Compellor
Compellor
Compellor
Compellor ®®
Owner’s Manual
Dual Mono/Stereo Automatic Level Controller
S Y S T E M S
11068 Randall St., Sun Valley, CA 91352 U.S.A.
Quick Start 2
Introduction 3
Installation 4
Specifications 5
Operating Instructions 6
System Description 7
Appendices 9
Safety Declarations
CAUTION: For protection against electric shock, do not remove the cover. No user serviceable parts inside.
WARNING: This equipment has been tested and found to comply with the limits for a Class A digital device pursuant to
Part 15 of the FCC Rules. These limits are designed to provide reasonable protection against harmful interference when
the equipment is operated in a commercial environment. This equipment generates, uses, and can radiate radio frequency
energy and, if not installed and used in accordance with the operating guide, may cause interference to radio communica -
tions. Operation of this equipment in a residential area is likely to cause interference in which case the user will be required
to correct the interference at his own expense.
The user is cautioned that changes and modifications made to the equipment without approval of the manufacturer could
void the user’s authority to operate this equipment.
It is suggested that the user use only shielded and grounded cables to ensure compliance with FCC Rules.
| Conforms to standards
UL60950 and EN60950. C
59887
®
US
Page 2
3. Introduction - Page 7
3.1 What Is A Compellor?
3.2 What Does It Do?
3.3 How Does It Work?
3.4 A BIt Of Compellor History
4. Installation - Page 10
4.1 Unpacking
4.2 Damage & Claims
4.3 Main Voltage Selection
4.4 Power Cord
4.5 Mounting In A Rack
4.6 Proper Ventilation
4.7 Panel Security
4.8 Tools & Equipment Needed
4.9 Safety Considerations
4.10 Remote Connector
4.11 Reference Level Setting
4.12 Input Connections
4.13 Output Connections
4.14 Summary
5. Specifications - Page 14
5.1 Inputs
5.2 Outputs
5.3 Audio
5.4 System Functions
5.5 Threshold
5.6 Ratio
5.7 Attack Times
5.8 Release Times
Page 3
6. Operation - Page16
6.1 Introduction
6.2 Recording
6.3 Mixing
6.4 Mastering
6.5 VIdeo Post Production
6.6 Sound Reinforcement
6.7 Live Concerts
6.8 Broadcast Radio Pre-processing
6.9 Broadcast STL/Phone Line Driver
6.10 Television Broadcasting and Cable Systems
6.11 Video and Audio Tape Duplication
6.12 Voice Processing
6.13 Hard Disk Recording
7. System Description - Page 20
7.1 Model Differences
7.2 Signal Flow
7.3 Processing Functions
7.4 Leveling Function
7.5 Compressor Function
7.6 DRC
7.7 DVG
7.8 Silence Gate
7.9 Stereo Enhance
7.10 Stereo Linking
7.11 Meter Selection
7.12 Limiter
7.13 Process Balance
7.14 Drive Control
7.15 Output Control
7.16 Process Switch
7.17 Input/Output
7.18 Operating Levels
7.19 Input/Output Metering
7.20 Gain Reduction Metering
9. Appendices
Apndx A. Balanced & Unbalanced Lines and Operating Levels - Page28
Apndx B. Dealing With Grounds and Hum - Page 29
Apndx C. Proper Wiring Techniques - Page 31
Apndx D. Standard Cable Wiring - Page 32
Apndx E. About Reference Levels - Page 36
Apndx F. Digital-vs-Analog; Peak-vs-RMS: How To Deal With The Confusion - Page 38
Page 4
Quick Start
1. Make sure there is signal going through the Compellor with Process both “In” and “Out”. If
not, check the input and output wiring. They may be reversed. Be sure to check for the cor-
rect input selection (analog or digital) on the rear panel. Leave Compellor in bypass (Process
“OUT”) until finished with set up.
3. Switch Compellor “Meter Select” to Input. The last red LED should indicate approxi-
mately ‘0’ on the meter. If not, adjust the rear panel REF LEVEL switch to the position which
gives you the closest reading to ‘0’.
5. Switch the Meter Select to G.R. (gain reduction). Adjust the Drive control to achieve 12dB
of gain reduction with ‘0’VU input. The last lighted LED shows the total amount of gain reduc-
tion occurring.
6. Set Leveling Speed to Slow if you are controlling full program, Fast if you are controlling
live voice.
9. Stereo Enhance: If using the Compellor for mono or dual mono operation, switch the
Stereo Enhance to “Off”. If using the Compellor for stereo program, switch the Stereo
Enhance to “In”.
10. Link: If using the Compellor for mono or dual mono operation, press the “Unlink” button.
If using the Compellor for a normal stereo program, switch the Link to “Leveling”. If using
the Compellor for any matrixed stereo program (e.g.- surround encoded), switch the Link to
“Leveling & Compression”.
