1176 story

The UREI 1176 Limiting Amplifier:

Sound quality and operational considerations of the

original and its digital emulations

Christoph Schaufler,

Student Number: 10468961

MA Audio Technology,

Thames Valley University,

London,

United Kingdom

Tutor: Tim Walker,

MA Audio Technology Project,

Module Code: MUM0610E

Wordcount: 10327

August 31, 2006

Contents

Synopsis iii

1 Analogue and Digital UREI 1176LN s 11.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2.1 The analogue original . . . . . . . . . . . . . . . . . . 21.2.2 Hardware clones . . . . . . . . . . . . . . . . . . . . . 5

1.3 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81.3.1 Bomb Factory / Digidesign . . . . . . . . . . . . . . . 81.3.2 Universal Audio . . . . . . . . . . . . . . . . . . . . . 101.3.3 TC Electronics . . . . . . . . . . . . . . . . . . . . . . 121.3.4 Focusrite . . . . . . . . . . . . . . . . . . . . . . . . . 121.3.5 Creamware . . . . . . . . . . . . . . . . . . . . . . . . 16

2 Mixing Techniques 172.1 Upward compression . . . . . . . . . . . . . . . . . . . . . . . 182.2 Michael Brauer’s Multi-Bus principle . . . . . . . . . . . . . . 18

3 Comparisons 203.1 Hard- and Software . . . . . . . . . . . . . . . . . . . . . . . . 20

3.1.1 Preparations . . . . . . . . . . . . . . . . . . . . . . . 203.1.2 Evaluation of the UREI and its clones . . . . . . . . . 253.1.3 Universal Audio UAD 1176LN . . . . . . . . . . . . . 67

3.2 Mixing Techniques . . . . . . . . . . . . . . . . . . . . . . . . 813.2.1 Preparations . . . . . . . . . . . . . . . . . . . . . . . 81

4 Summary and Opinions 854.1 UREI 1176LN Comparison . . . . . . . . . . . . . . . . . . . 85

4.1.1 Subjective thoughts . . . . . . . . . . . . . . . . . . . 854.1.2 Listening test . . . . . . . . . . . . . . . . . . . . . . . 874.1.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 884.1.4 Further investigations . . . . . . . . . . . . . . . . . . 89

4.2 Mixing Techniques . . . . . . . . . . . . . . . . . . . . . . . . 934.2.1 Subjective thoughts . . . . . . . . . . . . . . . . . . . 93

i

CONTENTS ii

4.2.2 Feedback . . . . . . . . . . . . . . . . . . . . . . . . . 944.2.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . 94

A Session Diary iA.1 10. Aug, 06 - ProTools HD3 Mix . . . . . . . . . . . . . . . . iA.2 13. Aug, 06 - Audient ASP8024-36 Mix . . . . . . . . . . . . iiiA.3 20. Aug, 06 - Otari Elite+ Mix . . . . . . . . . . . . . . . . . v

B Supplied data mediums viiiB.1 Data DVD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viii

C Research Folder ix

Synopsis

This thesis came into being as the written part for the final project of amaster’s degree in Audio Technology. It gives a detailed background onthe history and function of an audio compressor, the UREI 1176LN andcompares this classic unit to various hard- and software clones. The testexperiments include audio signal processing, like this equipment would beused in a mix, as well as more scientific methods, like measurements withthe help of test signals.

A second part gives an insight in advanced mixing techniques usingupwards- and multi-bus compression. A combination of those have beenused to improve the mix of a modern pop song, using different analoguemixing desks and digital audio workstations to see how well those methodstranslate on different systems.

The results are more than surprising, showing clearly how close digital sig-nal processing has come to its analogue counterpart. When it comes toemulating a classic hardware unit, with its own distinct sonic characteris-tics, some of the digital successors do an extremely convincing job, even atreally extreme, saturated and distorted settings. This clearly shows whatsophisticated DSP systems are capable of nowadays, as all of the really closeemulations are only available on dedicated processing cards or systems.

The project also shows besides this main part, how powerful digital audioworkstations are at this present time, since the performance, even withcomplex routings as used for the second multi-bus part is excellent. Weare clearly in an age now, where high quality mixes do not necessarily needcostly analogue systems, but capable engineers that are aware of the qualityand characteristics of digital systems and know how to take advantage ofthem.

iii

Chapter 1

Analogue and Digital UREI1176LN s

1.1 Overview

The first section of the project will focus on background information andhistory, plus the standalone function and technical specifications of the vin-tage test hardware. All available hard- and software clones are presented aswell, before the actual comparison experiment is executed.

For a comparison between hard and software processors the UREI 1176LNseemed like the best choice for several reasons:

• The analogue version is one of the absolute classic professional audioproducts ever made and still heavily used in every world-class studiothere is.

• Exactly for that reason, its design has been implemented in modernhardware units, some of them even 1:1 clones. The same counts fordigital systems, the 1176 was one of the first hardware classics, thatwas modelled for the use in a DAW 1; several clones of other softwaredevelopers followed.

• Two of the actual hardware classics are fortunately available for testingand all software- and one hardware clone could be sourced for a propercomparison.

1D igital Audio W orkstation

1

CHAPTER 1. ANALOGUE AND DIGITAL UREI 1176LNS 2

InputInput

Transformer

Gain-Reduction

Circuit

PreAmplifier

Gain-Red.ControlCircuit

OutputAmplifier

Output

Figure 1.1: The block diagram of the 1176 as shown in Universal Audio’smanual for their DSP card

1.2 Hardware

1.2.1 The analogue original

[7]2 The first design, of what became an industry standard compressor, orpeak limiting amplifier as the unit is called originally, saw the light of dayin 1966 engineered by Bill Putnam. The heart of the 1176 is the (which hadonly been recently invented) F.E.T or Field-Effect-Transistor, which endedup being used as the bottom leg in a voltage divider circuit. The nature of aF.E.T transistor requires a careful design, since the linear sweet-spot needsto be found, in order to minimise distortion. Followed by a model 1108preamplifier (the solid state successor to the tube-based 108 amplifier) andas a final stage, a Class-A output transformer, specially designed to matchthe (back then) standard load of 600Ω. For the compressors main design,Putnam decided to go for a feedback operation characteristic, therefore theoutput is used as a reference for the gain-reduction system, as opposed tothe input signal, which is used on feed-forward types.

Controls

On the left side of the unit are the two main input and output knobs, followedby the two smaller attack and release dials stacked vertically. The famousfour-button design for changing the ratio is found on the left side of the VU-meter, whereas on the right side the same button system is used to switchbetween the meter-view modes (Gain Reduction, -18dB / -24dB) and a finalOff knob to power down the unit.

Technical specifications

The compressor-relevant specs, as found in the original [9]3 UREI manualfor the later revision H models. Since the actual detection and compression

2http://www.nrgrecording.de/html/history.html3Operating Instructions for model 1176LN Peak Limiter, sn# 7652 and above


circuit pretty much remained the same over the years, those values applyfor other revisions as well.

• Attack Time: from less than 20µs to 800µs (for 100% recovery)

• Release Time: 50ms to 1.1s (for 63% recovery)

• Ratios (with according threshold in brackets): 4:1 (-30dB); 8:1 (-26dB);12:1 (-25dB); 20:1 (-24dB)

Function

The straightforward concept is very easy to understand and implements acommon way of increasing the amount of compression, simply by rising theinput level. There’s no dedicated threshold control like found on many othercompressor units, but each different ratio works with a default thresholdvalue. Again, unlike most other limiting amplifiers, both attack and releasedials work in the opposite way – therefore turning them fully clockwiseshortens- and counter-clockwise increases the time constants. Another lessknown feature of the unit is a bypass mode, that is activated simply byturning the attack knob fully counter-clockwise until the pot ”clicks”. Someengineers use the unit without gain reduction, just to color the source-signalwith the UREI ’s circuitry. A more known trick is to push all four-ratiobuttons in at once (often called British Mode, four-button trick, or nukemode). The ratio is then set somewhere in between 12:1 and 20:1 andthe bias points change all over the circuit. The initial transients are passedthrough even on fast attack setting, thus resulting in a punchy attack, that isinstantly followed by a sucking of volume, both time slopes react then reallyaggressive and snappy and result in crunchy distorted sounds, especiallyin the lower / mid frequency range, a characteristic often used on drumambience / overhead channels.

Revisions

[6]4 Several different improvements / changes have been made over the yearsof production to the original concept, each can identified via serial numberand (depending on the revision) by optical appearance.

• Rev A (Jun, 1967; #101-125): Silver brushed aluminium front with ablue outline

• Rev AB (Nov, 1967; #125-216): Slight circuit changes to improvenoise and stability.

4http://mixonline.com/mag/audio revision history/index.html


Figure 1.2: The look of the early 1176 unit is clearly identifiable by the blueVU outline

Figure 1.3: One of the later, more sought-after low-noise or LN models, thefrontpanel color was changed to black

• Rev B (Nov, 1967 to Jan, 1970; #217-1078): A bipolar design replacesthe FET preamp section.

• Rev C (Jan, 1970; #1079-1238): First introduction of the blackfacefront and the LN suffix, which points out the newly introduced low-noise operation, that is based on an additional circuit board imple-mented by Brad Plunkett.

• Rev D (into 1973; #1239-2331): No major changes, the low-noisecircuit was incorporated in the main PCB.

• Rev E (up to March, 1973; #2332-2611): Power transformer for 110Vand 230V operation was added.

• Rev F (March 1973; #2612-7052): Class A output transformer waschanged to a push-pull principle for more output drive. Also the me-tering circuit has been reworked to simplify calibration and testing.


Figure 1.4: A rare 1178LN stereo unit

• Rev G (unknown date; #7053-7651): Replace the input transformerwith a different, low-cost op-amp amplifier.

• Rev H (unknown date; #7652+): Front panel was changed back tosilver.

Stereo operation

Besides the more common mono version (that has a complicated stereo-linking connection at the back, which can only be calibrated with an exter-nal, battery powered stereo adapter) there has also been the 1178 stereoversion with separate input / output control (stereo-linkable) but a sharedattack / release / ratio control. It was introduced with the latest modelH (hence the silverface), and incorporates the same circuit changes thathave also been done on the mono version, which has generally not the highreputation as the blackfaced model D and E.

Test models

Both units available for testing are blackfaced revision F models with serialnumbers 6168 and 5277. Those versions are one of the later builds and differin the missing output transformer, thus changing the sonic characteristics,which puts another variable into the comparison, since most clones are basedon revision D & E models. Nonetheless the detection circuit remained thesame, with the original crunchy sound when pushed fairly hard.

