|
|
|
|
Revision Date:
250726
Information
contained on these web pages copyright W8KHK / N1BCG. Amateur or individual use
is encouraged.
Commercial use of
any kind is prohibited without the express written permission of the author,
Richard A. Maxwell
MAX Audio Processor – Technical Details
How It Works -
Circuit Description
Power
Supply:
We start with the power
supply, because without careful design in this area, performance of all
sections of the processor would be compromised.
The power supply is fully
linear in design, and no spurious noise producing
switching circuits or devices are employed.
The supply is transformer coupled to the mains,
providing full isolation, preventing shock hazards and ground loops. If an internal transformer is used, the secondary
provides approximately 24 volts AC, center tapped. The center tap is connected to the floating
ground on the main board, and the balanced legs of the secondary are tied to a
full wave diode bridge, providing full-wave rectification of the positive and
negative raw input power rails. In the
case of the external wall transformer, a single output of 12 volts AC is
applied between the floating ground and either one or both legs of the bridge,
thus forming a half wave voltage doubler, resulting in the unfiltered negative
and positive rail voltages to be filtered then regulated.
If an internal
transformer is employed, mains power enters through an IEC approved connector
with embedded EMI filter. The LINE (hot) input lead is fused for ½ ampere, fast
blow, and the primary current flow is then controlled by a front panel
switch. The two primary windings are
tied in parallel, fed directly from the NEUTRAL pin on the line filter and the LINE
output from the front panel power switch. The two secondary windings are tied
in series, with the virtual center tap connected to the grounded pins 2 and 3
of the AC input header, while the ends of the secondary are connected to pins 1
and 4 of the AC input header. This connection provides a power supply with a
pair of full-wave center-tapped supplies, one negative and one positive; the
rectifier actually configured as a bridge with the power transformer secondary center
tap referenced to the ground return.
If an external wall
transformer is used, 12 volts AC enters the back panel via an insulated
(isolated) barrel connector, controlled by the front panel switch, and then fed
to the AC input connector on the main processor board. The outer shell of the barrel connector is
wired to ground pins 2 and 3 of the AC input header, while the center pin from
the barrel connector is wired through the switch to pins 1 and 4 of the AC
input header. A fuse is not required when
the external wall transformer is employed.
This connection method provides balanced positive and negative supply raisl, implemented by two diodes in a standard voltage
doubler configuration. If the hot lead
of the transformer is connected to both legs of the bridge rectifier, then the
diodes are configured in a parallel, redundant circuit, even though there is no
attempt to balance the current load between the two parallel diodes in each
leg.
Each of the positive and
negative rails is filtered by a pair of 3300 uF
electrolytic capacitors, diode decoupled to further reduce any hum or noise.
Filtered unregulated rail
voltages are fed to negative LM-7905 and LM-7912 regulators, and to the
positive LM7805 and LM7812 regulators.
The regulator outputs are further filtered by 100 uF
electrolytic and .1 uF disk ceramic capacitors. Each regulator is protected by two diodes, one
diode from output to ground, preventing reverse voltage application, and one
diode from input to output, preventing the output from rising to a level which
exceeds the input voltage. While these
diodes are recommended per the 78xx regulator datasheet, they are absolutely
required In an application where the load floats on
the positive and negative rails. During
either power up or power down, without these protective diodes, the regulators
may be exposed to inappropriate voltage combinations, which causes a lockup
condition upon subsequent power-up cycles, rendering the processor inoperative.
Each modular section of
the processor is provided clean, isolated power through a series of decoupling
resistors and capacitors, and each IC device is bypassed by a .0.1 uF disc ceramic capacitor at every incoming power rail pin.
The decoupling and bypass
circuit components comply with the following naming conventions:
•
Bypass
capacitors will be labeled C12N and C12P, as associated with the negative and
positive supply rails for IC device U12
•
Series
decoupling resistors will be labeled R12N and R12P as related to the devices in
the modular section containing IC device U12
•
Decoupling
electrolytic filter capacitors will be labeled C12EN and C12EP as utilized by devices
in the modular section containing U12
•
Rather
than place these components on the page near the related IC, they are all shown
on the power supply page, to improve schematic clarity
•
C12,
or any other resistor, capacitor, or diode without the N, P, EN, or EP suffix
is related to the signal handling process, not power distribution
•
Devices
in the 300, or 3XX component number range are dedicated to the power supply
proper, and have no involvement with the active signal path
The above decoupling and
bypassing circuits perform the function of providing clean, steady power to
each modular section of the processor, and isolate sensitive sections from
circuits that demand a high level or pulsating current, thereby avoiding
spurious signals, feedback motorboating, etc.
All operational
amplifiers and the Analog Devices SSM-2019 mic preamplifier utilize the
balanced positive and negative 12-volt supplies.
Negative 12 volts is also
used in the dynamic compressor circuit.
Positive and negative
5-volt supplies are used by the Maxim MAX295 SCAF filter chips. These voltages are also used in the limiter
peak control circuits for LED illumination.
Positive 5-volt supplies
are used for the bandwidth Clock NE-555, the Level Indictor comparator LM-339,
and for all LED illumination in level indicator and dynamic compression hold
status LEDs.
No effort has been
expended to provide +48 volts phantom power for microphones. While it would be a trivial task to provide
this voltage, and while the microphone input circuit is compatible with such a
feature, the risk to expensive microphones, in the case a kit might be
assembled improperly, is too great to consider providing this function. For use with microphones that require phantom
power, it I suggested that a separate power adapter designed for that purpose be obtained and safely employed.
Overall power consumption
of the entire processor shall not exceed 7 to 10 watts, inclusive of the wall
transformer or the interior power transformer.
While all 78X and 79X regulators will have extruded aluminum heat sinks
attached, there is sufficient airflow to cool these devices without any
ventilation openings or a circulation fan.
Microphone
Input:
Low Level Balanced
Differential microphones may be connected to the microphone input via either
XLR female or TRS ¼” phone jacks. (The
microphone cable would obviously have a male connector at the end of the cable.) Sleeve or pin 1 is grounded, tip or pin 2 is
positive, and ring or pin 3 is negative.