11. Switch the Meter Select to Output. Adjust the Output control so that the red part of the
level meter indicates 0dB. Switch the Compellor into circuit (Process “In”). The Compellor will
now act as a unity gain device whenever the input level is at zero VU, and make gain correc-
tions for higher and lower incoming levels.
Page 6
Without a Compellor, it is usual to insert a compressor or limiter in the line to control varying
levels. That always results in degraded sound due to the processing by-products. Lost punch,
overly fat backgrounds, inversion (when a loud sound gets lower than average), suck-down
by transients, and noise swell ups are typical problems encountered with usual processing.
The Compellor was designed specifically to avoid all of these problems and more.
In contrast, a Compellor automatically detects and corrects the sound level according to how
we hear, and therefore seems natural and relatively undetectable. The unique and patented
circuitry in a Compellor resulted from years of experiments in audio processing and creates
the only level controller on the market designed specifically to be as “transparent” to the ear
as possible. Additional information about the processing circuits in a Compellor will be found
in the various sections of this manual.
Page 7
Some owners may be interested in how the Compellor was first developed. The story begins
in Hawaii in 1982 when Donn Werrbach, a consulting broadcast engineer, undertook to design
an advanced AGC unit for on-air processing to improve the sound of radio stations. Werrbach
The Compellor has become the world standard audio level controller.
Understandably, we are very proud of that fact!
had been experimenting with broadcast audio processing for many years but needed to find a
good enough VCA (voltage controlled amplifier chip) to fully implement all the new processing
techniques he had discovered. A chance contact with Boyd Collings, who was then the Aphex
agent in Honolulu, introduced Werrbach to the type 1537A VCA chip which was produced and
sold by Aphex. Given a free sample, a couple of weeks time, and the inspiration brought by
the VCA’s fabulous performance, Werrbach produced the first Compellor prototype.
Werrbach’s prototype found its way not only into on-air trials but into a tape duplicating lab,
an album recording studio, and several live showrooms where it quickly proved its usefulness
as a gain controller without processing artifacts. At Boyd’s urging, Aphex’s product manager
Jon Sanserino visited Honolulu and auditioned Werrbach’s prototype at the Audissey record-
ing studio where he was intrigued by its possibilities. Finally, in 1983, an agreement was
reached between Werrbach and Marvin Caesar, the president of Aphex Systems, to produce
the Compellor as a product line.
The first unit rolled off the line in 1984 as the Aphex Model 300 Stereo Compellor. Patents
were secured for key inventions of the Compellor circuitry and are assigned exclusively to
Aphex Systems.
As a premier product line, Aphex decided to build the Model 300 to the highest commercial
standards including only the best available parts and construction techniques. As a result,
not only is the audio processing performance outstanding, but the reliability and long lifetime
of the product was assured. Thousands of Model 300’s are still in constant use today, some
with as much as 19 years of duty under 24-hour service!
The next models introduced were the Models 301 and 303 based on the Model 300 design.
The Model 301 was a single channel version, while the Model 303 was a Model 301 with an
Aural Exciter (tm) added. These models are also still in widespread use.
Page 8
Now, in 2003 (as this manual is being written), the Compellor is still the most advanced and
effective audio level controller available.
Page 9
4.1 Unpacking
Your Compellor was packed carefully at the factory in a container designed to protect the unit
during shipment. Nevertheless, Aphex recommends making a careful inspection of the ship-
ping carton and the contents for any signs of physical damage.
AC Line power is supplied to the unit via an integral receptacle/fuse holder on the rear panel.
This receptacle meets the various international safety certification requirements, provides the
international mains power selection, and serves as a radio frequency line filter. The pro-
grammed voltage can be read near the left end of the fuse clip on the surface of the program-
ming card. If the incorrect voltage is seen, proceed to reprogram the voltage.
Reprogramming the mains voltage is easy if the following steps are followed. Remember to
check the fuse value and install the correct fuse as indicated.
FUSE
1. Slide window open PULL FUSE DATA
100-120VAC
0.375 Amp
Slow Blow
2. Pry out “Fuse Pull”
220-240VAC
3. Extract programming card 0.25 Amp
Slow Blow
Programming Card
100 240
Side 1 Side 2
120 220
Page 10
To minimize the risk of shock or fire, do not expose the unit to moisture. Allow adequate
ventilation around the unit for cooling. Make sure the mains voltage is properly selected. Do
not open the chassis cover: there are no user serviceable parts inside.
Page 11
Normalizing the Compellor is accomplished by a rear panel REF LEVEL switch provided for
each channel. Two standard reference levels of -10dBV and +4dBu are available. Simply
set the switches as required.
If you have a nonstandard operating level, select the closest setting to your operating level.
For DAT machines and other digital media that define operating levels according to a maxi-
mum level rather than an average level, we have found the -10dBV position most often pro-
vides the correct match.
The input impedance is 20 kilohms and the Compellor will not significantly load the source
when the unit is in-line. Inputs are made by means of 3-pin female XLR jacks. Pin connec-
tions follow conventional standards. Pin 1 is connected directly to chassis ground. Signal
pins 2 and 3 may be used either as pin-2 positive or pin-3 positive as you wish. Current U. S.
and international industry standards call for using pin-2 as the positive polarity lead.