1.2.2 Hardware clones

Purple Audio MC77

This is one of the early recreations with a very close copy of the revision Ecircuitry, the former MC76 model even had exactly the same input attenua-tor than the original. Unfortunately this part of the circuit was discontinuedin 2002 and after ceasing production for a brief period, Purple Audio thenrevised its copy and found a new solution for the input stage to match theoriginal in every way possible.


Figure 1.5: The colourful clone MC77 crafted by PurpleAudio

Figure 1.6: Ted Fletchers Edward the Compressor emulates the classic 1176as well besides several other modes

TFpro P8 Edward the Compressor

This unit has a Urei mode as one out of four mode settings. Although theP8 claims to recreate the original characteristics, it works with Mid-Sideprocessing to ensure the best possible stereo imaging for an optical detectioncircuit and offers advanced controls over stereo width (because of the M/Soperation), transient release (when in Green Box Mode) and different slopemodes (basically a ratio control). A recently released successor is the P38from the same manufacturer, but there’s no information available on whereimprovements where made and how the units differ, except from slightlychanged controls.

Empirical Labs Distressor

Dave Derr’s very modern design is not a real copy of the original UREIdesign, although certain hints in the manual describe settings to emulatethat classic sound. With the recently added British Mode (A 4 button em-ulation), that is available as an upgrade to older- and comes fitted as astandard option on new machines, it tries at least to recreate one particularsound of the vintage classic.


Figure 1.7: A very versatile modern compressor design: Empirical LabsDistressor

Figure 1.8: Universal Audio’s replicas (from top to bottom): 6176, a com-bination of the 610b tube - preamp and an 1176; 1176LN, the top-notchclone; 2-1176, a linkable dual-mono version

Universal Audio

Bill Putnam, responsible for the original 1176 design, started of with a com-pany called Universal Audio, which later became United Recording Electron-ics Industries or simply UREI. [8]5 Bill’s original name was relaunched in1999 by his sons, with the intention to reproduce classic analogue record-ing equipment designed by their father and his colleagues and research anddesign new recording tools in the spirit of vintage analogue technology.

All three available versions, as seen in picture 1.8, are based as closeas possible to the original, with no attempt to simply modernise the muchsought after original design. The unit is assembled using hand-wiring tech-nique and is tested to match the original in every achievable way.

5http://www.uaudio.com/company/history/index.html


Figure 1.9: Part’s supplied with the Purpleaudio DIY-version of their veryown clone

Do-It-Yourself clones

Beside ready-to-use products from esteemed manufacturers, there are severalclone-schematics available on the web. Purple Audio even sells their ready-etched PCB and hard-to-find components as a kit, another popular copy isGyraf’s design, which is based on the later revision F.

1.3 Software

1.3.1 Bomb Factory / Digidesign

[5]6 Bomb Factory started off as, and continues to be a top-class recordingstudio, owned by Erik Gavriluk, the main attraction beeing its collectionof historic instruments and studio equipment. The production of softwareequivalents to analogue processors, using complex mathematical models,began when he met Dave Amels during a restoration session of museumpieces

Digidesign acquired the rights to shipping all Bomb Factory plug-ins in2004.

Purple Audio MC77

Ironically, the plug-in part of the Classic Compressors Bundle, is a digitalreplica of a hardware unit, which claims to be identical to a ’real’ 1176 -therefore its a clone of a clone.

[3]7 As you can see from picture 1.11 the front of the unit is identical and

6http://en.wikipedia.org/wiki/Bomb Factory7Bomb Factory Plug-Ins Guide, p. 14


Figure 1.10: One of the many version of Gyraf’s popular design

Figure 1.11: Bomb Factory’s Purple Audio MC77


Figure 1.12: Bomb Factory’s BF76 replica

so are the controls. Only the time-constants are quite off the original withvery different attack values spanning from 0.4 - 5.7ms and closer releasetimes from 0.06 - 1.1s. The four-button mode can be achieved by double- orcommand clicking on any of the ratio buttons. Bomb Factory claims thatthe ratio stays at 20:1, but the compression knee changes drastically andthe processor starts behaving a bit like an expander.

BF76

There have always been rumours, that this version isn’t different from thePurple Audio clone, but no official statements confirm wether this is trueor not. Except for the color of the GUI, this plug-in looks exactly like theMC77 and according to the manual, there are no differences in controls orspecs as well - this is something which the tests will shed some light on.

1.3.2 Universal Audio

UAudio has a similar background to Bomb Factory with strong roots inhardware processors. As already mentioned in the UREI history was Uni-versal Audio the original company under which name Bill Putnam releasedhis first 1176 designs. Today Bill Putnam Jr. and James Putnam arerunning the company, after a re-launch in 1999.

Besides the reproduction of classic hardware, the company manufacturesthe so-called UAD card. Available for all PCI equipped desktop comput-ers, it incorporates a powerful DSP chip to accomplish the audio processing.That means that the card takes the strain off the main computer CPU, butalso that only plug-ins especially written for that card will run on them. InUAudio’s case the plug-ins are developed and programmed at their com-pany, with a special attention drawn to the porting of analogue classics tocomputer recording systems.

[1]8 Two different versions are available from them as well: The regu-

8UAD Plug-Ins Guide, p. 120


Figure 1.13: UAD DSP processing PCI card

Figure 1.14: Top: Universal Audio Blackface 1176LN ; Bottom: The white1176SE


Figure 1.15: TC Electronics’ PowerCore Firewire card, as a 1U 19” rack-mount device

lar 1176LN and a DSP-friendly 1176SE version, that uses less power fromthe chip, whilst still maintaining sonic characteristics similar to the 1176LN(UAD claims, that the differences only become audible with extreme set-tings, whilst under normal usage, the variation is negligible).

Both attack and release have the same specs as the original, the front-panel layout is equal and the four-button mode is switchable by shift-clickingon any ratio button. In this mode, the UAD manual states that the com-pressor will be working with a ratio of around 12:1, the release is a lot fasterand the shape of the release curve changes as well.

1.3.3 TC Electronics

This Danish company is another well-known manufacturer in the audioindustry. Besides the production of analogue and digital hardware effectprocessors, they offer a similar DSP -based product like Universal Audio,although their concept is available in more variations and open for devel-opment by third-parties. Pictures 1.16, 1.15 and 1.17 show the availablehardware units.

[13]9 The information found from the manufacturer is somewhat thin.The plug-in claims to have a modelled algorithm, but no information is foundabout the time constants, or anything else, but the looks clearly remind theobserver of an 1176. As a unique feature, the plug-in offers an auto-gainfeature, that adjust the output level according to the program material. Ofcourse the all-button-in mode isn’t missing either and is described as a quirkybehaviour, the buttons can be pressed in any combination with a shift-click,where ratios are simply added together.

1.3.4 Focusrite

This British company has long been known for its high quality and oftenesoteric audio designs. They focus on signal processing, like microphonepreamps, compressors and equalisers and has recently entered the world ofdigital processing with versatile solutions like the Liquid Channel and therange of computer audio interfaces called Saffire.

Both, the Liquid- Channel and Mix are based on a dynamic-convolutiontechnology invented by Sintefex. ’Regular’ convolution has been around for

9TC Electronics 24/7c manual


Figure 1.16: TC Electronics’ PowerCore PCI card

Figure 1.17: TC Electronics’ PowerCore Compact firewire device offers lessDSP power in a smaller housing


Figure 1.18: TC Electronics’ 24/7c UREI replica

Figure 1.19: Focusrite’s Liquid Mix hardware plus control unit


Figure 1.20: The software control window of Liquid Mix

several years, mostly used to capture reverberation. The standard techniqueis to exaggerate a real room, or reverb effect unit (or any other static au-dio equipment) with a sharp, short impulse (modern systems use a moreadvanced method utilising a slower sine sweep), where the output of themeasured system is recorded. This impulse response is then loaded into adedicated software like Logic’s Space Designer, where the incoming audiosignal is multiplied sample per sample with the impulse response (or shortIR), to print the sonic characteristic of the sampled room or equipment ontothe audio files that gets processed. This process is static, where the IR rep-resents a single point in time. People have used this technique to copy platereverbs, hardware equalisers, or the frequency spectrum of an analogue tapeonto their digital audio, but a complex procedure like compression couldnever be captured. Compression is a signal dependent process, a methodthat can’t be captured by a single burst.

Sintefex invented the mentioned dynamic convolution exactly for thatpurpose. An impulse response is captured, using a series of sequenced testimpulses of different levels in order to receive the response necessary torecreate compression.

[4]10 Therefore Liquid Mix offers more than ’just’ a digital copy of an1176. No less than 40 compressors and 20 equalisers are available, capturedusing Sintefex’s technique. As stated in the cutsheet, two classic units havebeen used for dynamic convolution, one early blackface (serial #1394 - la-belled STELLAR 1/ US CLASSIC SOLID STATE 1 ) and one later (serial#11854 - labelled STELLAR 2/ US CLASSIC SOLID STATE 2 ) silverfacemodel.

10Focusrite Liquid Mix cutsheet


Figure 1.21: Creamware’s Vinco vintage compressor emulation

Because the convolution technology allows a blending of settings, LiquidMix offers besides the classic functionality a free mode, where the limitationsof the original (like fixed ratios) are overruled and all common compressorparameters can be freely adjusted.

1.3.5 Creamware

[2]11 As the last known software copy, Creamware, similar DSP-based con-cept than the previously introduced systems, sells Vinco (short for vintagecompressor). The optical appearance, as shown in picture 1.21 clearly re-minds the observer of the UREI 1176LN design, although the obvious rolemodel is nowhere mentioned in the manual. There are no specs of attack andrelease times given, but you the front panel reveals some enhancements overthe original design. First of all there is a 2:1 ratio option (but no 4-buttonmode as far as the manual goes), a soft-clip function, the compressor has asidechain input and metering is switchable to a more accurate, digital PPMmode.

11Vinco operational manual

Chapter 2

Mixing Techniques

With every new project a mix engineer starts, he wants to hear an improve-ment over his/her last finished work. Assuming that the equally, if not moreimportant recording part was done well, everything relies on the skills of theperson operating the faders.

Speaking of records in the modern pop/rock genre, one of the main bigchanges over the past few years was, to craft dense, loud mixes, that havebeen made even louder during the final mastering stage. Whether this is agood or bad thing is not under consideration in this project; nonetheless, ithas become an important part of popular music production, where there ismuch competition about the loudness of mixes.