XLR connectors should have pin one directly attached to the chassis
ground terminal of the connector. The
Processor board is grounded via metallic spacer sleeves and machine screws via
the rear two mounting holes. Attaching a
ground wire from pin 1 on the board to pin 1 on the input connector is purely
optional.
Integrated circuit U12
provides the microphone input amplification.
The input resistors R17 and R18, in association with bypass capacitors
C6 and C7 provide RFI ingress protection.
In extreme cases, two 1.0 mH inductors may be
substituted for these two resistors.C8 and C9 provide
DC isolation, in the event a device connected to the input applies an unwanted
DC bias voltage. R8 and R9 provide a
galvanic DC return path, as required by the input chip specifications. If a balanced microphone with a center ground, or a coupling transformer with a grounded center tap
is used, these resistors may be omitted. C5 is also an RFI suppression device,
installed as close to the input pins as possible, per Analog Devices
recommendation for the SSM-2019 preamplifier device.
Positive and negative 12
volt filtered and isolated supply voltages are applied to pins 7 and 4,
respectively. Offset for output is
defined as ground by connection to pin 5, and output signal is taken from pin 6
to either a level control trim pot or a berg jumper header at location
R12.
Gain for the mic
preamplifier is set by a resistor from pin 1 to pin 8. 32 ohms provides a gain of around 60 dB,
while an open circuit provides unity gain.
Other values of resistors, for various desired gain levels, may be found
in the SSM-2019 datasheet, available online.
The gain set jumper at header GN allows the user to change between
microphone gain sensitivity and unity gain by moving the jumper from pins 2 nd 3 to pins 1 and 2.
It must be noted that to maintain maximum CMR and minimum noise and hum,
the wires to the gain set resistor must be kept as short as possible. This precludes running an eight-inch pair of
wires from the jumper to a front panel switch.
The SSM-2019 provides
exceptionally clean, noise and distortion-free amplification for even the
lowest output level studio microphones.
If one desires to connect
a single ended (unbalanced) microphone to the balanced input, connect the
shield to pins 1 (GND) and 3 (NEG) and the center coax lead to pin 2
(POS). These numbers apply to both the
XLR female connector pins, and the input terminal pins on the main processor
board.
Note that while it is
common practice in high-end audio applications to ground the chassis or
enclosure at only one point, to the ground reference point defined as the
lowest level (highest sensitivity) point in the circuit, with the goal being
total elimination of ground loops; in an RFI-laden environment it is often
necessary to bond the chassis or enclosure to the circuit at MULTIPLE points. The
processor accomplishes the latter requirement, when necessary, by grounding the
circuit board to the chassis at the two rear mounting holes, and additional
ground connections may be made from the GND land near the input terminal pins
on the board, as well as pin 1 on mic and line inputs, as well as output 1 and
output 2 headers. In addition, the processor chassis or enclosure should be
grounded to the station ground with a substantial cable or strap, to a bolt and
wing nut combination on the back of the enclosure.
Line
Input:
Integrated Circuit U11
provides the line level input circuit, as well as the second stage
amplification for both microphone and line input functions. Similar to the microphone input, the line
input is fully balanced differential and exhibits high CMR performance. The stage is run at unity gain in order to accommodate
very high-level inputs without the need for an input attenuator to avoid first
stage clipping.
C1, C2, C3, C4, R17 and
R18 provide RFI and EMI immunity. In
extreme cases, when these components are insufficient in blocking stray RF,
then a pair of 1.0 mH RF chokes may be substituted
for R17 and R18. Thus far, with over eighty
MAX Audio Processors operating in the field, this extreme step to mitigate RFI
issues has never been required.
R3 defines the stage
input impedance, and prevents noise or instability if no input device is
connected. R4, R5, R6, and R7 provide
feedback for unity gain, balanced differential input. Decoupled, filtered, and bypassed positive
and negative 12 volt supplies are applied to pins 8 and 4, respectively.
Output is taken from pin
1 and applied to header R10, which may be populated with either a 10K ohm trim
pot or a berg jumper header.
Input
Level Controls:
The microphone and line
input level control layout is flexible, allowing several customization options.
First, the selection of
either a trim pot or a jumper at locations R12 and R10 allow either a permanent
preset input level via the trim pot, a coarse and vernier setting via the trim
pot and front panel level control, respectively, or just a jumper to pass the signal
on, unattenuated, to the next component.
The vernier front panel control could be set at a standard position, for
example, three o’clock, and the trim pots could be set such that the input
level requires no major adjustment when switching between the line and mic
input sources.
Second, the MAX Audio
Processor is offered with a variety of enclosure and front panel
combinations. The Front Panel Interface
PC board (FPI) provides interconnections to the input level controls and the
mic/line selector switch. For configurations
using a single front panel level control, the input signals from mic and line
preamplifiers flow to the mic/line switch, and the selected signal then is
attenuated by the front panel level control before being routed back to the
second op amp In the line input circuit, U11-2. Conversely, when two front panel input level
controls are present, the mic input feeds the top input level control, while
the line input feeds the bottom level control.
Either mic or line use is selected by the mic/line
switch, which receives input from both level controls, and the selected
signal is passed directly to U11-2. In
all cases, the trim pot or jumper at R10 and R12 control the signal level sent
to the controls on the front panel.
Op amp U11-2 is a
non-inverting, gain stage, with R13 and R14 setting the stage voltage gain to
10, or roughly 20 dB. The level attenuator preceding this stage prevents overload
with high level input signals, but also is capable of dealing with very low
input levels as well. At the input, C10
and R15 remove any residual DC bias that may be present in the signal from an
external device. In general, the strategy
is to minimize the number of coupling capacitors in the signal path, in order
to avoid low-frequency roll-off and minimize any cumulative phase shift in the
program material. C11 provides a bit of
high frequency roll-off and noise suppression.
The signal from this stage is passed directly to the phase rotator, and
is optionally delivered to the Audio Sculpting stage if the phase rotator is
disabled.
Phase Rotator:
The phase rotator, also known as an
all-pass filter, is employed in order to equalize or suppress asymmetric audio peak
energy, thus allowing the compressor to provide a much denseer
modulation envelope. Originally a passive,
LC phase shifting filter by Kahn, a simpler, more practical implementation leverages
the use of active components, resulting in a zero dB insertion loss component
in the audio chain. On this website’s landing page, a brief discussion of the
phase rotator function includes external links that reveal more of the history
and science of the all-pass filter application.