For unbalanced use, tie pin 3 to pin 1 for the ground and use pin 2 as “hot”.
Whether using balanced or unbalanced wiring, be sure to follow the same connection scheme
for both channels of the input and output wiring to avoid audio phasing problems.
Page 12
Output connections are made by means of 3-pin male XLR jacks. The pinout follows the
same conventions as the input jacks described above, and you should exercise the same
care about wiring as described for input wiring. Refer to Appendix 1 for complete details
about wiring and interfacing to other equipment.
4.13 Summary
If you pay attention to the line voltage setting, reference level, and i/o wiring you should have
no trouble operating the Compellor. If any difficulties are experienced while installing the
Compellor, other information contained in this manual will probably supply adequate assis-
tance. Please study this manual before contacting the factory for assistance.
Page 13
5.2 OUTPUTS
Connector: 3 pin XLR male
Type: electronic servo balanced (unbalanced without 6dB loss)
Impedance: 65 ohms bal or unbal (nominal load 600 ohms or greater)
Max level out (bal): +25dBu(ref = +4), +10.8dBV(ref = -10)
Max level out (unbal): +20dBu(ref = +4), +10.8dBV(ref = -10)
5.3 AUDIO
Frequency response: +/- 1dB 10Hz to 65KHz
Hum & noise: measured for a 1KHz tone at unity gain
No gain reduction: -67dBu(ref = +4), -86dBV(ref = -10)
10dB gain reduction: -74dBu(ref = +4), -89dBV(ref = -10)
Crosstalk @ 20KHz: -65dBu(ref = +4), -78dBV(ref = -10)
Dynamic THD typically .05% for 1KHz at 20dB gain reduction
Static THD: .025% at maximum output level
IMD, max output: .13%(ref = +4), .4%(ref = -10)
5.5 THRESHOLD
Compressor: 30dB below nominal level
Leveler: 30dB below nominal level
Limiter: 14dB above nominal level
Page 14
Page 15
6.2 Recording
Regardless of how consistent a musician or a vocalist may be, when there is a ‘take’ the levels
can vary dramatically from the rehearsals. Trying to ride faders in real time is impractical, if
not impossible. You want to have real time control over the levels, but not change the musical
character of what you are recording. The Compellor is the perfect solution.
Set up the Compellor for no more than 4 to 6dB of gain reduction with nominal ‘0’ VU in, use
fast leveling speed with Process Balance at 12 to 1 o’clock. This setting will give you 4 to 6dB
of increase in the lowest level signals and keep overly enthusiastic performers from overload-
ing the recorder. When using the Compellor as a dual mono unit make sure that the Stereo
Enhance is off and the Unlinked is in.
Using the Compellor while recording will give you more consistent levels to work with when
you mix, thus making the mix process faster and easier. It will also allow you to maintain better
signal to noise performance in the recording medium.
6.3 Mixing
Layering elements within a mix so that each ‘sits’ in its own pocket, not interfering with other
elements is a critical part of assembling a high quality mix. Typical multitrack mix downs
become very complex. Gain adjustments on the different elements can be manually made
through the mix, programmed through automation, or the Compellor can be used. By using
the Compellor on an individual track or subgroup, that element can be ‘fit’ into the mix more
easily and require less attention from the mix engineer.
A background vocal subgroup, for example, may disappear into the noise or become lead
vocals. By using the Compellor, the background vocals stay in the background, but at the
right level.
Use one channel of the Compellor as a pre-fader insert, or on the output of a buss and return
the buss into a fader. This will give you the ability to automatically control the levels of the
element and also adjust the level of the return if necessary. Note: If you insert the Compellor
post-fader, the Compellor will ‘unwind’ any fader moves.
Using the Drive control, adjust the amount of total gain reduction for the amount you want to
bring up the lowest level signal (usually no more than 6 to 8dB). Process Balance should be
set more towards Leveling (CCW) for a more dynamic sound and more towards Compress
for tighter instantaneous dynamics.
6.4 Mastering
A recording studio is not the typical listening environment. Most listening is done in a higher
noise environment with less than optimal acoustics (e.g.- a car). Monitoring in a studio is
usually done at higher levels than in a typical playback situation. Furthermore, if the master
Page 16
The goal, of course, is to maintain the sound quality of the studio while you are trying to
‘tighten’ the mix. Adjust the Compellor for 2 to 4dB of gain reduction with the Process Balance
at 11 o’clock. Stereo Enhance should be in and Leveling Link in. If the mix was encoded for
surround or has some other matrix processing, use the Compression and Leveling Link. Note:
If there is one dominant element in your mix, the Compellor will work on that element causing
the lower level elements to be ‘ducked’. It is therefore important to control the elements in the
recording and the mix down before the mastering process.