My biggest concern about the way I crafted my ’beginners’ mixes was,whenever a track was finished, with a transparent sound, good depth anda balanced frequency spectrum, I started to get in trouble during the mas-tering stage when it came to increase the overall loudness. Yes, I do usecompression during mixing, most of the time quite a lot, but my averageRMS level is nowhere near a modern record. So the main problem is allabout the final compression / limiting of the finished stereo mix. As soon asthe solid mix gets brought up by dynamic processing, the compression startsto suck up the overall signal, limiting the peaks, but also bringing out thelower signals, like reverb and delay, ruining the depth of the mix. The samecounts for multi-band processing. If this is done excessively, the frequencyspectrum of the mix falls out of order and your mix loses its balanced sound.

These are extreme examples, but this is what happens and even insmaller amounts it is still audible. The ear gets tricked by a louder sig-nal, but the overall quality of the mix is lost. To solve this problem, thefollowing two lesser known and practiced methods will help to craft a densermix with higher RMS values, where the overall sound gets preserved overthe mastering stage. What is loud as a starting material for mastering, can’tbe ruined by excessive compression.

17

CHAPTER 2. MIXING TECHNIQUES 18

2.1 Upward compression

[11]1 Rumours say that what is commonly know as the New York MixingStyle, has is roots back in the days of Motown music, where expressive, dy-namic, soul singers had to be kept under control without ruining a brilliantperformance. What engineers started to do was, to split/send the unpro-cessed vocal track onto a second channel and compress and equalise that in away, so every word had a constant level and could clearly be understood andthan add that ’help’ channel to the original signal. The musicality of thesinger was preserved, but the vocals still sounded up front and had impact,a key quality of a good pop mix.

The successor of that technique was then adapted for other instruments,as in the actual case of the New York Style, especially for the rhythm section.All drum channels are sent to a bus, often together with the bass, whichgets heavily squashed (10dB gain reduction or more) and often a drasticshelving-boost of 6-10dB for the low- (100Hz) and top end (10kHz). As withthe motown original, this, effected signal is added underneath the originalchannels, so it can just be heard. The same result is achieved, with a tight,locked rhythm section, that still has all its original dynamics and transientspreserved.

This method can be applied for every signal that one can imagine. Lock-ing kick and bass together, creating guitar-walls of sound and so forth.

2.2 Michael Brauer’s Multi-Bus principle

[10]2 In an attempt to take upwards compression to the next level, wellknown mix-engineer Michael Brauer is constantly developing and improvinghis unique way of crafting powerful, modern mixes. The very basic idea is tosplit instruments into four subgroups (or busses) plus the stereo main-mix,that can be seen as some sort of multi-band system with no real crossovercircuit for separate frequency processing. Instead he groups instrumentstogether that play in the same frequency range and are usually in a similarmix-space.

His system of choice is an analogue setup, based around an SSL 9000desk, which has four stereo busses and the main mix. The four subgroupsare labelled alphabetically and are in combination with the main stereo busas follows:

• A: Represents the top-end and is processed by a Neve 33609 compres-sor and a couple of Pultec equalisers.

1Bobby Owsinksi, The Mix Engineers Handbook2TapeOp Magazine, Michael Brauer on Compression

CHAPTER 2. MIXING TECHNIQUES 19

• B : Stands for bottom and is sent through a pair of Empirical LabsDistressors into Avalon eq’s.

• C : Is the center of the mix, which is compressed by a Pendulum ES-8.

• D : The last subgroup is a taste bus, used to warm-up sounds with acustom-made Vac Rac limiter

So as an example for a standard pop/rock mix, drums and bass wouldgo to group B, guitars to C, keyboards and backing vocals maybe to acombination of A and D and strings to D on its own. Lead vocals arealways treated on their own. Each single track on the desk is sent 100%through a subgroup, or any combination of them, initially there is no upwardcompression going on. His concept is more a way of having separate mastersections incorporating their own sound, with each section working, maybeeven pumping on its own, without affecting other parts of the mix. If twobusses are used for one signal, like the backing vocals in the example, thedecision, if this sound is needed, is made beforehand, since a routing like thatdoubles the signal over two groups, thus resulting in higher level. The stereobus is then a safe way out of using too much compression, for instance whena snare is not popping out enough and increasing its level will only drive thesubgroup compressor even more; reducing the amount of compression willeffect other signals that are going through and sound good the way they are.So sending the snare to the dry, uncompressed main bus, is the only way tobring its transients back into the mix.

That is the main idea behind his grouping technique. In conjunctionwith this technique, he never patches compressors across channel inserts,instead he always goes for the upward method of having a separate channelwhere a more slammed version of the original signal can be added, whichthen goes to any of the groups or just the main mix.

Chapter 3

Comparisons

3.1 Hard- and Software

Comparing and testing on device against another requires careful planning.A crucial part of comparing hard and software with each other is, to ensureexactly the same audio files are used and that it is possible to align and loopthem properly for convincing blind testing. When it comes to adjusting thedifferent compressor models, it has to be ensured, that settings are matchedas close as possible and this can only be done by actually measuring theattack & release times and amount of gain-reduction, otherwise differencesin sound could be reduced to differences in processing settings. Since nohardware unit behaves in exactly the same way, even with all the dials andknobs in the same position and as stated in the software section, the plug-incopies differ a fair amount from the original in terms of time constants andcontrols.

3.1.1 Preparations

Test sound files

To bring out sonic qualities of a compressor models, several types of audiotest signals have been chosen. After extensive testing and research, thedecision was made to use the 1176LN’s on following three audio examples:

• Drum overhead: Probably one of the most famous applications forthe UREI compressor, because of its distinctive, aggressive characterwhen driven hard, is used on room microphones and drum overheadtracks. The very transient signal can be used to show how fast thecompressor can react on spikes created from kick and snares.

As a test file, an uptempo loop, that starts of with a quieter side-sticksnare + ride cymbal groove and then changes via a fill-in includingcymbals and toms to a full, snare-smacking beat has been chosen.

20

CHAPTER 3. COMPARISONS 21

• Bass Guitar: Almost the exact opposite of a peaking drumloop is theelectric bass guitar: Most of the time it needs evening out of levelvariations between the strings and an increase in sustain, for a solidbass foundation, that doesn’t jump up and down in the mix.

Since bass guitars can be played more sustaining (usually with fingers)or slapped, which produces a signal closer to that of drums, the examplesoundfile includes both playing styles in a 45 seconds loop. This willmake it harder to find a setting that suits both playing styles, butrepresents a real-life problem.

• Vocals: The leading element of nearly all songs in the pop/rock genrealways needs special attention. It has to be up front of the mix, with-out standing out, but with careful attention paid to preserve the artistsexpression, which can be a very dynamic, emotional performance de-pending on the song. So for vocals, you usually want a very transparentcompressor, that can be pushed hard, without creating unwanted sideeffects, like obvious pumping.

Therefore a vocal line, that fits all those criteria and represents a verycommon vocal take was chosen.

Test tones

As already explained, a critical part of the comparison is, to keep the settingsof all test-processors as equal as possible, a method of actually measuringimportant parameters has to be developed.

The initial idea was to use white-noise test signals, to measure attack,release and gain reduction in ProTools or Logic’s sample editor. But afterseveral attempts, the conclusion was drawn that this method is unusablebecause of the more or less random waveform of a white-noise signal, asseen in picture 3.1

Especially when it comes down to really fast time constants, this methodis inappropriate, which led to the following, quite obvious improvement:Regular sine-waves were used instead, as they deliver a much more usableresult. After experimenting with different frequencies, a 3kHz tone has beenchosen, since it offers a short enough wavelength to be useful within a 5msrange, whilst still providing a useful signal for the compressor to react to.Experiments with different frequencies (100, 1k and 10kHz) will show, ifthere are differences in the detection circuits.

Gain reduction can be measured with the same test tone principle anda simple level meter, which is the last important setting that needs to bemeasured.


Figure 3.1: A compressor attack time of 9ms should be verified with thehelp of white-noise burst - as you can see, the ’random’ waveform is notsuitable for this purpose

Figure 3.2: The 3kHz test sine used to determine attack and release timesas well as the amount of gain reduction


Test equipment

Since the whole testing environment should stay as consistent as possiblefor all compressor models, a portable recording system was chosen. Sinceall the test audio files and tones were prepared on a Logic 7 Pro system,all the recording and testing was done with that software as well. The testtones / loops were arranged in two tracks, one for mono- and one for stereooperation. The hardware units and the plug-ins only available in ProToolswere integrated in Logic with the I/O plug-in that can be found in theHelper category.

A MOTU 828mkII firewire audio interface was used for the analogueto digital conversion and the digital integration of ProTools. AKG 271khave been the monitor headphones of choice, a speaker setup was not anoption, since the different hard / software required the constant change ofthe testing-room.

The only exception were made when testing the Universal Audio andTC Electronics DSP-cards, since student colleague Akis Kollaros kindlyprovided both systems in his homestudio. A RME Multiface audio card andSteinberg’s Nuendo was the test system for those software replicas.

Test sequence

To ensure convincing test results, the next step requires careful preparationas well. Following approach seems to be the best option for the main testexperience:

• A day of ensuring that the planned method of measuring both hard-ware and software compressors will be needed. This is the last chanceto change the main concept of the actual experiment

• Both UREI 1176LN of Vestry Control 1 will be measured first, whereunit #1 will act as a reference for all further models.

• The test audio files will be processed on one hardware original first,recording two different files. First the unit will be adjusted, as it wouldbe used in a mix, then a second harder setting, where the compressionis more audible will be dialed in. This will make listening tests a biteasier, although sounds like this won’t be used in an everyday situation.

• Each setting is measured with the sine-tone method, which is neces-sary for adjusting further hard- and software replicas. Each settingwill also be analysed using the ETF software, to check the frequencyresponse and transfer function.

• As soon as these reference files are finished, all available clones willbe tested. First for the minimum and maximum attack and release


Computerrunning

Logic Pro

Motu 828mkII Interface

Firewire

Analog I/O Digital I/O (Adat)

ComputerrunningProTools

Digidesign 192Interface

Analog

DigiWire

Optical TosLink

Computerrunning

ETF

Analog

FocusriteLiquidMixFirewire

Analog

HardwareCompressors

Figure 3.3: The block diagram of the test session. The ETF system waseither connected to a hardware unit, or the Motu audio interface if softwarewas measured


Figure 3.4: Both original test units

times, then those will be matched as close as possible to the referencesettings and finally the audio files will be processed. Again with anETF analysis in between.

Matching of test hard/software The most crucial part of the wholetest procedure was done as follows: The amount of gain reduction and thelevel after the release phase was measured with the freeware RMS 1 meterRMS Buddy (by DestroyFX2). So Input / Output (and other settings likethreshold, compression on not 1:1 clones like the TFPro or Focusrite Liq-uidMix recreations) were adjusted until the same RMS values as the UREI#1 (the reference unit) had produced.