As implemented in this processor, a series
of eight poles (eight op amps in four chip packages) are connected in cascade
as unity gain inverting amplifiers, with the addition of an RC delay circuit
connecting the output of each pole to the non-inverting input of the following
stage. The net result is a phase shift
that increases in magnitude with increase in input signal frequency. While four stages, or poles, may be adequate,
the results of testing on-the-air with four, eight, twelve, and sixteen poles indicated
that eight poles were optimal, any more being overkill and adding unnecessary
distortion and noise.
The all-pass filter does not alter the
amplitude as frequency changes, nor does it alter the gain as the input level
changes. It simply alters the phase (delay)
of the incoming signal components with respect to frequency. While you would not desire the use of this type
of filter in a home high fidelity reproduction system, it has been proven that
the benefits in overcoming atmospheric noise and signal fading in broadcast and
other communications applications are well-worth any negligible, inaudible distortion
that might result from its employment.
The phase rotator in the
MAX Audio Processor may be enabled or disabled via the phase rotator toggle
switch on the front panel. Depending
upon the switch position, the signal from either the phase rotator, or from input
amplifier U11-2, respectively, is passed to the Audio Sculpting section of the
processor
Signal
Flow and Off-Page Connector Notation:
At first glance, five sheets
of schematic diagrams may seem a bit overwhelming. But taken as bits and pieces, in conjunction
with this descriptive dialogue, it is relatively easy to follow the signal
flow. While the power supply stands
alone on sheet five, the signal flows logically from sheet one through sheet
three, with the level indicator and dynamic processing control logic all
exposed on sheet four. “Off-Page Connector”
symbols, with descriptive legends, are used to guide the reader from one sheet to
the next. In addition, when a connector points
to a different page, there is usually a script near the connector, directing
the reader to the appropriate sheet.
Off-Page Connectors are
also used in an attempt to make the schematic sheets less cluttered, and
therefore more readable, in comparison to a large number of long parallel wire
lines that are difficult to trace. A
special application of the off-page connector is used to provide a logical
interconnection between the main processor board and the “Front Panel Interface”
PC board, to which all front panel controls and indicators are directly
soldered. This “front panel interface”
eliminates a myriad of hand-wired connections from the main board to each
switch, potentiometer, and LED on the panel.
As a result, the only hand wiring needed to complete the processor
interconnections are the input and output connector circuits to the back panel,
and the AC input wiring from either an internal or external power transformer.
An attempt has been made
to enable the off-page connector legends to be self-descriptive, such that,
after one understands the legend strategy, it is unnecessary to follow the
connector to the related sheet, but instead one may visualize the related
component or circuit based upon the abbreviated description in the legend. While the front panel interface schematic is
documented on yet another separate sheet, the descriptive legends are intended
to eliminate the need to look at that page when studying or analyzing each
section of the processor’s circuitry.
The legends not only define the related component, but also the
functional pin connection to that component.
Examples of some connectors are:
PRSW1 is the phase rotator
switch, pin 1; DENSW2 is the compression density switch, pin 2; LOCW is the low
gain potentiometer clockwise pin, and “visualize” PEV is the pre-emphasis
potentiometer variable pin.
Here are some representative
examples of the switch and potentiometer pin notations:
•
DPCsw1
is the Dynamic Processing Control switch pin 1; pin 1 is always the Common switch
pin
•
DPCsw2
is the Dynamic Processing Control switch pin 2; pin 2 is always the Normally
Closed switch pin
•
DPCsw3
is the Dynamic Processing Control switch pin 3; pin 3 is always the Normally Open
switch pin
•
The
“DPC” prefix identifies explicitly the switch in question
•
PEccw is the Pre-Emphasis potentiometer counter clockwise
terminal
•
PEv is the Pre-Emphasis potentiometer variable terminal
•
PEcw is the Pre-Emphasis potentiometer clockwise terminal
•
When
looking at the three potentiometer connectors on the schematic sheet, one can
easily visualize the control device connections
Beyond the common
notations listed above, an effort has been made to standardize on these
notations with regard to the 40-pin interface connector attaching main to front
panel interface circuits. The same
legend will be used in the main circuit proper, the main board connector, the
front panel interface connector, and the front panel component. Thus it is
unnecessary to inspect the interface connector legends, but instead the viewer
may go directly from the circuit in question to visualize the remotely located
component pin, without ever leaving the main circuit sheet.
Audio
Sculpting:
Prior to the absolute
latest version of the MAX Audio Processor, three switches (Low Cut, Low Boost,
and Pre-Emphasis) were provided to make the desired adjustments to the audio
response curve. There were repeated
requests for either a full internal graphic equalizer circuit, or support for
an external EQ box within the signal path.
All suggestions were carefully considered. Internal EQ implementation was deemed impractical,
as it would exceed the acceptable level of assembly difficulty, as well as
escalate the cost of components, and exceed available circuit board real
estate, as well as front panel capacity.
Issues related to connecting external EQ devices include interfacing
without risk of ground loops, RFI ingress, incompatible signal level, etc. Instead of either of the impractical options
described above, the decision was made to implement an equalization section
specifically designed for the needs of voice communication using Amplitude
Modulation on the amateur HF bands. The “Audio
Sculpting” feature was the end result of the study and several prototyping
efforts.
The Audio Sculpting
section employs two op-amps in one device, U21-1 and U21.2. It is designed to provide unity gain, such
that it may be inserted or removed from the signal path with no apparent level
shift. Jumper header EQ may be used to
either enable (pin 2 to pin 3) or disable (pin 1 to pin 2) Audio Sculpting,
completely bypassing this section of the circuit. Both U21-1 and U21-2 function
as inverting amplifiers; the two in cascade restore the original polarity of
the incoming program material. The first
of these op amps runs at unity-gain, while the second provides make-up gain for
the passive components. Capacitor C21 and
C29 achieve attenuation of unnecessary high frequency signals, maintaining
overall stability.
R30, R31, R32, and C28,
in conjunction with the Low Gain front panel control, provide both reduction
and enhancement of the lower voice frequencies.