Use the Compellor on individual elements or subgroups and you will not have to work the fad-
ers as often or as hard. The Compellor can be used as a final gain rider so that the average
level of the concert stays within defined dynamic range.
Be careful not to use too much gain reduction on vocals and instruments which are miked so
that you avoid feedback when the Compellor is fully released.
Page 17
The Compellor recognizes the differences in density and adjusts it release times based on
the peak to average ratio of the input. If an input is already heavily processed (e.g.- a com-
mercial), the Compellor slows its release time. When the Compellor sees a very high peak
to average ratio (e.g.- a live news broadcast) it speeds its release time. Segues between
program and commercials are much smoother.
Another complaint made to cable systems is the difference of levels channel to channel. If
each channel of a cable system has a Compellor and all the Compellors in a cable system
are set similarly, channel to channel differences will be eliminated.
The Compellor is also useful when assembling a single tape from multiple sources. The level
differences between the sources are automatically smoothed out, obviating the need for stu-
dio time to level adjust each source before mastering.
Consonant recognition is a critical part of intelligibility. If the processing is not done correctly,
the consonants will be ‘crushed’ causing the loss of intelligibility. The Compellor effectively
controls the levels without loss of intelligibility, indeed because the levels are more controlled,
the intelligibility will be enhanced.
It is recommended that Leveling Speed be set for ‘fast’ when processing voices.
Page 18
Page 19
The leveler has two operating speeds which can be selected by a front panel switch: “fast”
and “slow”. The slow speed will not affect musical dynamics but will act fast enough to follow
the general density trends of a program mix. The fast speed is better for plain voice work
as it can follow the faster and more unpredictable voice changes of announcers and singers
as they weave around the mic or use expression. The frequency discriminate leveling in the
series “A” Compellors has so improved the leveling function that fast leveling is now feasible
for use with musical programs to materially increase the loudness density of a mix without
causing objectionable bass pullback and bass pumping.
The leveler is influenced by both the silence gate and dynamic verification gate whereby the
gain control generated by the leveler can be frozen by either of the two gates. This means
that either gate can make the leveler stop changing the VCA gain and hold the most recent
leveling value.
The frequency discriminate leveler, as opposed to the previous leveler, responds more slowly
to low frequencies than to higher frequencies. This represents a significant improvement
whereby the ear can perceive much less effect of the bass signals controlling the Compellor
gain. The prior leveler responds to all frequencies at the same rate.
Page 20
The attack and release times of the compressor are program dependent as a function of the
audio waveform’s complexity. Thus, most of the sonic artifacts of compression are minimized
or eliminated. It can be generalized that transient sounds will cause faster attack and release
as well as greater compression than continuous and slow changing sounds. Like the leveler,
the compressor is also influenced by the DVG and silence gate. Either gate can force the
compressor to freeze and hold its gain control at a steady value.
7.5 DRC
The dynamic release computer, or “DRC” is directly imbedded in the compressor to control
the release time. This circuit detects the audio waveform and directly affects the compres-
sion detector. The result is a compressor which responds differently to “fat” and “dense”
sounds than for “thin” and “peaky” sounds. This is how the Compellor can better match the
changing elements in a program. For example, the DRC helps the Compellor match up a live
announcer’s voice level and density with the relatively heavier density of a recorded cut-in or
segue without resorting to excessive compression effects.
7.6 DVG
The “dynamic verification gate” continuously detects the Compellor’s processed VCA output
signal and computes the historical average of peak values. It also “verifies” when the present
peaks exceed or equal the historical average and outputs a “stop and go” control for both the
leveler and compressor functions. Whenever the present peak amplitude is below the histori-
cal average, the leveler and compressor gain controls are frozen by “gating” their respective
level detectors into a “stop and hold” mode. Otherwise the level detectors are gated into the
“ track and go” mode. Thus, the word “gate” does not indicate the audio signal itself is being
gated in any way, although our use of the term “Silence Gate” has confused some uninformed
users.
The DVG, described above, needs a continuing source of audio to maintain its operation.
If the audio should stop for any reason, the DVG first freezes the VCA gain then simply
relinquishes all control after about 1.5 seconds of silence. This would allow the Compellor
to begin recovering gain and thus bring up the background noise level. To prevent this, the
Silence Gate is used. Whenever the input audio signal drops below the user selected silence
threshold for one second or longer, the silence gate circuit freezes the VCA gain in lieu of
the DVG circuit. You should set the Silence Gate threshold to a level above the typical back-
Page 21
If, for some reason, you want to prevent the Compellor from bringing up program fades below
a certain point, simply set the Silence Gate threshold to the level where you want the fade to
become “uncorrected”. The Compellor will then freeze its automatic gain control and let the
program continue fading out naturally.
The audio level selections set up the meter in two colors to simultaneously indicate the aver-
age, or “VU meter” level, and the peak level. VU is shown in red and peak in green. In the
gain reduction setting, the meter uses two colors to simultaneously indicate the leveling and
compression gain control.
