” effect to modifying the knee in a digital compressor such as Pro-C2. The -4dB Input is akin to a harder knee, which has better transient reproduction. The 0dB Input is akin to a softer knee, which is smoother and more suitable for transparent compression.

To produce the bias voltage that controls the parallel-connected tubes, audio signals must be rectified into DC. A six-position switch provides a choice of the rectifier diode and smoothing capacitor value. The unique characteristic curve of each element (germanium, silicon, and LED) has a direct impact on attack and release time constants. When auditioning the different Rectifier positions, SPL recommend setting Attack to position “2” and Release to position “1”. With these settings, the compression action is mostly dependent on the Rectifier’s time constants.

Position 1 - GE 1mF

Position 2 - GE 2mF

Position 3 - LED 3.3 mF

Position 4 - Si 330 nF

Position 5 - Ge 220 nF

Position 6 - Ge/Si 100nF

In practice, the first three positions produce a tight and clear transient response, with GE 1mF (position 1) having the fastest time constants, and LED 3.3 mF (position 3) having the slowest. GE 2mF (position 2) and LED 3.3 mF (position 3) were reminiscent of Manley Vari-Mu at certain settings. Si 330 nF (position 4) has a response reminiscent of Fairchild 670, I particularly like this setting on the drum bus, but I could also see it being very useful on the master bus for transient control. I have not yet utilized Ge 220 nF (position 5) or Ge/Si 100nF (position 6) in practical applications but could imagine them being well-suited on individual instruments that are particularly transient rich.

The attack and release times are not constant values; they heavily depend on the audio material, and the position of the other main parameters (Input, Rectifier, Threshold, Tube Bias, Side-chain EQ).

Therefore, the values listed below should be considered approximate rather than absolute values:


The bias of a tube is the voltage present on the tube’s grid. The higher the voltage; less signal makes it from the cathode to the anode of the tube. Therefore, the Bias setting arbitrates the final degree of compression. With the rest of the controls kept the same, higher bias equates to more gain-reduction and saturation.

The below images demonstrate this behavior:

The Sidechain EQs can make the response of the compression be influenced by a specific frequency range. For example, if the low frequencies are attenuated, the compressor will not react as promptly to the kick and bass. This can prove beneficial when these elements are prominent within the source material. Conversely. If you boost specific frequency ranges, the compressor will react more strongly to them.
In my opinion, the included side-chain EQ curves, are neither typical nor intuitive. The response plots presented in the user manual are overlaid and difficult to interpret.
There are five sidechain EQ positions: Off (actually a 20Hz HPF is in the circuit all the time), EQ1, EQ2, EQ3, EQ4, and Ext (external input).

EQ 1: This setting is relatively flat in the low frequencies, it has an approximate 3dB boost at 750Hz and a considerable dip (15dB or so) around 2.2KHz that flattens out around 5KHz. In practice, this side-chain EQ setting could be useful for a “boxy” sounding master, or on a drum bus; as the 750Hz boost will initiate more compression on the over-exaggerated midrange, the dip at 2.2KHz would leave the high midrange sounding “open and “transparent”.
EQ 2: has an approximate 5dB shelf boost at 400Hz, and a 7dB cut at 520Hz. In practice, I have not yet found a suitable application for this side-chain EQ curve. Any suggestions are welcome.
EQ 3: has an approximate 5db sub frequency boost, and a 3dB dip centered around 175Hz. In practice, this side-chain EQ setting could be useful for a stereo master that has an unbalanced low-frequency response, below say 60-80Hz but a balanced LF response above this range; the dip centered around 175Hz will help keep the LF “defined” and “punchy” while controlling the “weight” in the subs. It is likely that a threshold adjustment will be required after engaging this setting, as it will cause a little more compression overall.
EQ 4: has an approximate 5dB shelf boost in the low frequencies that flattens out around 400Hz. With a 4dB resonant shelf boost at 4KHz (3dB cut at 3KHz). In practice, this side-chain EQ setting is great for vocals, the 4KHz shelf boost makes IRON compress more whenever the vocals get “shrill” or “reedy” sounding, and the 3KHz dip will help keep the vocal “open” and “transparent”.

IRON provides two tone-shaping EQ curves called “AirBass” and “Tape Roll-Off”. This passive EQ is placed after a Lundahl output transformer and before SPL’s 120-volt DC Audio Rail output op-amps.

AirBass: is a smooth “smiley” curve with a gentle 2dB shelf boost beginning at approximately 2kHz, and a 2.5dB shelf boost starting at around 500Hz. This EQ curve can make the input sound “rounder”, with a powerful low end and a “bright” and “silky” high end.
Tape Roll-Off: is a subtle roll-off EQ curve with a 12dB/oct low pass filter at 15KHz and an approximate 6dB/oct filter at 20Hz. Tape roll-off is well suited for input signals that are overly “harsh” or “shrill”, it produces a nice “rounding” of the high and low end.*

TMT is Brainworx’s patent-pending Tolerance Modelling Technology. It takes the inherent tolerances of components found within audio circuits into consideration. The result is digital audio that sounds analog, where each channel of a stereo instance will react slightly differently, due to variances in frequency response, time constants, etc.
The “Stereo Mode” toggles between using the same TMT channel for both units (Digital) and using two adjacent TMT channels (Analog).