The circuit is similar to a Baxandall bass control, in that the control
potentiometer receives both a positive and a negative feedback signal on
opposite ends of the variable resistor, providing a flat low frequency response
near the center of rotation.
R24, R25, C23 and C24, in
conjunction with the Pre-Emphasis front panel control, provide an increased
boost in higher audio frequencies as the control is rotated clockwise. Because the CCW end of the potentiometer is
returned to ground potential instead of the negative feedback signal, this
control can boost, but it cannot cut the high frequency program content. Typical pre-emphasis settings are near the three-o-clock
position of the control. R33 and C30
tame the increasing high frequency response above the range desirable for
transmission, and additional high frequency attenuation is afforded later in
the signal path by the SCAF (Switched Capacitor Audio Filter).
The components around the
Null front panel control implement a notch filter, such that response is flat
when the control is rotated fully clockwise, due to ground potential being applied
to the CW pot terminal. As the control
is rotated counter-clockwise, a null of increasing depth is created in the
frequency range near 280 to 300 Hz, providing a deep “mud cleansing” filter for
the vocal range of frequencies that dominate the modulation envelope without
providing any increase in intelligibility.
The null is narrow enough that the overall low frequency components of the
speaker’s voice are not expunged.
Level
Indicator:
Several LEDs are used to
convey various input content level parameters.
Positive or negative dominant polarity of the asymmetric program material
is indicated by the blue and green LEDs to the left of the panel.
The signal at the input to
the Audio Sculpting section is passed through capacitor C151, which is charged
to the input DC offset level by resistor R151, thus removing any bias offset
that would render the dominant polarity indication invalid. The signal is then passed through a dual op
amp, both stages at unity gain, the first non-inverting, and the output of the
first is inverted by the second stage.
The balanced audio program
material is then rectified by a series of diodes, D151, D152, D153, and
D154. The outputs of the first two
diodes are filtered and the time constant is controlled by R156, C152, R157,
and C153, resulting in similar DC levels when the program content is primarily
symmetrical, however when it becomes significantly asymmetric, the voltage
across C152 will be greater when positive peaks dominate, and the C153
potential will be in excess when negative peaks dominate.
These two varying DC levels
are applied to Comparator U152 channels 3 and 4, with the polarity of the
signals reversed as applied to the latter comparator. Output 3 is normally high when positive peaks
do not dominate. When positive peaks
reign, output 3 is pulled low, forward biasing the emitter-base junction of
transistor Q153, and the resulting collector current causes the green positive
peak LED to illuminate. The same action
takes place on output 4 when negative peaks dominate, causing Q154 to conduct,
illuminating the blue negative peak LED.
In order to indicate the
overall signal level, diodes D153 and D154 rectify both the positive and
negative program content, and the resultant DC voltage charges C164, and the
discharge time constant is set by shunt resistor R161. This varying DC level is
passed to comparator channels 1 and 2, such that the required normal and peak
input levels may be determined. A
voltage divider chain, sourced with positive 5-volts DC, consisting of
resistors R158, R159, and R160 provide the necessary reference voltages for the
normal and peak level comparator function.
With a total resistance of 59.2K, sampling the “normal” reference level across
the 2.2K resistor, provides a voltage of 2.2 / 59.2 = 0.03716 * 5 volts = 0.1858
volts, or 185.8 millivolts, as the normal input level. For the peak level, the reverence is taken across
the combination of the 10K and the 2.2K resistors, for a total of 12.2K / 59.2K
= 0.20608 * 5 volts = 1.0304 volts, or 1030.4 millivolts. When the sampled DC level at C154 reaches or
exceeds the normal or peak fixed reference levels applied to comparator channels
1 and 2, the channel 2 and 1 outputs transition from high to low, turning on transistors
Q151 and Q152, thus illuminating the red peak LED and the yellow normal LEDs,
respectively.
The various resistor
values for R169, R170, R171, and R172 are chosen to limit the current through the
various LEDs, because the different color LEDs require
varying current levels to produce a similar brilliance when illuminated.
Earlier revisions of the
processor exhibited random blinking of the dominant polarity LEDs, even when no
signal, or very small signals were present. This is due to the narrow threshold of the
LM-339 comparator device. In order to suppress
the random blinking, the supply to the transistors controlling the dominant
polarity LEDS (Q153 and Q154) the supply to these two transistors is gated by
the signal from the normal LED switching transistor, Q152. This circuit change prevents the dominant
polarity LEDs from being illuminated when the input voice level is below the
minimum, or normal, volume level. In
retrospect, it would seem illogical to allow these indicators to illuminate
with a complete absence of input signal
The input level control
should be adjusted such that, under normal voice conditions, the yellow normal
LED is illuminated almost all the time during speech, while the red peak LED
should be illuminated occasionally to quite often. The reference voltage divider described
earlier sets the comparator reference levels such that no adjustment or
calibration is required for the level indicator, insuring the compressor
receives the audio program signal level within the proper dynamic range to
achieve the desired modulation density without overdriving its gain reduction
capability.
The final feature of the
level indicator circuit enables the novel “Dynamic Processing” feature of the MAX Audio Processor. While most generic compressors will reduce
gain when the input signal is very large, and increase the gain when the input
signal is small, this basic scheme results in an undesirable output level
increase when there is no input signal whatsoever. Some compressors employ downward expansion or
noise gate circuits to deal with this situation but these create undesirable
artifacts during speech pauses, and some even totally block the input signal,
creating dead silence during these pauses.
In order to preserve the natural sound of speech, even during momentary
pauses, it would be desirable to avoid a gain increase when there is no voice
present, but then continue to adjust the compressor gain upward or downward
when speech is actually present. The
goal here is to provide a consistent, dense modulation envelope, but avoid the
sudden level jumps, both up and down, when speech pauses are encountered. To accomplish this goal, the compressor must
be placed in “limit only” mode when there is no speech, (gain reduction only) but
enable full compression, upward or downward, when speech is present.
Considering we already
have a method to detect whether speech is present (the illumination of the yellow
normal LED) we simply need to gate that signal to the compressor section of the
processor, such that the behavior of the compressor may be altered on-the-fly
when speech is interrupted or continues.
It would be nice to allow the operator to choose whether this feature is
functioning, so we take the active low normal LED activation signal at the gate
of Q152 and pass it through a front panel switch, before submitting that
control signal to the hold circuit.