7.11 Limiter
The peak limiter has a very fast attack of less than 1 microsecond. This positively stops audio
peaks from escaping the VCA if the limiter is turned on. The threshold of limiting is about
14dB above the output zero VU reference level. With such a high threshold, the limiter only
activates occasionally when excessively high transients are present in the output audio.
Page 22
7.16 Input/Output
The Compellors use the Aphex active servo balanced input and output stages. We like to
specify pin 2 as positive but this is purely arbitrary since the balanced input and output pins
are directly in phase and there is no dedicated unbalanced output which would be specified in
relation to the input phase. You could just as well call pin 3 positive if that is what you want.
This would translate to an average signal level of +4dBu with a peak value of +13dBu if the
Compellor were normalized to +4dBu. Note that, since the peak value is always greater than
the average value of an audio signal, the green bar always appears to the right of the red
bar.
6 Reds 3 Greens
Page 23
6 Reds 3 Greens
1 Red 7 Greens 2 Dark
Figure 7-2 Bargraph G. R. Meter Example
If all gain reduction is due to compression, then the indication will be a totally green bar, and
the whole bar indicates compression. If all the gain reduction is from leveling, then the green
bar will have a red dot at the right most point. Figure 7-2 shows an example of indicating
16dB total gain reduction with 4dB of leveling. 12dB of compression is therefore inferred.
Page 24
SCOPE
All defects in workmanship and materials. The following are not covered:
a. Voltage conversions
b. Units on which the serial number has been defaced, modified, or removed
c. Damage or deterioration:
1. Resulting from installation and/or removal of the unit.
2. Resulting from accident, misuse, abuse, neglect, unauthorized product modification or failure to follow
instructions contained in the User’s Manual.
3. Resulting from repair or attempted repair by anyone not authorized by Aphex Systems.
4. Occurring from shipping (claims must be presented to shipper).
WHO IS PROTECTED
This warranty will be enforceable by the original purchaser and by any subsequent owner(s) during the warranty
period, so long as a copy of the original Bill of Sale is submitted whenever warranty service is required.
LIMITATION OF WARRANTY
No warranty is made, either expressed or implied, as to the merchantability and fitness for any particular purpose.
Any and all warranties are limited to the duration of the warranty stated above.
Some States do not allow limitations on how long an implied warranty lasts and/or do not allow the exclusion or limita-
tion of incidental or consequential damages, so the above limitations and exclusions may not apply to you.
This warranty gives you specific legal rights, and you may also have other rights which vary from State to State.
Page 25
Grounding
There is the notion that some king of earthly “ground”
exists out there that sinks all the noise and acts as
some kind of a shield. You see wires connected to
ground rods and water pipes that are supposed to
get a good ground. This is not a correct interpreta -
tion of grounding from an audio standpoint. Proper Figure 2 Unbalanced Line Model
grounding of equipment and wiring is important and
you will gain a better understanding of that as you
read along. Many people, because they have more experience
with unbalanced wiring, think that balanced is
Balanced -vs- Unbalanced confusing. Believe it or not, balanced lines are really
Every audio signal is connected through a circuit. The easier to understand than unbalanced. There is no
circuit must contain two conductors to create a com - grounding issue with balanced, and the way it works
plete return path. In other words, a signal voltage is is perfectly natural and simple. Balancing naturally
conducted to a piece of equipment by injecting a cur - rejects hum and noise and eliminates all sorts of
rent into a wire. That current has flow though to the complications in interfacing.
destination through the wire and return back to the
source through another wire. Since audio is an alter - Balanced transmission works something like this.
nating voltage, swinging through negative and posi - Your balanced input stage looks at the two wires
tive polarity, the current through the two conductors and detects only the potential (voltage) difference
changes direction each alternate half cycle. Which between them. Anything that is the same on the two
wire is the source and which is the return alternates wires (for all practical purposes as seen measuring
accordingly. In this regard, balanced and unbalanced from ground) is called a common mode signal and
Page 26
Page 27
Page 28
Page 29
audio line from a low impedance and receive into a The following instructions show all the different ways
high impedance. Generally, a minimum 1: 10 ratio is you will probably ever need to hook up your 320D as
possible. This is called “bridging”. This has become well as any other equipment you may own. You will
modern practice and all balanced inputs are normally see that connecting balanced outputs to balanced
running 10K ohms or higher impedance. Because inputs is ultimately simple and the same cable will
of these developments, it is no longer as critical to work for all flavors of output stages.
consider impedance when dealing with interfacing
pro line level equipment (impedance “matching” is Connecting a balanced output to an unbalanced
mostly a requirement of the past). input requires a little more knowledge and care.