As soon as this was achieved, the attack and release times / slopes werecompared and adjusted to the closest matching possible - the whole proce-dure took about 5-10 tries for each unit.

3.1.2 Evaluation of the UREI and its clones

UREI 1176LN Model #1 - The Reference

As you can see on picture 3.4 the top 1176 has a metering fault, when there’sno gain reduction happening the meter is stuck to full right, so it’s unpre-dictable how much dB are taken off the signal when the unit is compressing.There was no audible error, which was the reason to still use this unit as areference, as all the following test models were approximated with the helpof sine waves anyway.

The pictures 3.5 and following show the units average and extreme set-tings in full effect (make sure to observe the unit in the time-ruler bar).Each of these overview tests were done with a 8:1 ratio and an input setting

1RMS: Root Mean Square = measures continuous AC (Alternating Current)2http://destroyfx.smartelectronix.com/


Figure 3.5: UREI #1 : Attack (fast setting) - The UREI’s ultra-fast perfor-mance is impressive for a vintage circuit board design

that led to maximum compression, whilst the compressor was still releasingto 0db gain reduction when the quieter part of the sine wave kicked in.

When you look at the overall reaction slope in picture 3.7, you noticethar the compressor reacts on the initial transient, but still lowers the signalover a longer time period until full gain reduction is achieved.

Attack times are really hard to measure properly, since 20 microsecondswould require a 50k sine wave, which is an inappropriate frequency for mea-suring audio devices inside their usual operating range. It is a lot easier tospot release times, both screencaptures clearly identify the 50ms - 1.1s range(remember for 63% recovery!).

When you look at image 3.7, you’ll notice that even if the initial attack,that reacts on the first transients, is very fast, the compression is still loweredafter that, but a lot slower and smoother. Because hardware was integratedvia the I/O plug-in, it was more difficult to meter the incoming signal andthe measurement started off with fast time constants, which led to slightclippings on the slower times for the sine waves, but all processed audiotest-loops are un-clipped.

UREI#1 Bass As explained earlier, the compressor was recorded withtwo different settings, as shown in figures 3.10 and 3.11. The more averagesetting for balancing the bass was based on a 4:1 ratio on all models, whichoffers the smoothest compression knee that results in the least audible pro-


Figure 3.6: UREI #1 : Release (fast setting)

Figure 3.7: UREI #1 : Medium settings


Figure 3.8: UREI #1 : Attack (slow setting) - In comparison to the fastsetting the attack time doesn’t seem really slow, but you can see how muchgreater the peak values are.


Figure 3.9: UREI #1 : Release (slow setting)

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.10: Setting UREI #1 : Bass

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.11: Setting UREI #1 : Bass driven


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.12: Setting UREI #1 : Vocal

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.13: Setting UREI#1 : Vocal driven

cessing of the signal. The attack time was carefully adjusted so that thecompressor reacts fast enough, therefore the level decrease isn’t noticeable,but without muffling the attack at the same time, which is a side effect ofan attack time that has been set to quick. Release was then a compromisebetween the slap- and the sustaining part, because for slapping you want afaster time, since the compressor should be already back to zero gain reduc-tion before the next string is plucked, whereas for the second part a longertime is more appropriate, to enhance the sustain of the bass.

The second driven setting was achieved by pushing the input of the1176 a bit more and lowering both time constants. The latter setting ismore audible and really brings out the edge of the bass guitar. Comparedto the dry loop, particularly on the driven setting the crunchy sound of theUREI really shines. The string attack is more accentuated and the low endgrowls, perfect for an assertive bass sound in a complex mix.

UREI#1 Vocal When it came to compress the vocal line, a higher ratioof 8:1 was chosen since the performance spans quite a wide dynamic range.A faster attack time is necessary to level out quick volume rises, for therelease a short setting is a good starting point as well, but it has to beensured that the time is long enough to prevent the sound from unnaturalpumping, so that the volume for trailing of words is not increased again.


On both settings the UREI really shines, even if the vocals are heavilysquashed and have nearly no dynamics left like on the more driven sound-file,the compression is still not obviously noticeable and introduces no unwantedartefacts.

UREI#1 Drums Since the hardware test unit is a mono processor, thestereo processing of the drums is found in the next section, where bothvintage 1176 compressors were used for stereo operation.


Figure 3.14: UREI #2 : Attack (fast setting)

UREI 1176LN Model #2

As figure 3.14 and following show, in terms of attack times both originalunits are nearly identical, the only difference is the second gentle slope,after the initial attack is a bit slower on the second device.

A surprisingly huge difference is noticeable on the release time. UREI#2 is a lot slower and offers a softer release curve than the steep slope foundon test model #1.

UREI#2 Bass The matching of the attack and release time was not aseasy as assumed, since both units differ quite a lot from each other. Ureimodel 1 has a similar slope, but reacts a bit slower on transient signals. Thelonger release, as seen in the picture 3.18 is strange, because the overdrivensound, that is usually associated with short release times, starts to get audi-ble at the same knob position, but which results in a different, much longerrelease time. So the final setting resulted in a longer release time, in orderto keep the obvious difference in tone on a really short setting (which wouldhave matched unit #1 a lot better) subtle, which does not affect the resultsof listening tests.

As a first impression the second hardware unit does sound close, butnot as snappy as number one, which can be clearly identified in figure 3.21,where the yellow overlay displays a more aggressive compression, but that


Figure 3.15: UREI #2 : Release (fast setting)

Figure 3.16: UREI #2 : Medium settings


Figure 3.17: UREI #2 : Attack (slow setting)

Figure 3.18: UREI #2 : Release (slow setting) - The light gray overlaydisplays the much shorter release curve of UREI model #1


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.19: Setting UREI #2 : Bass

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.20: Setting UREI #2 : Bass driven

kicks in with a slight delay, letting more transients pass. Other than that,the compression is less audible when processing with model two, which canbe explained because of the slower release time. Character-wise they arenearly identical, offering a similar growl and crunchy sound.

UREI #2 Vocal When you compare the release graphs 3.7 with 3.16 itis noticable that the slope of model two is convex 3, whereas original unitone has a concave4 form. Screengrab 3.24 shows both slopes in one picture,the decision was made to adjust the second unit so the 63% recovery pointwas nearly identical on both processors.

The input / output controls, to dial in the overall amount of gain re-duction is very close to the reference model one, which was also noticableduring bass processing. When it comes to compare the audio files for vocalprocessing there is virtually no difference.

UREI Drums When it came to compress the stereo-drums, the first bigproblem was encountered - both models could simply not be matched closeenough. Unfortunately, the original Stereo Adapter was not available, so the

3Convex: Based on a lens that is bent outwards4Concave: Based on a lens that is bent inwards


Figure 3.21: UREI #2 : Attack setting for processing bass. The light yellowoverlay displays the waveform of UREI #1

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.22: Setting UREI #2 : Vocal


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.23: Setting UREI #2 : Vocal driven

Figure 3.24: UREI #2 : Release slope of model #2, the gray overlay displaysunit #1.


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

1176 PEAK LIMITER

RATIO METERATTACK

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

LEFT CHANNEL

RIGHT CHANNEL

Figure 3.25: Setting UREI : Drums

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

1176 PEAK LIMITER

RATIO METERATTACK

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

LEFT CHANNEL

RIGHT CHANNEL

Figure 3.26: Setting UREI : Drums driven


Figure 3.27: UREI : Time constants for stereo-drum compression. Left:Attack; Right: Release

goal was to adjust the second unit as close as possible to the first, whichwas a difficult procedure.

Figure 3.27 visualises both time constants, attack is matched very closely,but the release time has the same issues as encountered during bass process-ing. The time is significantly longer, which leads to obvious stereo-imageshift, but if the time would have been reduced to match the first device,audible distortion would have been the result.

Since there is no reference unit, the processing can’t be compared toanything yet, but as a first impression, the performance of the hardwareoriginals working in stereo sounds excellent, especially the famous 4-buttonmode results in a very unique sound, that works well on the drum loop.


Figure 3.28: TFPro P8 : Attack (fast setting)

TFPro P8 Edward the compressor

Even though the TFPro compressor utilises an optical principle, it managesto operate pretty fast for a usually smooth and slow circuit design like that.The release slope is gentler, but in the same speed range than the referenceunit.

Both slower constants show some sort of settling process, especially thesecond bump during the attack phase is conspicuous. The actual slope isnot so different from the fastest possible attack setting, but when the knobis turned to slow, there signal is passed through uncompressed for a longerperiod before the processor kicks in, a similar behavior to the bass-processingof the reference unit.

TFPro Bass Since this unit works with completely different controls com-pared to the original 1176 design, there is no way to transform the usedsettings to the classic UREI controls. The mode dial was of course set to1176 and after a few experiments slope 4 (which basically control the ra-tio between 5:1 and 8:1, but also affects the reaction slope, as the namesuggests) was chosen, because it resulted in the fastest and most 1176-alikebehaviour (fig. 3.33). Compression is a threshold and output control in oneknob and was used in conjunction with input to match the desired amountof gain-reduction, as produced by the reference UREI #1.

Even though the design is fast for an optical concept, the attack control


Figure 3.29: TFPro P8 : Release (fast setting)

Figure 3.30: TFPro P8 : Medium settings


Figure 3.31: TFPro P8 : Attack (slow setting)

Figure 3.32: TFPro P8 : Release (slow setting)


Figure 3.33: TFPro P8 : Aggressive Slope 4 on top of the yellow, more gentleSlope 1

had always been set fully counterclockwise, which resulted in the quickestpossible reaction time.

Soundwise the TFPro is not far off, the sound has the sluggishness ofoptical circuits in terms of reacting on peaks, therefore the attacks are nicelyaccentuated, but don’t provide the same top end crunch that the originaldelivers.

TFPro Vocal The release times were a lot easier to adjust, although theTFPro has a multi-stage circuit as you can see in fig. 3.35. Both curves fitneatly together, but the P8 reacts a bit faster at first, only to lose speedand approach the non-compressed final level slower.

When it comes to comparing the performance of vocals to the referenceunit, the differences are inaudible. When doing double-blind5 tests, I’m notable to spot either processor.

TFPro Drums A big plus of the TFPro, as with all other following 1176recreations is noticeable on stereo tracks. For the two single tracks, the width

5Double-Blind or ABX tests are used in professional listening tests to evaluate equip-ment, from audio compression-codecs, to signal processors like this. It represents a verysophisticated procedure, which helps to find out if a difference between to audio files isaudible, whilst keeping the chance of guessing at a minimum.