The switch-enabled “speech
present” signal is routed from the switch to the base of NPN transistor
Q155. Configured as an emitter follower,
when the input signal is low, the transistor does not conduct, therefore there
is no output at the emitter. This
condition exists when the switch is enabled and voice is present, or when the
switch is disabled (open) and resistor R163 maintains the base of Q155
low. At times when the dynamic
processing switch is enabled, and voice is NOT present, the signal at the base
of Q152 goes high; this same signal is passed to the base of Q155 via the
dynamic processing switch, so in this condition, Q155 conducts, providing a
high-level voltage at the emitter. This
voltage illuminates the blue hold LED, indicating voice is not present, and the
compressor is in hold mode, (limit only, no compression) so it will only
respond to louder program material, thus reducing gain, but will not increase
gain in the hold mode. Resistor R173 limits the LED current, controlling
brilliance. Capacitor C155 extends the illumination
time of the LED, such that very short periods during voice absence will visibly
illuminate the LED. Diode D155 prevents
the charge on capacitor C155 from erroneously extending the hold time of the
compressor dynamic processing switching mode.
Capacitor C156 provides a
bit of delay, such that the hold signal does not toggle off between
closely-spaced syllables, while resistor R163 ensures C156 gets discharged when
voice is present. Resistor R164 limits
the base current on Q155, as it is not necessary or appropriate to apply the
full 5-volt supply potential to turn on the base current. In addition to illuminating the blue hold
LED, the Q155 emitter output voltage is passed to the compressor section to
enable toggling between full compression and limit only mode. The action of the compressor with this
switching voltage will be described in detail in the next section.
Dynamic
Processing Compressor:
The compressor section of
the MAX Audio Processor is detailed on schematic sheet two. At first glance, it seems rather complicated,
but after tracing the signal flow, it will appear ultimately simplistic. In order to preserve the quality of the input
audio as much as possible, the program material is subjected to as few components
as possible, including one resistor, one FET gain reduction transistor, and a single
non-inverting op amp with a gain of 10.
All other components in the compressor circuit are involved with
controlling the gain reduction bias signal applied to the FET, but in no way
affect the signal path or audio quality.
In a nutshell, compression,
or output leveling, is accomplished by a 20K series resistor R101, a 2N3819
transistor (Q101), performing as a variable shunt resistor, this combination attenuating
the input signal which is then amplified by op amp U101-1. The gain of this op
amp is set by 1K R102 and 10K R103, with a capacitor in parallel (C101) to suppress
high frequency and supersonic amplification.
There were some considerations that it might also be necessary to take a
secondary step in suppressing higher frequency signals with shunt capacitor
C106, but as yet that has not been necessary.
With the above component
values, the output level of the compressor averages 0 dBm referenced to a 600-ohm
load. This is in fact the voltage
defined as the basic ”system level” for input and
output to and from each interconnected
stage of the compressor. With a
selected, high-quality FET, the input level may range from -18 dBm to around +
5 dBm, while the compressor output is maintained within 2 to 3 dB of the
desired system level.
The remaining circuits in
the compressor section control the FET gate bias, thus maintaining the desired
leveled output program material. The compressor
audio output, taken from U101-1 pin 1 is passed to non-inverting, unity gain op
am U101-2, which provides isolation, such that nothing in the control circuitry
may alter or corrupt the compressor audio output signal. Output of U101-2 is fed to U102.1, configured
as a unity gain inverting amplifier. (U102-2
is employed elsewhere in the peak limiter section of the processor.) In combination, the balanced differential
outputs of these two op amps are rectified by diode D101 and optionally diode D102.
Creating the raw gain reduction negative DC bias voltage. This negative voltage increases in amplitude
as the program material level increases, but considering the compressor runs in
a closed loop, and the gain reduction signal is derived from the compressor’s stable
output, the gain reduction level, in normal operation, does not incur large
excursions
To obtain the densest
modulation envelope for AM operation, it has been determined that it is optimal
to simply rectify the negative-going program material, and ignore the positive-going
material when creating the gain reduction control voltage. Asymmetric positive
peaks may be dealt with later in the peak control section). For FM and some other modes, it is preferable
to reduce gain on both polarities of the input material, providing a more
symmetric modulation signal to the transmitter.
For this purpose, jumper header LDR is provided to select either half
wave or full wave rectification of the program material.
In order to reduce
compressor output level (gain reduction) the control FET Q101 must have a
positive-going change in potential. The bias on the FET gate is always
negative, so another way to state the required change is to say that to reduce
gain, the gate bias must become less negative, causing a lower impedance
between the source and drain terminals. It is obvious the raw gain reduction
bias polarity is incorrect to achieve the desired result. The raw gain reduction signal is applied to
the base of PNP transistor Q103, via a voltage divider network of 4.7K R108 (or
optionally a 10K trim pot) and 47K resistor R109. A trim pot at R108 allows control of the
effectiveness of the bias voltage change, but this adjustment has been found to
be unnecessary in practice. When no
negative bias is applied to Q103 base, it does not conduct. Thus the voltage divider from the regulated
negative 12 volt supply to ground, consisting of R115 and R116, sets the
quiescent bias level at the collector of Q103.
When program material is present, the raw bias on the base of Q103
becomes negative, causing the device to conduct, increasing the current through
resistor R115, resulting in a decrease in the negative voltage at the
collector. We now have a control voltage
of the correct polarity to apply to the gate of FET Q101, such that gain will
be reduced by increased attenuation. We
may also adjust the quiescent bias by changing the value of R115. But this adjustment should not be necessary,
as the current process includes the prequalification testing of all FETs to be
used at Q101, thus assuring the proper dynamic range and output level from the
compressor.
The bias could be applied
to the FET gate through isolation resistor R112, and with either
C104, C105, or both in parallel to provide a reasonable attack and decay
time constant, we now have a functional compressor circuit. Unfortunately, it
would behave as any other typical compressor, increasing the gain during voice
pauses, such that background noise is increased. Then upon the next syllable uttered,
overmodulation or limiting would occur, and the compressor must at that time
reduce the gain back to the appropriate attenuation level. There must be a better way. Continue reading about the “Hold Function”.