You should refer to your equipment manuals and
determine the type of balanced output stage that is
A word on impedance and interfacing adapters: provided, then use the correct “transition cable” as
If you are connecting between two line level devices depicted in this section. Improper transition cables
and they have different connectors (example: 1/4” can cause crosstalk, hum, and distortion problems
phone to XLR or vice-versa), you do not need to use within your system.
an impedance matching transformer. With very few
exceptions you are strictly dealing with a difference TYPES OF BALANCED OUTPUTS
in connector types and should only use hard-wired Believe it or not, there are at least 5 types of balanced
adapters (or cables) for this situation. output stages in use today. They may be placed
into two main classes: transformer balanced, and
transformerless balanced, usually called “active bal -
anced”. Transformer balanced outputs are becoming
APPENDIX D: Standard Cable Wiring outdated because of their high cost and their sonic
limitations. However, they can still be found on a lot
In relation to 1/4” phone jacks, you may see the terms
of older equipment.
“TS” and “TRS” as abbreviations. Here is a what that
means: TS refers to the Tip-Sleeve or “mono” 2-
Within the transformerless class, there are several
conductor type and TRS refers to Tip-Ring-Sleeve or
types of circuits that are used by different manufac -
“stereo” 3 conductor type 1/4” phone connectors.
turers. These different types of output circuits all look
This applies to jacks (female connectors) and plugs
just about alike to any balanced line, but they act dif -
(male connectors).
ferently when driving an unbalanced line. You need
to observe the proper cable wiring for each type of
Page 30
OK
Standard store-bought cable. Shield is grounded at both ends.
Positives: Both ends are interchangeable.
Negatives: May cause ground loops through shield contacts.
No Connect
BETTER
Custom cable. Shield is grounded at receiving end only.
Positives: Stops ground loops and reduces noise.
Negatives: Should be oriented so lifted shield is at sending end.
No Connect
Stops Ground Loops
Page 31
Female XLR
SIMPLIFIED SCHEMATIC
Female XLR
Ground Pin 3 Directly to Pin 1. Impedance Balanced Outputs
Alternatively, Carry Pin 3 Through Twisted Pair Cable
and Ground at Other End
Female XLR
Pin 3 Doesn’t Matter
OK Grounded or Not Grounded
SIMPLIFIED SCHEMATIC
Page 32
Mono (TS) Phone Plug Stereo (TRS) Phone Plug Mono (TS) Phone Plug Stereo (TRS) Phone Plug
(Guitar cord of Part 3 above usualy works just as well) Not Used
Male XLR Male XLR
Mono (TS) Phone Plug Mono (TS) Phone Plug Not Used
Male XLR
To Equipment Input
Ground Shield This End Only
To Equipment Input
Ground Shield This End Only To 207
Insert Jack
From Equipment Output
Mono (TS) Phone Plug Ground Shield This End Only
Page 33
ANALOG SYSTEMS
Systems declaring the average reference level are very different than systems declaring the
peak reference level. In the United States, most analog systems still use the VU meter and
we declare the +4dBu (for example) reference level to be the average program level. Peak
program levels may greatly exceed this level but sufficient headroom is allowed in the elec-
tronics to safely carry any unseen peaks. In a peak declared system such as practiced in
Europe, a maximum signal level is declared as the reference and a Peak Program Meter is
used to observe the program levels.
In an average reference system, peak levels may exceed the reference level by as much
as 20dB.
Thus, a +4dBu referenced system may see peaks as high as +24dBu. If we carefully con-
trolled a mixed program to keep its sound level constant, we would see fairly consistent VU
indications, but extremely variable PPM indications. Likewise, if we mixed the same program
to keep the PPM indications consistent, the sound level would vary.
Since the Compellor is expressly concerned with controlling the sound level as the ear
perceives it, only the average level bears relevance. This is an important concept to
grasp if you are used to dealing with peak responding level meters, because you can-
not see the Compellor’s benefits on peak meters.
When you set the Compellor’s REF LEVEL switch to match your system reference, an
assumption is made of an average reference level.
For peak referenced systems, such as the +6dBu German system, the average program level
will reside far below the reference level (typically 10 to 15dB below, or around -8dBu) but will
be uncertain and variable depending on the peak factor of the particular sound.
DIGITAL SYSTEMS
Digital recording has almost universally adopted peak level metering. Digital level meters
have 0dB at the top of the scale. That is defined as 0dBFS, or 0dB referred to full scale. It is
impossible to have a digital signal that exceeds 0dBFS in peak value. If recording a signal that
frequently peaks to 0dB, it is likely the signal is clipping since it is probable that some of the
peak waveform is “going over the top”. Therefore, it is necessary to keep all signals peaking
below 0dB in a digital system.
There are further constraints on peak levels in digital audio. Any type of digital audio process-
ing can add peak overshoot. Digital effects like reverb or phasing add a lot of overshoot, as do
bit reduction schemes such as MP3. For example. transmission codecs used for ISDN audio
can easily add 6dB of overshoot. Therefore, it is wise to record and maintain digital audio
streams at lower than -6dBFS maximum peak at all times.
In an attempt at dealing with this problem, the Society of Motion Picture and Television
Engineers (SMPTE) has declared a standard of practice where 0VU is equal to -20dBFS.