Figure 3.34: TFPro P8 : Attack phase for bass processing. The yellowoverlay displays the attack curve of UREI #1

Figure 3.35: TFPro P8 : Release phases of the driven vocal processing set-ting. As always the grey overlay represents the UREI reference unit.


control was of course set to mono, whereas on the stereo drums the unit offersvery flexible width control, where the Side signal gets more compressed,which results in a wider stereo field. For this test the default value 100%width was chosen, to keep the comparison with other competitors fair.

The behaviour of the TFPro P8 on drum-overheads is no comparisonto the original. It catches the transients in a very restrained manner anddoesn’t offer that famous snap of the reference unit. You can clearly hear theunit work, which results in a continuous level change when the compressorreacts on snare peaks. The driven setting is of course unfair, because Edwardthe Compressor doesn’t offer the british mode and was generally hard toadjust for similar settings.


Figure 3.36: Bomb Factory BF76 : Attack (fast setting)

Bomb Factory BF76

As the pictures 3.36 and following show, the BF76 is very close when itcomes to time constants. Attack curve is similarly fast, even though thefinal gain reduction is reached a lot quicker than on the original. Release isvery close, yet with the same Concave slope.

BF76 Bass The attack slope, as with all the other copies so far, reacts toofast on transients and arrives the final amount of gain reduction a lot quicker.It is closer to hardware test model #2, which leads to the presumption, thatthe reference unit might not be working as it should be.

During listening, the performance was as good as suspected from thegraphs, it even has more low end growl and sounds slightly punchier thanthe original. A quick look at the processed sine wave file of the BF76 revealsthat the attack time was a little bit off and reacted a few milliseconds slower.The time was tricky to adjust for all copies, because of the already mentionedlaid-back reaction of the reference model.

BF76 Vocal When you compare the settings 3.41, 3.43, 3.45 to the orig-inal, you’ll notice that the BF76 needs less input gain for the same amountof compression. As with all other copies so far, attack and release knobsdiffer a bit as well (as they did between the vintage units themselves).


Figure 3.37: Bomb Factory BF76 : Release (fast setting)

Figure 3.38: Bomb Factory BF76 : Medium settings


Figure 3.39: Bomb Factory BF76 : Attack (slow setting)

Figure 3.40: Bomb Factory BF76 : Release (slow setting)


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.41: Setting BF76 : Bass

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.42: Setting BF76 : Bass Driven

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.43: Setting BF76 : Vocal

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.44: Setting BF76 : Vocal driven


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.45: Setting BF76 : Drums

What seems to be noticeable is a slightly more open sound on the origi-nal, although I can’t verify that with the help of an ABX test, which (onceagain) prooves the closeness of this replica.

BF76 Drums The stereo image stability is again perfect, something thatcannot be said for the original. Input has to be reduced compared to theoriginal for similar behavior, even if the time-slopes are nowhere near to theoriginal. Figure 3.46 shows clearly the over-compression during the attackphase, where the signal gets louder after a hefty initial compression anda shorter, gentler release phase compared to the quick and steep originalcurve.

As you can assume from viewing the screencaptures of the BF76’s strangebehaviour, that the drum sound differs a fair bit from the original. Particu-larly the all-button-in setting results in heavy pumping, which doesn’t soundusable at all in my opinion.


Figure 3.46: BF76 : The driven slopes as used during drum processing. Thegrey overlay is the sine processing of the original UREI

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.47: Setting BF76 : Drums driven - Input had to be reduced in orderto achieve the same amount of gain reduction than the original


Figure 3.48: Bomb Factory MC77 : Attack (fast setting)

Bomb Factory PurpleAudio MC77

Both fast settings are as impressive as the original, the release time is in-credibly quick and outshines even the fast classic unit #1. Slopes are againslightly different, more clinical / mathematical if you like.

None of the copies so far were able to reproduce the two-stage attackbehavior of the original. As screenshot 3.50 shows, after the attack phase aconstant level is held (without any further level reduction) until the originalsine-tone is dropped in its level and triggers the release phase.

Even the slow attack setting is still really fast and no comparison to theUREI nor the BF76, as is the much more versatile release time that spansa wide range from under 200 milliseconds to over long 20s for full recovery!Besides an extension of the original design (the PurpleAudio hardware cloneis based on the same time constants than the original), which first of allclearly prooves that the MC77 and BF76 are definitely not identical plug-ins, as you would guess from their identical manuals.


Figure 3.49: Bomb Factory MC77 : Release (fast setting)


Figure 3.50: Bomb Factory MC77 : Medium settings


Figure 3.51: Bomb Factory MC77 : Attack (slow setting)


Figure 3.52: Bomb Factory MC77 : Release (slow setting)


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.53: Setting MC77 : Bass

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.54: Setting MC77 : Bass driven

MC77 Bass Attack had to be set a fair amount slower for close matching,whereas the release was tricky to adjust, because the control-range is reallytight from ultra-fast to very-slow. As with the former Bomb Factory plug-inthe amount of input was a hint lower than on the original, which concludesto finer adjustment range and more possible input drive.

Due to it’s ultra-fast design it reacts even on the slowest setting, asused for bass processing, too fast on transients and clearly distinguishablefrom the original. During the driven setting the low frequency distortionwas remarkably well reproduced, it seemed even more present than on theoriginal.

MC77 Vocal Similar to the bass processing with the Purple Audio cloneplug-in, there had to be some adjustments made to adapt the setting fora similar performance. But as soon as both attack and release spots werefound, the result is as undistinguishable as the other clones when it comesto vocal compression.

MC77 Drums Exactly as on its Bomb Factory sibling, the unit startsto overcompress on transient signals. A similar, inferior performance wouldhave been expected, but critical listening revealed a surprisingly good per-


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.55: Setting MC77 : Vocal

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.56: Setting MC77 : Vocal driven

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.57: Setting MC77 : Drums

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.58: Setting MC77 : Drums driven


formance. The compression is less crunchy, but does not result in heavypumping like with the BF76. If you waive the missing two-step attack reac-tion and extended release times, this digital clone is definitely a close match.


Figure 3.59: Focusrite Stellar : Attack (fast setting - yellow overlay repre-sents model #2)

Focusrite LiquidMix Stellar #1 & #2

Before we go into any technical details, the control options of both modelsshould be mentioned. As described in chapter 1, both copies are part of theLiquidMix DSP-card, that is a control surface at the same time. So a bigplus, for people that work mainly on a computer is the ability to controlthe plug-in with a tactile control surface. The knobs are unfortunately abit coarse to adjust, which means for fine tuning you have to reach for yourmouse again.

After comparing the waveforms on sine processing it was noticeable thatboth modelled units are suspiciously close. The suggestion that both replicassound similar was confirmed when the audio files were compared - therewas virtually no difference noticeable. This was really surprising since theimpulse responses were taken from two completely different originals.

That’s why both copies were consolidated into one section, to show howclose they are, and unfortunately, how different they are from the origi-nals. Especially, the release times (fig. 3.60 and 3.63) are incredibly long.When the unit is set to maximum the release constant spans over massive


Figure 3.60: Focusrite Stellar : Release (fast setting - yellow overlay repre-sents model #2)


Figure 3.61: Focusrite Stellar : Medium settings (yellow overlay representsmodel #2)


Figure 3.62: Focusrite Stellar : Attack (slow setting - yellow overlay repre-sents model #2)


Figure 3.63: Focusrite Stellar : Release (slow setting - yellow overlay repre-sents model #2)


Figure 3.64: Focusrite Stellar1 : Drums driven compression settings appliedto the test sine. Similarities to both Bomb Factory clones are show with theyellow overlay of BF76’s result

25 seconds and both LiquidMix reincarnations are not even back to 0dBgain reduction. What surprised me the most is, that neither stellar copyrecreates that second, slower attack slope because that is a really conspic-uous characteristic of the original and one would suspect that this sort ofsampling procedure would have captured that.

Stellar Bass As you have spotted in chapter 1.3.4, the control surfaceof both the LiquidMix hardware and software panel has a more commonlayout, with a threshold knob instead of the more input, more compressionprinciple. There is an input dial available, but tests have shown that this isjust used to increase quiet signals, since a boost of level over 0dB results inclipping in the digital domain.

Due to the much slower release time, you can hear both copies workingmore obviously, especially on the quicker slap part. It also doesn’t reproducethe low end growl on the driven setting, as heard on some of the other digitalemulations.

Stellar Vocal This is the first unit where a difference is noticeable becausethe compressor doesn’t back up quick enough. So the overall signal reductionis much higher, resulting in an overly compressed sound with absolutely nodynamics left. Once the recreation has detected a signal peak and reacts onit, the gain reduction stays on during the whole vocal snippet.


Stellar Drums The harder working settings of both Stellar copies ondrums raises the conjecture that the Focusrite copies had the Bomb Factorydesigns as a role model, or that both original hardware units are differ a lotfrom other revision models after those plug-in emulations were modelled.

Audio wise, both Stellar clones are way off the original. The amountof compression is much lower and saturation and distortion artefacts, par-ticularly noticeable on the peaking snare on the reference files, are missingcompletely, resulting in a much cleaner sound.


Figure 3.65: UAD : Attack (fast setting)

3.1.3 Universal Audio UAD 1176LN

A drawback of this recreation, which is noticeable on first adjustments, arethe tiny attack and release knobs, there is probably a way to fine adjustthem with a shift-click or similar, but if it would not have been possible touse a dedicated control surface, the adjustments with a mouse would havebeen really fiddly.

As you have probably spotted yourself by comparing the screenshots 3.65and following with those of the two original units, the UAD one is incrediblyclose to original #2. This is the first copy that is spot on when it comes tothe two-stage compression slope for the attack section.

Also the release times are closer to UREI 1176 #2 which probably con-cludes again that the reference was not a perfect choice.

UAD Bass The next conformity was found with input and output con-trols. Settings as used on the original could be copied over to the DSP-basedclone nearly 1:1, which portends on a very carefully matched digital copy.

Listening tests show a close performance to the original, with a nicecrunchy sound, as reproduced by some other copies, but also reveals slightbuzz, that isn’t noticeable on any other hence resulting in an unnatural


Figure 3.66: UAD : Release (fast setting)


Figure 3.67: UAD : Medium settings


Figure 3.68: UAD : Attack (slow setting)


Figure 3.69: UAD : Release (slow setting)

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.70: Setting UAD : Bass

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.71: Setting UAD : Bass driven


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.72: Setting UAD : Vocal

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.73: Setting UAD : Vocal driven

sound.

UAD Vocal Screengrab 3.74 shows again the very unique slope of thereference unit, which is another indication that UREI model #2 is likely tobe more ’original’ than model #1.

Performance-wise, there is again no real difference to spot when it comesto vocal compression. As with all other candidates the differences are neg-ligible.