Hold
Function:
If you have not yet read
the section on the Level Indicator, consider studying that narrative before
continuing. The control signal from the
level indicator is necessary to complete the “Dynamic Processing” functionality
unique to the MAX Audio Processor. The
moniker “Dynamic Processing”, in this context, simply means that the compressor
mode toggles instantaneously between “full compression” and “limit-only” mode.
While in full compression mode, the gain (or reduction), may increase or
decrease at any time, based upon the input signal and the resultant gain
control bias voltage. Limit-only mode
allows the compressor to reduce gain, or increase reduction, anytime a stronger
signal is presented to the compressor input, but it will not allow an increase
in gain (or a decrease in gain reduction) when in limit only, or “hold”
mode. This capability is accomplished through
bidirectional gating of the control signal, such that it only reaches the FET
gate during limit-only mode when a level decrease, or gain reduction, is
commanded by the bias voltage.
The required
bidirectional gating is accomplished by the combination of diode D103 and
transistor Q102. The direction of D103
is such that anytime the generated control voltage at Q103 collector becomes “less
negative”, this change in potential is coupled to the FET
gate, as well as the time constant capacitors C104 and C105. The new bias potential is “remembered” by the
charge on the density control time constant capacitors, C104 and/or C105. When the input audio signal drops, such as
when speech is not present, the negative DC voltage at the base of Q103 is
reduced; Q103 conducts less, causing the collector voltage to become more
negative. If this change is allowed to
pass to the gate of FET Q101, gain reduction would be less, in other words
compressor gain would be more, raising background noise.
But the increased
negative potential at the collector of Q103 will not be coupled to the FET gate
by diode D103, as it is in an open circuit state when the junction is reverse
biased. In order to increase the negative
bias at the gate of Q101, the change must be coupled by transistor Q102. And under normal circumstances, IE Full
Compression mode, this transistor is biased to the ON state by the negative 12-volt
supply rail by resistor R114. When the
dynamic processing switch is OFF, as described in the Level Indicator section,
the HOLD signal is not produced by NPN cathode follower Q155 on sheet
four. But when the Dynamic Processing
switch is turned ON, and no voice is present, with the blue hold LED
illuminated, this positive control signal from the cathode of transistor Q155 toggles
the right end of resistor R113 in the compressor circuit to positive 5 volts
DC. This counteracts the negative bias
to the base of Q102, causing the transistor to go “open circuit” or high impedance,
preventing the ability of the negative increase in the gain reduction signal
from increasing the negative bias on the gate of FET Q101, In hold mode, the
FET base remains at the potential stored in either C104 and/or C105, magically “remembering”
the compression level relevant to the amplitude of the last spoken
syllable.
The compressor circuit
provides three different levels of “density”, from a Dense, aggressive, fast-acting
compression, through a medium, conversational density, to a fully open, or very
relaxed, slow time constant density. These
three choices are enabled by three different capacitor connection configurations. Dense mode uses just the smallest, 22 uF capacitor C104 to control the faster, more
aggressive time constant. Open mode
connects both the 22 uF C104 and the 100 uF C105 in parallel, for the longest, most relaxed time
constant. To enable the medium time
constant, the 22 uF capacitor remains in circuit,
while the 100 uF capacitor is connected to ground via
a 4.7K resistor, buffering its ability to constrain changes in the gain
reduction bias voltage.
SCAF
Bandwidth Filter:
The SCAF (Switched
Capacitor Audio Filter) consists of two Maxim MAX295 chips in cascade,
providing a very sharp cutoff brick wall filter, at the bandwidth chosen by the
operator. Bandwidth is determined by the
clock frequency applied to the Maxim chip(s).
A single Maxim chip installation would require a clock frequency of 50
times the desired audio response, which is one half the occupied bandwidth of
that signal when operating with Amplitude Modulation. With two Maxim devices in cascade, the clock frequency
is somewhat less than 50 times the desired audio response.
The entire SCAF section
of the processor may be enabled or disabled (bypassed) by the setting of the
jumper at header SCAF. Place the jumper
between pins 1 and 2 to bypass the signal from the Compressor section directly
to the Balanced Differential Output section.
Place the jumper from pin 2 to pin 3 to insert the SCAF into the signal
path, enabling the desired tight bandwidth control.
The NE-555 Clock Section:
An NE-555 chip is employed
to generate the required clock square wave.
The frequency of the clock is determined by a resistor and a capacitor
time constant. Either a variable
resistor, switch selected fixed resistors, or selectable trim pots may be used
to provide various audio response bandwidths.
This processor leverages the latter, providing three discrete, user
selectable bandwidths, initially set via internal trim pots, allowing any one
of the three bandwidths to be selected on the fly by a front panel toggle
switch.
The entire SCAF bandwidth
filter section may be enabled or disabled (bypassed) by placement of the jumper
at header SCAF. Place the jumper from pin
1 to pin 2 to bypass the SCAF circuit, or place the jumper from pin 2 to pin 3
to insert the SCAF filter into the signal path, enabling bandwidth management. The SCAF filter section is positioned between
the dynamic processing compressor and the absolute peak limiter stages.
Referring the the processor schematic sheet two, the NE-555 clock chip is
powered from a separately isolated and filtered positive 5-volt supply,
preventing clock noise from infiltrating other sensitive circuits in the
processor. The time constant, and therefore the square wave clock frequency, is
set by capacitor C123, at 1500 pF, and the combination of trim pots R123, R124,
and R125. The clock may be adjusted to
provide an audio response bandwidth from around 2 KHz up to approximately 10
KHz. Typical settings for the three
switch positions might be 3 KHz, 5 KHz, and 7 KHz, or possibly even 4 KHz, 6
KHz, and 8 KHz, depending upon operator preference, band conditions, and
receiver/transmitter technical limitations.
The total resistance of each of the trim pots has been chosen to permit
accurate calibration of each bandwidth setting, as each sequential trim pot is
expected to manage a somewhat higher clock frequency.