Most of the world’s audio industry has accepted this standard in principle, but it is widely mis-
understood. The problem is the difference between peak and average level measurements,
and how reconcile them.
Page 34
In a world where all audio levels would be monitored by VU meters, this SMPTE standard
would make things simple. Since VU meters measure something close to the average level,
we could simply equate 0VU to -20dBFS with a calibration tone. We could mix and track on
the VU meters, knowing there is 20dB of headroom in the digital domain for peaks. Since
most audio has peaks that go 10 to 14dB over the average level, then we would still have 6 to
8 dB of digital headroom left over to allow for subsequent digital overshoots. However, most
of the digital gear is shipped only with peak responding dBFS meters, not VU meters.
When working only with dBFS meters, it is not possible to apply the SMPTE standard, and
therein lies the confusion.
Early in digital audio history, the manufacturers of DAT recorders taught everybody to record
at an average level of -18dBFS, but they forgot to tell you something important. They forgot
to tell you that the -18dB should be the actual average level, not the running average of the
peak levels. Unfortunately, the DAT machines only come with dBFS metering, and people
went around recording peak levels way down in the mud - at least 10dB too low for good
digital quality. That’s why so many DAT recordings sound like crap.
That’s not the worst of it. With no other sources of edification, recordists and engineers have
applied the same principle in other digital audio work. This has led to a serious problem in the
recording industry where tracks and mixes are ridiculously variable in level and quality. You
will find CD’s that were mastered with peaks slamming against 0dBFS and clipping all to hell.
You can also find CD’s mastered so peaks hardly ever exceed -8dBFS. These discrepancies
reveal the widespread problem of misunderstanding the technology.
If, after the average levels have been corrected by a Compellor, it is felt the PPM or dBFS
indications should be made more consistent, you can use an Aphex Dominator “Precision
Multiband Peak Limiter” after the Compellor. The Dominator will not act on the average
levels but will transparently control the peaks and bring the program closer to having consis-
tent peak levels without disturbing the average levels. The Compellor and Dominator are
designed to work together and no other peak limiter will perform with equal transpar-
ency to the sound quality.
Since the Compellor’s reference system is average, you will not be able to find a direct match-
ing REF LEVEL setting for a peak referenced system. However, remembering that averages
are usually 10 to 12 decibels below the peak level of typical sound, you can use the -10dBV
setting on the Compellor to get a reasonable match for peak references of about 0 to +6dBu
(-10dBV equals -7.8dBu).
An excellent way to set the Compellor’s REF LEVEL switch is to pass a signal through
the Compellor at standard levels and observe the input level meter on the Compellor.
The red bar part of the indication is similar to a VU indicator. Select a REF LEVEL
switch setting that brings the red bar closest to hitting 0VU.
Page 35
Digital–vs–Analog; Peak–vs–RMS
How To Deal With The Confusion
By Donn Werrbach • 10/03/03
The Confusion
The matter of audio level measurements and specifications can be very confusing at times.
That is because some specs relate to peak measurements and some to average or RMS
measurements. There is no one standard in use throughout the industry.
If all sound were nothing but pure simple sine waves (the most fundamental wave of nature),
the measurement of sound would be very simple. Measurements, whether peak weighted or
average weighted, would almost come out the same. A sine wave’s peak level is only 3dB
higher than it’s average level, and what’s more important, the ratio of peak to average, also
called the “crest factor”, is always the same no matter what the level is. Both peak and aver-
age level meters could be calibrated in the same relative units (like VU) and would read the
same.
However, since sound waves are complex, their peak to average ratio varies depending on
the sound characteristics, and that ratio can vary from 3dB to as much as 15dB. So, in the real
world, peak and average meters will disagree by as much as 15dB. The ear hears loudness
based on the power level contained in a sound wave. The power level is proportional to the
average signal level, so averaging meters will respond to level more like our hearing. Peak
measurement of audio cannot infer the volume level except with pure test tones because the
crest factor of program audio is large and variable, kicking the peak meter well above the
average measurement.
The problem is that we find both kinds of meters in use and they cannot be easily recon-
ciled.
Dueling Standards
An advantage of this method is that audio electronics need not be built with very much head-
Page 36
room above the maximum PPM indication. By controlling the audio levels to maintain good
PPM readings, there can be no possibility of the electronics clipping the audio. The disad-
vantage is that to maintain a good average volume level, it takes very clever people riding
the gain who can accurately guess at the crest factor of all the sounds. The BBC of the U.K.
has actually created standards on where to allow music, voices, and commercials to peak on
their own version of a PPM. It just seems so ridiculous when you consider they could all just
adopt the American VU standard.
The VU Standard
Throughout the United States and its sphere of influence in the world is traditionally found the
VU meter as specified by the ASA (American Standards Association, now extinct). The name
VU comes from “Volume Units”. It is the intent of VU meters to indicate the audio level as we
hear it. It does not indicate the peak levels of the audio.