UAD Drums The UAD copy really shines in the top class of all 1176models, the drum processing. The same amount of saturation is clearlyaudible, this clone is much closer in this overdriven setting than any othercompetitor so far. A look at the waveform indicates that the reaction time ontransients is faster when used with audio material, because the performanceon the sine tone is nearly identical. After a few concentration listeningpasses the differences was noticeable, but the performance is nonethelessunbelievable close, a hint longer attack time, would have probably solvedthis problem as well.


Figure 3.74: UAD : Release slope of the UAD, with the light grey overlay ofreference unit #1 when processing the driven vocal

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.75: Setting UAD : Drums


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.76: Setting UAD : Drums driven


Figure 3.77: PowerCore: Attack (fast setting)

TC Electronics PowerCore 24/7

The last product for this comparison is visually in the same league than bothoriginals and the UAD clone, but overcompresses on fast attack settings.Unfortunately is the release constant fairly off, with a much longer time asfound on other copies.

The maximum attack is still a hint faster as the original, whereas therelease time is extended to five seconds for full recovery.

PoCo Bass As you can see on figure 3.82, the PowerCore model hadto be set-up very different from the original in order to achieve the sameamount of gain reduction. Throughout the whole test with the TC 24/7,its release time was set to seven in order to match the original settings asclose as possible and even though this fastest setting possible was chosen,the PowerCore plug-in was still too slow.

The behaviour on bass processing is very close, the tonal character ispreserved, although it starts to distort audible on the sustaining part, whichmight be a side-effect of the shortest release setting. Strangely enough thismisbehaviour is not noticeable on the driven part, where it performs on asimilar level as the reference.


Figure 3.78: PowerCore: Release (fast setting)

Figure 3.79: PowerCore: Medium settings


Figure 3.80: PowerCore: Attack (slow setting)

Figure 3.81: PowerCore: Release (slow setting)


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.82: Setting PowerCore: Bass

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.83: Setting PowerCore: Bass driven

PoCo Vocal Surprisingly is the overall level noticeable louder as on allthe other compressors. When checking the RMS levels of the processed sinewave, the level bump is measurable. Because the levels (respectively in- andoutput values) were set before the time constants were matched, this unitreacts more sensitive to changes in attack and release time. Since this effectwas never noticed on any of the other units, it was unfortunately not checkedin this test, which makes the comparison not very fair for vocal limiting.

But other than an even more compressed sound, their digital recreationperforms well again.

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.84: Setting PowerCore: Vocal


1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.85: Setting PowerCore: Vocal driven

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.86: Setting PowerCore: Drums

PoCo Drums A specialty of the PowerCore clone is its all button mode.Where the other copies paid more attention to capture the behavior of thevintage original, the 24/7 simply adds the ratios together, depending onwhich button-combinations are pressed. So the approach of trying a britishmode would result is 44:1 ratio, nowhere near the UREI reference. Thedecision was made to choose the 20:1 ratio and perform the test with thissetting.

The results are less saturated and more tame than the original, whichputs it in the same league as the TFPro model. But this is understandable,

1176 PEAK LIMITER

RATIO METERATTACK

RELEASE

OUTPUTINPUT

24

30

36

48

8 0

6

12

18

24

30

36

48

8 0

6

12

18

3 5

71

3 5

71

20

12

8

4

GR

-18

-24

OFF

UNIVERSAL AUDIO

VU VU

-20-10

-7-5 -3 -2 -1 -0

+1+2

+3

Figure 3.87: Setting PowerCore: Drums driven


because as we know the distinctiveness of the all button mode is not onlya ’new’ ratio, but a series of weird behaviours, that need to be reproducedproperly for similar results.


3.2 Mixing Techniques

3.2.1 Preparations

Reference track

A tricky question, that needs to be decided early for a project like this,was to find a suitable recording to mix. Since I was involved in two biggerprojects over the last month, there were a few songs to choose from. Themain criteria was to find a song that hasn’t too many tracks and can bemixed in a straightforward way, without any excessive automation, sincemixes will be done on analogue desks, that don’t offer flying-faders, as well.The best option was found with a sort of slower soul-pop track, it offered asimple arrangement with the following instruments:

• Drums: 3x Kick (subkick out, dynamic in, boundary layer on pillow),3x Snare (three channels: dynamic top & bottom, condenser top),Hi-Hat, 2x Toms , 2x Overhead

• Bass: DI Box

• Electric guitars: DI, 2x dynamic

• Piano: 2x condenser

• Vocals: Lead, 2x harmony, 2x adlibs

Reference mix

Even if it might not be the best mix, a lot of time was spent on the original,native Logic mix. Several audio specialists and the musicians that played onthe song have given feedback on the rough mix, to get it to the final versionas it is now. The concept of this early mix is not so different from Brauer’sbussing, but subgroups are mainly used for processing the drums, guitarsand backing vocals together, in order to save processing power.

There is no upward compression going on, but all subgroups have acompressor inserted that is taking off a couple (4-6) dbs at the maximum.The whole mix includes no huge specialities:

After ordering and labelling the tracks a loop was set, so the song playsback seamlessly, for a rough balancing of levels, and even more important,placing all the instruments in the stereo field. Drums were positioned froman audience view, with kick and snare dead center, hi-hat and rack-tomslightly right, floor-tom medium left and overheads hard left and right. Af-ter starting off with a common stereo approach for guitars, piano and back-ing vocals, a professional engineer suggested to place the backing singersopposite the guitars, like a band was actually set up and recorded on oldersoul-classics, which turned out great. The BV’s seem like a compact unit,


Figure 3.88: Drums-, Guitar-, and Backing Vocal bus setup on the referencemix. The open plug-in windows represent the guitar signal chain.

that gives you the impression of a choir, rather than four overdubs. Thepiano stayed wide, but low in the mix.

Nearly every part of the drum kit has a compressor plus equaliser in-serted, to shape the sound of each instrument. Kick and snare spillage iskept quiet with the use of gate/expanders and the following compressoraccentuates the attack phase of both drums. Toms were automated in, be-cause they were hardly played, with a similar compression approach as forkick and snare. On the overheads worked a tape-saturation plug-in, for awarmer top-end and longer sustain on the cymbals. The di’d bass guitar waspepped up by Logic’s bass amp for a more authentic sound and kept undera constant level with the help of a moderate compressor. The same concept(except for the use of Guitar instead of the Bass Amp) was used for the di’dsignal. The difference is a harder working compressor (for more sustain)and a phaser plus stereo-delay, because the di guitar signal is automated induring the guitar solo for a distinctive lead sound, over the other parts ofthe song it sits really low in the mix. Main-vocals are of course compressedas well and put in a space on their own, using a combination of delay andreverb. Last but not least the piano with a hint of compression, which isalways a tough call, but in a dense pop mix like that, where the piano playsthis lower-midrange filler, there’s no other possibility but to keep it undercontrol.


Delay compensation and phase shift

Early research has shown that the big difference, when working with buss-compressed signals and their parallel, dry versions, between an analoguedesk and a digital computer mixing environment is the processing latencyand phase-shift that plug-ins introduce as opposed to analogue processors.Latency is a delay in milliseconds range (0 - 2056 samples equal 0 - 50msat 44.1kHz sample rate) that is required by the plug-in to process the data.Phase shift is introduced in several processors, a simple eq for instance workswith a delayed copy of the original signal an then mixing them back to-gether, because of phase cancellation/boosts, frequencies can be attenuatedor boosted.

In the past years there was always a big issue when it came to use plug-ins on sends, or subgroups. My sequencer of choice, the recent version 7release of Logic audio, improved the latency compensation from channelsonly to all available track types.

Figure 3.88 shows my test setup6 to proof fully working latency com-pensation. A stereo drumloop was assigned to the main out and sent to sixbusses in parallel. There was no signal sent to Bus 3 from the main channel,but bus 2’s output was assigned to it’s successor, to create a complicatedrouting. Each bus had at least one plug-in inserted, which are known to in-troduce a fairly huge amount of latency, like Linear Phase EQ. No insertedplug-in was actually processing data, so for example eq curves were flat andcompressor ratios set to 1:1 with the highest possible threshold. By skilfulinserting of phase reversing ’Gainer’ plug-ins, the split signal could be madecompletely silent, because of total cancellation. That ensures that latencycompensation is fully working.

6This experiment can be found on the supplied data-DVD


Figure 3.89: Perfectly working latency compensation in Logic7

Chapter 4

Summary and Opinions

4.1 UREI 1176LN Comparison

4.1.1 Subjective thoughts

The critical listener has already noticed where the recreations are close,and where they fall off compared to the original. As most of the normalprocessed audio files show, the differences are negligible since some settingscould not even be reproduced for certain clones (e.g. too long release times)but they still sound very similar, which leads to the conclusion, that thesame job could have been done by virtually any compressor.

As soon as the compression gets audible, input transformers are over-driven and transients get crushed, the contrast is more obvious. And in myopinion these are the aspects that count if a plug-in wants to sound like an1176 in all-button mode. Sure this is not your everyday setting, but if youreach for a distinctive processor, you might as well slam the meters to getthe sound the unit is famous for. I feel, there is nothing against insertingthe 1176 for subtle levelling, but what is the point if a modern, reasonablypriced workhorse like a Focusrite Compounder or a software compressor dothe same job? Let us not forget that Universal Audio’s hardware reissueretails at around £1,395 for a mono version and vintage units in good con-dition are priced up to twice as much!

For me, if I focus on the driven settings, the favourites are easy to pick:

1. Universal Audio 1176LN: If you stumbled across of any of the greatreviews that this DSP system has received in the past, the first placeis understandable. I’m usually very critical against exciting opinionslike that, but when I first saw the waveform on sine-processing, I knewthis reincarnation would sound good. Except from the fiddly controls,the UAD nails even the british mode very closely.

2. Second and third place would be a split between the PowerCore 24/7and Bomb Factory’s MC77. Both models maintain the distinctive

85

CHAPTER 4. SUMMARY AND OPINIONS 86

character of the original very well, the only obvious difference can beheard when it comes to drum processing.

3. All the rest: It sounds a little drastic, because most units are verygood sounding compressors. Keep in mind that this experiment (northis ranking) is not about which model is the best compressor, butwhich model is capable of sounding like a hardware 1176LN.