In order to allow three
bandwidth selections via a single toggle switch, a SPDT switch with a
center-off position is used. The narrowest bandwidth requires the lowest clock
frequency, and therefore the highest resistance for the time constant. For the narrow setting, trim pot R123 is set
to the desired clock rate, with the switch in the center-off position. R123 is ALWAYS in circuit, and it is in
parallel with either R124 or R125 when the middle or wide bandwidths,
respectively, are selected by the toggle switch. When calibrating the clock frequency, the
narrow bandwidth must be set first, then either of the
middle or wide frequencies are subsequently adjusted. If the narrow clock is changed, the middle
and wide clocks of course must be recalibrated.
The operator may choose
to substitute a variable resistor for one of the trim pots, thus providing two
fixed settings, and one continuously variable bandwidth setting. In that case, care must be exercised such
that the pot resistance does not venture past the range of the NE-555 chip’s
ability to oscillate. If there is no
clock presented to the Maxim MAX295 devices, there will be no output signal. In order to extend the operational clock
frequency of the NE-555 to the maximum practical value, bootstrap resistor R121
and capacitor C121 provide some feedback from the clock output to the trigger
input at the highest selected frequencies.
The Maxim MAX295 SCAF Section:
Two Maxim MAX295 Butterworth
SCAF filter chips are used in cascade to provide the optimum sharp slope brick
wall filter to contain the modulation envelope within the desired occupied
bandwidth, as determined by the applied clock frequency. While the 295 Butterworth device
appears to be preferable in this application, it is also possible to employ the
MAX296 Elliptical SCAF device with no circuit changes. The clock signal from the NE-555 output pin 3
is applied directly to the clock input of the MAX295 chips at pin 1, A test point is also provided, such that the clock may be measured
by a frequency counter, oscilloscope, or DSO.
In order to avoid loading the signal with the test cable capacitance, a
10K ohm resistor should be placed in series between the test point and the
probe tip.
Referring to the SCAF
section of schematic sheet 5, we can trace the signal flow from pin 1 and back
to pin 3 of the SCAF enable/disable header.
The SCAF chips not only include the filter, but each also has a general-purpose
op amp, which may be configured only in inverting mode. The gain of the op amps may be set via the
external feedback resistors. The SCAF
sections must be capacitively coupled to the remainder of the processor, in order to avoid undesirable DC-offset as the
signal is passed on to the peak limiter and output amplifier stages.
The input signal enters
the op amp section of U141 via resistor network R141 and R142, setting the gain
at unity. The output of the op amp is
capacitively coupled to the SCAF input by C143. The output of the first SCAF is
coupled directly to the input of the second SCAF, U142. The output of the second
SCAF is then coupled to the input of the embedded op amp. Now that the signal has traversed two inverting
op amps, the polarity of the signal is returned to match that of the original
input signal.
Because the output of the
SCAF may not be totally free of any clock signal content, this output signal is
passed through a low pass filter consisting of R147 and C144. This filtered audio content is now passed to
the input of the embedded op amp via resistor network R145 and R146. Input level to the SCAF is maintained at the
processor “System Level” of approximately 0 dBm into 600 ohms, which is roughly
0.774 volts RMS, or 774 millivolts.
Based upon the controlled input level, the values of the gain setting feedback
resistors R145 and R146 were carefully calculated to provide the correct input
level to the peak limiter, ensuring the limiting effect is optimal at all
times, with no calibration required.
During extreme contest conditions, if harder limiting is desired, it is
simply a matter of setting the compressor audio input level slightly higher,
driving the compressor at the higher end of its dynamic range. The harder limiting will be evidenced by
somewhat brighter flickering of the two red Peak Control LEDs.
In order to minimize the
number of required coupling capacitors in the chain of processor modules, the
output DC blocking capacitor, C142, is placed AFTER the SCAF enable/disable header,
such that a single capacitor may serve to block DC from entering the following
peak limiter, whether or not the SCAF section happens to be enabled. Minimizing unnecessary coupling capacitors
reduces undesirable tilt and low frequency attenuation.
The SCAF section is
powered by a dedicated decoupling circuit for both + 5-volts and – 5-volts. Each chip includes 0.1 uF
bypass capacitors on the input power pins.
Peak
Limiter:
The peak limiter consists
of a resistor-diode network, followed by an op amp gain recovery stage, with
minimal low pass filtering implemented in the feedback circuit. LED indicators are included to display the
relative instantaneous limiter performance.
The peak limiter is one of the few circuits in the processor that includes
no jumper option to bypass the limiting function.
The primary positive peak
limiter consists of series resistor R131, and totem pole shunt diodes comprised
of transistors Q131 and Q132. When the
incoming signal positive peaks exceed the conduction threshold level of the
combined transistor junctions, the conduction causes a voltage drop in the
series resistor, gently rolling off the peak to the desired limit. Q132 includes a LED in the collector circuit,
which illuminates during limiting when the emitter-base junction is forward
biased by the program content peaks. A series
limiting resistor, R133, protects the LED from overcurrent, and sets the
maximum brilliance.
Negative peak limiting is
essentially identical to the positive peak-limiting circuit, however, it
employs PNP transistors Q133 and Q134 to act on the opposite polarity program
peaks. The collector of Q134 is
connected to the negative peak control LED through current limiting resistor
R134. Positive 5-volts is provided to
the positive peak control LED, and negative 5-volts is provided to the negative
peak control LED, via dedicated power rail pins within the 40-pin front panel
interface connector.
In order to provide both
symmetric and asymmetric peak limiting options, a front panel toggle switch connects
the variable wiper of trim pot R130 to ground when in +100% symmetrical limiting
is selected. The switch allows the trim
pot wiper terminal to float when asymmetric peak limiting is desired. Trim pot R130 and resistor R135 provide an
adjustable positive bias to the bottom of the limiting transistor string, such
that clockwise rotation of the trim pot allows ever increasing positive peak
percentage. Upon initial assembly, the
initial setting of this trim pot is fully counter-clockwise. It may then be adjusted upward while the unit
is in operation, and the positive peaks are evaluated via a modulation monitor.
In the event an even
tighter control of negative peak content is desired, the secondary negative
peak limiter section may be enabled. If
an undesirable negative peak escapes through the first negative peak limiter,
R132 and Q135 may be configured to put an even tighter rein on the negative
peak program content. This limiter
functions identically to the primary negative peak limiter, with the exception
that the bottom of the limiter circuit is tied to a variable negative bias, initially
set to maximum via fully clockwise rotation of trim pot R140. This high negative bis on the emitter of Q135
prevents conduction, as there is insufficient voltage differential to reach the
required junction threshold potential.