In VU meter practice, audio electronics must be designed to have sufficient peak headroom to
allow safe passage of all unseen audio peaks. To allow this, at least 20dB of headroom above
the 0VU reference level is designed into professional equipment. However, the advantage is
that monitoring and controlling levels by VU indications yields pleasing consistency of sound
levels without any guessing about the crest factor.
Both Meters
Show 0VU
-20 VU VU
Shaded Area = -27
8
-20
Peak Meters
Very Different for Same Loudness
Figure F-1, Sine Wave and Complex Waveform of the Same Sound Level Compared on VU
and Peak Meters Demonstrating the Peak and Average Metering Differences
Page 37
Birth of dBFS
In the analog world, there can always be found a little more headroom. Magnetic tape has
a very soft and spongy nonlinear area above the maximum operating level. It compresses
peaks without hacking off the tops. Most other electronics have some headroom to spare. It
is seldom catastrophic, from a sonic perspective, when a few peaks hit analog clipping. In the
digital world, the same is not true.
Digital audio has a very hard peak ceiling that literally shaves off any and all excessive peaks.
That causes severe audible distortion and needs to be avoided. True also is that, when the
best digital audio had only 16 bits, it was readily discovered that the best sound came from
recording at the maximum level to capture all the digital quantization possible and stay out of
the low level grunge.
To assist with that cause, digital audio equipment makers disavowed the VU meter in favor of
a new kind of peak responding meter. After a few early experiments, the digital audio meters
have emerged with instant peak response (no peak integration like the PPM) with 0dB at the
very top of the scale. The new scale is called dBFS (dB referred to full scale). This allows you
to accurately see how your audio waves fit below the digital ceiling so you can avoid digital
clipping.
DAT Tragedy
That may seem all well and good considering the fact that 16-bit audio has such limitations.
But, now with 24-bit digital audio prevalent with its much greater useful dynamic range, the VU
meter has not been reintroduced and probably won’t be unless a stroke of luck knocks some
sense into somebody along the line. That is because of DAT machines.
Digital audio users get precious little technical training. What little there is comes from the
equipment’s user manuals. DAT machines were the first popular digital recording media.
Through the DAT manuals, users were taught to record the average levels at -18dB. OK, fine,
if that means you are to record the average levels at –18dBFS. That leaves 18dB for swells
of volume level and the host of variable peaks that may rise above by up to 14dB. The DAT
manuals forgot to tell you that, however, and it was wrongly interpreted to mean the recording
of average peaks should be at –18dBFS. That has unfortunately stuck as a general digital
audio practice that needs to be corrected. The Compellor Model 320D can truly help.
Page 38
Most simply stated, the Compellor will accept an audio input, digital or analog, level it out
and add some compression making it more consistent in average level. The resulting aver-
age output level will target around 0VU. That means –20dBFS in the digital audio world and
+4dBu (or –10dBV depending in the level reference settings) in the analog world. The digital
output will have peaks that may rise up to 0dBFS but will probably not consistently rise above
–8dBFS. That is because audio’s typical crest factor is 10 to 14 dB.
If the Compellor’s limiter is switched in, it will stop peaks at the –6dBFS level for digital audio,
or 14dB above the 0VU reference for analog signals.
If the digital audio input was previously held consistent on a dBFS meter, as if a peak limiter
had been used, it may not look as peak-consistent at the Compellor output. That is because
the Compellor acts to correct the average levels at the expense of letting the peaks fly where
they may. This should not deter you because you’re actually getting what you wanted. If you
want to also see solidly consistent peaks after the Compellor’s processing, then you can add
an Aphex Dominator multiband peak limiter. It will flawlessly bring the peaks to consistent
levels without affecting the average level first established by the Compellor. The Compellor-
Dominator pair is the best audio packaging system there is for effectiveness and sonic trans-
parency.
ISDN LINK
Page 39
outputs into the codec, the level will then shift and be louder with the analog input. That may
give the effect of a fuller on-air sound when the coder is driven by analog because the on-air
audio processor at the decoder side is driven with higher input level.
What, Me Worry?
This level mismatch need not be a problem, especially if you intend to use only the digital out-
put. Simply readjust the final audio processor to be optimal with the –20dBFS average digital
audio input level. If you want to also use the Compellor’s analog output for comparison or for
a backup plan, then you can readjust the coder’s analog input gain, if available, or add an
external attenuator to the coder’s analog input to get the –20dBFS digital reference conver-
sion. Reducing the Compellor’s analog output gain will not solve the problem because
it will also drop the digital level proportionately.
Summary
Increasingly, audio monitoring is a mixed bag. Some equipment with VU, some with PPM and
some with dBFS. Only if you understand what the meters show, will you be able to use them
properly. Being aware that VU meters indicate relative volume level without regard to peaks
and that PPM or dBFS peak meters indicate the available headroom below clipping without
regard to perceived loudness is most important. Secondly, it is important to know how to set
up an operating level. With these two bits of knowledge, you will be able to find a satisfactory
solution to all of your audio interfacing problems. Confusion: Gone!
Page 40