The big plus that a hardware unit has for me, is the total control aboutevery parameter. There is a dedicated knob for all important values andwhen it comes to adjust the unit to the perfect sound for your source signal,it is easier to achieve that with a proper hardware unit for several reasons.First of all, and that counts for any analogue hardware, from compressorsto full mixing desks, you focus on the most important thing when it comesto audio processing: The Sound. No visual distractions from fancy graphicaluser interfaces, no half-usable presets, just you and a pair of ears to find thatideal setting. I know that this is more a psychological thing, but not manyengineers are aware of this problem in my opinion. You can work againstthat issue by using tactile control surfaces and focus on the sound. As soonas you are familiar with how to remote control your DAW with the help ofa fader box, you can close your eyes and listen to adjust settings. Again,I’m not an advocate of the digital is bad philosophy, in fact I’m quite thecontrary, because visualisation can be a great help for a lot of critical appli-cations in the audio field, but when it comes to fine adjustment of musicalparameters, nothing beats a hands on control and the total focus on thesound. Furthermore, as silly as it may sound, working with famous hard-ware units like the 1176 pushes me to accomplish that extra 10% percent ofperformance, because you know that a unit like this was used on countlessexcellent sounding records.

On the other hand, the most obvious advantage of the digital copies (as wellas the one hardware model) is perfect stereo operation. 1178’s are rare andthe bad user reviews I’ve found on using the stereo adapter clearly speaks forsoftware clones on this one. This thought immediately leads to the abilityto use multiple instances in a single mix. Nowadays you might be lucky tofind a pair of vintage originals in your studio of choice, whereas, dependingon the plug-in model and your system of choice, you can easily load morethan 30 UREI clones if you want to. And because this thought is goingmore in a general digital and analogue comparison, you might as well thinkof the missing maintenance costs. A digital copy is (providing that yourcomputer system is fully working) always in perfect shape, sounding goodfrom the second you load it and will never need any adjustments to workperfectly. Favourite settings can be stored and more importantly you areworking in a total recall environment, that will allow you to boot sessions


from the past and they will sound exactly as you last saved them. Of coursethere is always the argument that not every computer workstation runs thesame plug-ins and eventually some of the will become obsolete over the timeand lose compatibility with newer systems, but a hardware unit could breakor be removed from the environment you are working in.

4.1.2 Listening test

To get external feedback besides my own opinion, a dedicated listening testwas compiled1. Both original hardware units were clearly labelled as UREI#1 & #2, but all the emulations were name Copy #X, so people had noclue what they were listening to. The files were uploaded on my webserveras 320kbps mp3s, which is of course not ideal to have a compressed audioformat for critical comparisons, but as research and test experiments haveshown, when it comes to encode single audio files with the highest possibleresolution, the quality is definitely good enough. Sure this question wouldlead into a more detailed discussion, but as the uncompressed audio fileswere about 350MB big, there was no other way to distribute them over theinternet, as this was the most suitable way to get enough responses.

The two big questions the project attempted to answer were:

1. Can you hear a difference between the two originals and the recre-ations? If yes, what- , and how obvious are the differences? Which ofthe copies do you prefer (include a short ranking)?

2. Have you used an original 1176 and a digital copy (or any other classichardware unit that has been digitally emulated)? Which one do youprefer and why? Where do you see the strengths and weaknesses ofboth systems?

As an overall summary it can be said that the majority of people re-sponded with similar thoughts as my own. Real differences could be pickedup between the harder driven files on bass and drums. In the more subtlecompression settings listeners said that it is possible to spot a differenceand after focused, concentrated listening a ’winner’ could be picked, butthat this would be pretty pointless, as all of them were so close that a tinytweak on any of them would relativise the outcome. Copies that came closewhen used in harder situations were in their opinion: The UAD 1176LN,Bomb Factory MC77 and the PowerCore 24/7 (the same result as my earlieropinion).

Even more interesting was that a majority of the recording engineers thatparticipated in the test didn’t insist on the typical ’analogue is so much bet-ter’ opinion that a lot of experienced people in the industry stand for. The

1A dedicated Logic session can be found on the supplied data-DVD, besides a standardred-book compatible audio CD


obvious reasons why people like to use the hardware was suggested earlier -the ears and the more focused listening when you are controlling a hardwareunit with knobs and dials and not a mouse. One of the most interesting andthoughtful replies were posted on a German Homerecording2 forum:

Okay, the software simulates an 1176, but which one exactly? The explana-tion: If you would have bought six different units over the years (no panic,sometimes you found a mint unit for nearly nothing, the marketing hypefirst kicked in by the end of the 90’s) they could have been labelled as fol-lows: two are stored in the basement, two have a small blue point ( ’once ina blue moon’), one has a red (standard for drums), and one a green (suitsrock vocals). Is the software trying to simulate the blue, red, green, or eventhe crummy basement unit?

The former head of development of a UREI production place explainsthe question for the quality spread in mass-production: ”Forget that! Thosedevices were assembled in an area where people used to smoke and drinkbeer at their workplace. Sometimes we were lucky if the lights lit up on thefront.”

Counter-question: All the models that could not compete in comparisonto an original 1176, are those maybe good, or really good software compres-sors? Does software have to emulate hardware to be good?

Although this was not the original question of this project, it pretty muchnails it. Software processing is now on a level, where it can easily competewith any hardware unit out there. In terms of reliability and functionality itis ahead of hardware anyway. It all comes down to the knowledge of the userin the first place - if you are not able to achieve a great result on a modernProTools, or any other native digital audio workstation, it is certainly notthe software that can be blamed for.

4.1.3 Conclusion

Finally, to answer the question that this project tries to cover in the firstplace: Yes there are differences audible between hard- and software, butso are between hard- and hardware! Some of the emulations are so close,that they only can be distinguished from each other when used with reallyextreme settings. In every other aspect, the clones behave in a very similarway, they look the same and can be adjusted (depending on the copy) usinga familiar control layout.

2www.homerecording.de


Figure 4.1: UREI #1 : Frequency Response

4.1.4 Further investigations

Research carried out during the test experiments revealed a secret, thatnot many people (including me) know: The UREI 1176LN is a programdependent compressor. This information leads to further questions, thatwould be interesting to answer: How could you measure program dependentcompression? If a solution can be found, how would have been the resultsof the test candidates?

The next point where no further examination was done is the evaluationof the ETF measurements. As pictures 4.1 and following show, can someinteresting information be found in those graphs. The UREI has a signif-icant treble boost starting at around 4kHz, that the second example, theBomb Factory MC77 plug-in, does not introduce.

Also, the worse phase response in the higher frequency range would beworth further investigation. Data like this could be used to compare thetest candidates on a more scientific level.

As a last idea, the measurements that were done with different frequen-cies to check the attack time have not been included in this dissertation.The decision was made, because it was just measuring the medium attacktime for all processors.

Screenshot 4.5 displays the medium attack setting on the reference unitand the UAD copy. The plug-in reacts a bit slower, with a smoother oper-ation. This is another area that could be researched further into.


Figure 4.2: UREI #1 : Phase Response

Figure 4.3: Bomb Factory MC77 : Frequency Response


Figure 4.4: Bomb Factory MC77 : Phase Response

Figure 4.5: UREI #1 : Medium attack for a 100Hz sine wave. The yellowoverlay shows the same medium setting of the UAD plug-in copy


Generally speaking, the intention of this project was to ensure as closelymatched settings as possible on all processors so as to ensure a fair compar-ison. For deeper investigations in the subject matter, the mentioned mea-surement methods would have revealed on what the differences are based,but the collation of all the files was always planned with a listening test,since it always comes down to whether a difference is audible.


4.2 Mixing Techniques

4.2.1 Subjective thoughts

The more subjective part of this project also turned out to be very inter-esting. The initial question if multi-bus- or upwards compression translatesfrom analogue to digital systems cab be answered with a simple yes forme. Although the concept wasn’t overdone, since the straightforward ar-rangement of the song, and the style of the music did not demand an overlycompressed mix, but there have been a series of instruments running in par-allel during each mix and none of the reveals obvious problems in terms ofphasing or unstable stereo imaging.

All of the four mixes are surprisingly close together, I would have thoughtthat the results have to be further apart from each other, if you think of thedifference in system and processors that have been used. Besides that, therewere different speaker systems and acoustics in every control room, whichwas the reason to use the Logic reference mix in the first place. AppendixA, the session diary gives an insight at the exact setup that was used foreach mix. I try to summarise the obvious differences between each versionin the following sections:

Logic Reference Mix

Surprisingly the Logic mix is the loudest of them. I would have thoughtthat parallel compression would help in getting a denser mix, and an overalllouder result. The reason for this might be, that my approach on the Logicsystem was not so different from Michael Brauer’s idea, as explained insection 3.2.1 and more importantly, there is a lot more bottom end comparedto all the other mixes, which clearly pushes the RMS level. The compressed,sustaining bass is the main reason for the higher average level, although thenoticeable difference, when switching between the files is not too great.

Audient ASP Mix

The first analog mix is at the same time, the untidiest one. I really missedfilters3 on this desk, because those are the first knobs for which I reach afteran initial balancing of all levels, to roll off high and/or low frequency bandsthat are not occupied by the track signal. Instead, the Audient offers highand low shelving4 equalisers, that can’t be used in a similar way like filters,because of their much gentler slope.

3Filter: Part of an equaliser that allows to cut signal from the top and bottom frequencyrange

4Shelving: Part of an equalisers where all frequencies over or under a certain centerfrequency are boosted or cut


Otari Elite Mix

In my opinion the mix done on the Otari is in the same quality range as thereference track, even if it wasn’t possible to run the Logic mix in parallel,as done on all the other systems, because there weren’t any spare channelsleft. I felt very comfortable working in that analogue space and tried hardto get a very good result. The only noticeable problem in there is the highnoise level compared to the other mixes. All of them are a bit hissy, whichis especially noticeable at the beginning, because of a slightly faulty DI box,as though this one is not acceptable. Every channel was soloed in order totrack down where the most hiss was coming from, but each single one wasfine, it seems like the noise simply added up.

ProTools Mix

On this mix I have clearly overdone compression and equalisation. It soundslifeless and thin in comparison with the other files. I also thought I knew themonitoring and the room sound very well, since I have worked in that studioa lot over the last months, but nonetheless I was unable to recognise the lackin bottom end and ended up with a very clinical sounding mix. Everythingturned out really dry, which might be due to the quite reverberant acousticsin that room compared to the other studios I’ve mixed in.

4.2.2 Feedback

From all the persons who gave me feedback on all the mixes, none of themcomplained about any audible phase problems. Most of them criticised somesmall issues on the mixes in general, like a little bit of sibilance problemson the vocals here and there, or the volume level of the guitar solo onsome mixes, nothing major. When I revealed that two of the mixes weremixed digitally and two on an analogue desk, most people gave the samerection that I though in the first place: The difference is marginal. Thereis absolutely no part of any of the mixes were you can go and say: ”Thissounds definitely better”.