This behavior is demonstrated in the image below:

In practice, this technology can help make your audio sound a little more 3D, sometimes analog emulations can sound “closed” compared to their analog counterparts. A little trick I have done for years is to utilize dual-mono versions of plug-ins when I like the tone of the processing but think it sounds too closed. Depending on the non-linear behavior of the plug-in, simply using the dual-mono version is enough, other times a very subtle adjustment to one channel will be the ticket. Remember, do not go overboard on channel differences, as the variations in analog channels were typically minute.

Headroom adjusts the internal operating level of the Plugin. Rotating the control clockwise allows input signals to be pushed harder before being compressed; this results in less compression overall. Conversely, by rotating the control counter-clockwise, headroom is decreased, resulting in a greater amount of gain-reduction, color, and saturation.

Mono Maker collapses the audio signal to mono, the parameter range is from 20Hz to 20KHz. A typical setting is between 200-250Hz, below which bass frequencies reside. This should produce a tighter and more defined low-frequency response.

I have manufactured a “wide” low-frequency signal so that you can visualize Mono Maker at work:
Mono Maker - Off

Mono Maker - On

Stereo Width is a mono-compatible M/S widener. Rotating the parameter clockwise boosts the volume of the “side” signal (it is not a crossfade between “M” and “S”, so the “mid” signal is not attenuated). Conversely, rotating counter-clockwise will attenuate the “side” until only the “mid” signal remains.

The below images demonstrate Stereo Width in action:
Stereo Width - 100%

Stereo Width - 200%

In practice, I have found a setting between 105 and 125% adds a subtle but pleasing amount of width on stereo masters. Remember not to overdo this feature, because additional width will often, at first, sound pleasing to the ear. However, with more critical listening, you can easily make your song sound disjointed and unfocused.

Unfortunately, the IRON plug-in aliases, which is surprising in a mastering-grade plug-in. In addition, it also appears that IRON is truncating, which is very strange in any plug-in. I have emailed support about these issues and had no response. Sadly, this is the second time I have reported bugs to Plugin Alliance and been ignored, the same happened with Metric A/B.

The below images demonstrate IRON’s aliasing behavior:
18dB sine wave at 12KHx - 44.1KHz

18dB sine wave at 12KHx - 96KHz

To be fair, aliasing is actually not too bad, and since only using one instance of IRON, it will not have an accumulative effect. Nevertheless, it still shouldn’t happen on a mastering compressor!

Aliasing is not typically heard at low levels, but its accumulative effect can be audible on higher frequencies; adversely impacting “clarity” and transient reproduction.

Truncation - 44.1KHz

There is a considerable amount of truncation, I am really shocked by this.

Truncation adds low-level noise and distortion. With modern DAWs processing audio at 32-Bit float or 64-Bit double-precision; important information such as ambiance, space, and depth would be lost if the lower bits were simply truncated.

The IRON Mastering Compressor is very versatile, with a smooth and transparent sound. It has a tight and clear transient reproduction and controls low frequencies with dimension and impact. It is a staple on the master bus but also sounds great on vocals, drum or bass busses. I nearly stopped using IRON due to aliasing and truncation. Truncation is an outrageous bug to have in any plug-in and should be a high-priority bug fix. Aliasing in a mastering-grade plug-in is also disappointing, and an option for over-sampling should be implemented. However, despite its flaws, IRON does sound very good and I still recommend it to peers.

Since the SPL IRON has truncating and aliasing, would an analog mastering compressor plugin like the Slate VBC or the UAD Manley VariMu perform better without the truncating and aliasing?

I would try them out on a few masters and see for yourself. For uptempo dance music, I mostly always prefer SPL IRON over Manley Vari-Mu.
Truncation is >110dB down, in loud music, like dance, everything is often squeezed into the top 2 Bits, so very unlikely to be audible. If audible, it would be very low level distortion, similar to an undithered 16-Bit file but 15dB or so quieter, below the noise floor of many reproduction systems.

With the “low level distortion” (if the SPL IRON plugin audibles), I can just “CUT” any low level distortion anyhow right?

No, it will always be a part of the audio, dithering after the fact will not work.

Dithering pre IRON would replace quantisation errors with noise but that just feels wrong…

That makes sense.

The reason why I’m so interested in the SPL IRON discussion is because of the special sale that Plugin Alliance is offering for $29, $39, & & $49 (for most plugins) and I’m debating if I should add this particular plugin with my Slate & UAD plug-ins for audio mastering purposes.

1 Like

I would say yes, it is a fantastic sounding compressor. Arguably the aliasing and truncation is more of a technical issue rather than a sound one.

The SPL IRON now has a special place in my audio mastering toolbox. Thanks, for the insight.