4.2.3 Conclusion

Of course all mixes have been crafted by myself and I would never claim my-self as a perfect engineer, but this seemed like the only possibility to keepvariables, that would affect the result, at a minimum. Since this part isn’treally about digital and analogue summing, but how well upward compres-sion, or Brauer’s multi-buss principle translates on each system, the resultsfor those question are very satisfactory. There is no obvious degrading insound quality when running differently processed signals in parallel on ei-ther system. Sure, the amount of processing hasn’t been overdone in my


examples, but there is not much more than a healthy amount of compres-sion and equalisation that makes sense for parallel processing. The ProToolssystem has clearly shown that it is capable of doing that, and the Logic busexperiment prooves at least a fully working latency compensation systemand plug-in compressors work similar to the ones available in ProTools Isuspect.

Generally speaking, the multi-bus technique is easier to set up and aboveall more versatile than on a small analogue desk, which often don’t offerenough busses to accommodate a reasonable amount of stereo busses (atleast four make sense for regular projects). ProTools offers great flexibilitywith a unique routing concept, only found on really large desks. If youwork in Logic you have to use an auxiliary send if you want to split thesignal between multiple busses. The approach to use three stereo busses asa multi-band processor seems very useful to me, but certainly needs moreexperience to take full advantage of this technique.

Bibliography

[1] Universal Audio. UAD Plug-Ins Guide. Universal Audio, 2006.

[2] Creamware. Vinco operational manual.

[3] Digidesign. Bomb Factory Plug-Ins Guide. Avid Technology, Inc., 2004.

[4] Focusrite. Liquid Mix Cutsheet. Focusrite, 2006.

[5] http://en.wikipedia.org/wiki/Bomb Factory. Wikipedia information onBomb Factory.

[6] http://mixonline.com/mag/audio revision history/index.html. UREI1176 Revision History.

[7] http://www.nrgrecording.de/html/history.html. UREI 1176 History.

[8] http://www.uaudio.com/company/history/index.html. Universal Au-dio Company History.

[9] United Recording Electornic Industries. 1176LN Peak Limiter Manual.Urei.

[10] TapeOp Magazine. Michael Brauer on Compression.

[11] Bobby Owsinski. The Mix Engeineer’s Handbook. Mix Books, 1999.

[12] Michael Paul Stavrou. Mixing with your mind. Flux Research, 2003.

[13] TC Works. 24/7c Manual. TC Electronic, 2006.

96

Appendix A

Session Diary

As a preparation, to start with a blank ProTools session, the already finishedLogic mix has been cleared from nearly all inserted plug-ins and all files werebounced out using the export command ’All Tracks as Audio Files ...’, toensure timing accuracy with a common starting point. Nearly all plug-ins were removed because of the Match EQ1 used to weld lead-vocal takestogether that were recorded on two different days, in different rooms. Thatsaved some time fixing aesthetic problems in order to focus just on the firstmulti-bus mix.

A.1 10. Aug, 06 - ProTools HD3 Mix

A mix session in the all-digital Vestry 2 studios has been finished today.After importing all tracks and putting them in a logical order, the bussystem was set up. Brauer’s four subgroup system was copied as good aspossible, since the exact hardware processors he uses are not available inProTools. I ended up with a Purple Audio MC77 + Sony Oxford Eq for thebottom, a Bomb Factory Fairchild 660 for middle, the Bombfactory BF-3Afor top and the Joemeek SC2 on the flavour group as shown in screengrabA.1.

Everything was assigned to the groups as the inventor suggested. Drums/ bass to bottom, guitars to mid, piano to mid and flavour, the backing vocalsto top and the vocals to the dry main bus. A useful function of ProTools thatisn’t probably that common can be found at the output assignment buttonof each track. If the control button is pressed when a new output target isselected, the new destination is added to the old one, like assigning tracksto multiple busses on an analogue desk, which is very handy especially forBrauer’s technique (figure A.2.

To suit his style even more, all single signals were equalised to taste andno compressions was used on them at first. The first difficulty was then the

1An equaliser that is able to apply the frequency spectrum of one audio file onto another

i

APPENDIX A. SESSION DIARY ii

Figure A.1: Bottom, Center and Top compressors (from top to bottom)used for the ProTools Mix

APPENDIX A. SESSION DIARY iii

Figure A.2: The summed snare channel is assigned to the bottom group andnow routed to the main out in parallel.

bottom group, as it is extremely hard to adjust one compressor that suitsall signals. I wasn’t able to get a dense sound without obviously pumpingone of the instruments, so I decided to compress the bass in its own channeland split the signal between bottom and main mix, so I could back off thechannel fader, which resulted in a less hard driven bottom compressor, butstill balanced bass tone. The same approach was taken with kick and snare,which were routed to the main output in parallel. In addition to that I alsoput a channel compressor across the kick, not so much for compression, butfor shaping the sound, so the attack phase of the signal gets highlighted.

The mix itself was crafted with the Kurzweil KSP8 used as an externaldual-reverb processor and except for a short delay on the guitar and a verysubtle chorus/delay on the main vocals, no other effects were used. Sincethe goal behind those mixes is to get into multi-bussing, the main attentionwas drawn to that and not much time was spent on writing automation orspecial effects in the mix.

A.2 13. Aug, 06 - Audient ASP8024-36 Mix

The first analogue mix session for this project. Working on an all analoguesetup (well except for ProTools as a playback medium) is really enjoyable,it feels like actual work as opposed to sitting in front of a computer. Un-

APPENDIX A. SESSION DIARY iv

Figure A.3: Wiring up the group returns of the ASP8024 desk

fortunately it isn’t nearly as flexible as a digital system and needs a lot ofpreparation work. For starters the group sends of the 8 mono subgroups arewired to the patchbay, but not the returns, so an 8-way loom from bantamto quarter-inch stereo jack had to be made. As soon as the adapters for abantam to jack, cinch, female and male xlr (in pairs each) multicore weresoldered, the bus compressors were chosen and wired up.

For bottom an SSL MixBuss clone was used, center was processed bya Focusrite Compounder and so was the top (because the favoured stereoprocessor, a TC Gold Channel wasn’t wired up) and a Drawmer MX30 wasworking on the fourth flavour bus.

So another Compounder was available for lead vocals and bass, the insertchain for kick and snare was a Drawmer DS301 (working as an expander)followed by the DL241 compressor, the latter was also used on the toms.Unfortunately the Focusrite Green-3 VoiceBox and Equaliser were not wiredup either, so I ended up using a Bomb Factory BF76 for the DI’d sologuitar, that was manually automated in on the desk, and an BF-2a for thefirst adlib-vocal track, that gets really expressive and dynamic at the end.Except for two mutes on the piano and volume automation on one of thebacking vocal tracks, nothing else was processed in the computer.

For ambience / reverb and delays two Eventide Eclipse, a TC ElectronicD-Two and a Lexicon MPX1 were used as send delays, where the Eventide’sprovided a smaller room and a longer chorused vocal verb, the D-Two setup a slap-delay for guitars and the MPX1 used for a longer tap delay onlead vocals, guitar solo and a hint of piano.

The room and particularly the PMC IB1 speakers were new to my ears,so the Logic reference mix, was constantly soloed in, to keep the overall

APPENDIX A. SESSION DIARY v

Figure A.4: Layout of the Audient’s 8-mono groups

balance of the mix similar.

A.3 20. Aug, 06 - Otari Elite+ Mix

For the second analogue- and last mixing session for this project, the univer-sities flagship analogue studio was the control room of choice. Unfortunatelywere some parts of the studio in not fully working condition. For a start,the reasonably small session of 22 tracks had to be brought down to 16, be-cause 12 channels on the desk weren’t working (stuck meters and no signalin/output whatsoever) and the remaining 12 channels were a split of threestereo busses and three auxiliary returns. So each three channels for kickand snare where summed in ProTools to one output for both instrumentsand the same was done with the backing vocal tracks.

When it came to record the mix back into the computer another mis-behaviour was found. Only three of the first eight inputs of the Digidesign192 interface were working, after swapping outputs of the desk and changingpatch-leads, I noticed that all the other ones were distorting when I changedthe hardware inputs into the DAW.

Because of the missing channels, the fourth flavour bus was waived infavour of an effect return. The Otari is nearly as flexible as a digital desk,each channel input can be freely assigned and so can the output routing,which makes it perfect for complex subgroup set-ups like this. Bottom andCenter was sent through Focusrite Compounders and the Top group pro-

APPENDIX A. SESSION DIARY vi

Figure A.5: Messy patchbay of the Otari Elite. If you look on the last 12channels on the left half of the desk, you can spot the stuck meters

APPENDIX A. SESSION DIARY vii

cessed by a Behringer Composer MDX2100. The routing similar to the othermixes, with drums and bass going to Bottom, guitars and piano to Centerand backing vocals, overheads, piano and vocals to Top. Vocals and the sologuitar were the only two signals running in parallel.

The rest of the signal chain included Drawmer DS201 gates and theAmek System 9098 compressor on kick and snare. Drawmer DL241 ex-pander/compressor for the toms and two UREI 1176 for bass and vocal,where the bass was sent through a subtle chorus program of the EventideH3000. The same plug-in processors as in EFS1, a Bomb Factory BF76were used for the di solo guitar and BF-2a on one adlib vocal track.

A short room was provided by a Lexicon PCM70 and the main vocalreverb came from a 480L. The guitar slap-delay was once again created withthe TC Electronics D-Two unit.

As with EFS1, I’m not used to the room nor the speakers and since Ihad no channels left on the desk, the new mix could not be compared on thefly. So after the a mix was recorded back into the computer it got comparedto the reference mix and changes were made, to ensure a sonically similarresult.

Appendix B

Supplied data mediums

B.1 Data DVD

• 1176Comparison: The Logic file can be used to examine the audiofiles in a very easy and flexible way. Markers are set up to switch be-tween the different test loops. It is recommended to use the (alreadyactivated) solo function and switch between tracks per keyboard ar-rows or mouse clicks. If you scroll further to the right you will findthe appropriate sine-test region for each setting and the same for themaximum and minimum settings on each unit.

• Documents: Contains all relevant research documents in electronicform

• MixingTechniques:

– BetterDays Logic: This folder contains the reference mix craftedin Logic 7

– BetterDays Tools: Includes the native ProTools mix, as well asthe files for both analogue sessions, with dedicated output assign-ments

– Bounces: All final mixes can be found in the Mix folder. Mastershouses the cut and limited to a radio-competitive level of -10dBRMS final files

– LatencyTest: Contains the test experiment to prove Logic’s la-tency compensation on busses

• Presentation: A power point file, for presentation of the project

viii

Appendix C

Research Folder

ix