Again, using a modulation monitor, the modulation envelope may be raised
very close to negative peak overmodulation, then the
bias provided by trim pot R140 is reduced to the point where secondary negative
peak limiting is observed. This
completes the calibration of the secondary negative peak limiter section.
The limited program
material is applied to the non-inverting input of op amp U102-2, which is the
second half of the second op amp in the dynamic processing compressor section. Resistors R136 and R137 provide feedback to operate
at unity gain, and capacitor C133 provides a low pass filter function to remove
any slight harmonic distortion that might be generated by excessive peak
clipping. The output signal from the
peak limiter section is passed on to the output level control circuits.
Output
Level Controls:
The output level controls
are configured a bit differently than the input level controls and jumper
settings. For the Lite version, there is
one output channel, with the level control trim pot header at R181. The Full 540 main board includes two complete
low impedance balanced differential output circuits, where the second output level
trim pot header is located at R201. The
Lite board supports front panel circuits that include just one output level
control, while the Full 540 board supports panels with either one or two output
level controls. While the input trim
pots and panel input level controls are connected in cascade, which requires
both locations to be populated, the output level trim pots are directly in
parallel with the front panel control or controls. If there is a front panel level control for
the first output, then R181 header MUST NOT have a trim pot installed. If only
one output level control exists on the panel, then the second output level may
be controlled by a trim pot at header R201. For any outputs that have a panel
output control, the corresponding trim pot header must be left empty, or a male
berg pin assembly may be installed at that location to provide convenient test
points.
In the event the builder
has a single output level control on the panel, for example, the 1U rack model,
and it is desired to control the output level of both output amplifiers from
one front panel control, then berg male pins should be installed at both r181
and R201. Then a female-to-female Dupont
wire may be connected from pin 2 (center pin) of header 181, to pin 2 of header
201, slaving both amplifiers to the output of the single front panel output
level control.
Balanced
Differential Output Amplifier:
All MAX Audio Processors
provide a low impedance, balanced differential output,
which may be interfaced to either balanced or unbalanced device input circuits.
The Lite 540 version of the processor contains a single differential channel,
while the full 540 version provides a pair of discrete output circuits to drive
multiple transmitters or other devices, as desired. The output level may be adjusted
up to as high as +17 to +18 dBm, referenced to a load impedance of 600
ohms. The source impedance is on the
order of around 100 ohms.
The design of the circuit
is such that it may be loaded rather heavily, either by resistive or reactive
loads, and long cables, while still providing a distortion-free signal. While not recommended, on the occasion that
one leg of the differential output is shorted, no damage will occur, and the alternate
output leg will continue to perform without no ill
effects. A low impedance balanced
differential output is realized by a pair of op amp in a single chip, one
configured for signal inversion, while the other operates in non-inverting
mode. Several circuit features improve
stability under unusual loads and abuse.
The inverting stage runs at unity gain, as set by the matched resistors
R182 and R183. The non-inverting input
of the inverting device is tied to ground via a resistor of half the value of the
feedback resistors, as recommended by Analog Devices engineering notes.
To provide high frequency
stability, 33 pF capacitors are connected between ground and the non-inverting
inputs, as well as from output to the inverting inputs. For maximum stability, resistive feedback,
rather than direct connect feedback, is preferred for the non-inverting
stages. This configuration precludes a
unity gain non inverting stage. In order
to provide equal amplitude transitions on both the positive and negative outputs,
it is necessary to provide a 2 to 1 signal attenuation
on the non-inverting stage to counteract the gain of 2 in the feedback network. This is accomplished via resistors R185 and
R186. In order to prevent damage with
output overload, and to define the output impedance, 47-ohm resistors R189 nd R190 are placed in series with the output rails.
The output terminals on
the board consist of three pins: ground, positive, and negative, which comply
one-for-one with the standard XLR audio connector pin assignments.
MAX Audio Processor – Main Circuit Board Revision History
Thank you for your interest in the MAX Audio Processor. All processor kits include silk screened and
solder masked main processor printed circuit board, the front panel interface
printed circuit board (which eliminates hand wiring to the various front panel
LEDs, switches, and pots) and a printed circuit dress panel providing a
background (solder mask) color of black for the 1U Rack versions, and blue for
all desktop versions. They are enhanced
with white legends indicating the function of each control and indicator.
The MAX Audio Processor,
Series 500, has been available in several different models. Each model is provided as a kit of parts, with
printed circuit boards and documentation, everything needed except an enclosure
and power transformer. These two items
must be selected separately, or obtained locally by the builder. Available kit
models are 500, 510, 520, 530, 540 and 540L (Lite); the various features are
described below:
All 500-series kits
include all the features of the previous model 500, but each adds some
incremental feature updates. Version 500
was the first to eliminate hand wiring of the front panel controls, introducing
a new “front panel interface” printed circuit board that completed all the
panel interconnections, using pin connectors and prefabricated ribbon cables
between the front panel board and the main processor board.
Model 510 also employed
the ribbon cable interface, and all other features of model 500, except the output
attenuator pot for microphone input compatibility on the newer transceivers. Unveiled with revision 510 is an enhanced
peak limiter design, with the asymmetric “greater than 100%” positive peak
limiter being adjustable from 100% to approximately 135%. There is also an adjustable second-stage
negative peak limiter, to be employed when absolutely no zero-crossings are
permitted in the negative direction. The
high-impedance utility output was replaced by a second discrete balanced
differential output driver, allowing multiple transmitters to be connected and
adjusted, thus enabling much more convenient QSY to different bands or transmitting
equipment.
Model 520 provides a
feature set identical to model 510, but offers a much-improved method of
interconnecting the front panel controls.
Instead of several flat ribbon cables, a row of pins on the main
processor board connects to a female header at the bottom of the front panel
interface PC board, directly completing all the required connections; no
cabling needed.
Model 530 also provides
an identical feature set to the previous models. In addition, it includes a pair of jumper
interfaces to enable the insertion of either an internal or external EQ
component. In order to interface an
external balanced differential EQ, the second differential output might be dedicated
to route the “send” signal, while the line input circuit may provide a path for
the return signal from the EQ. This
configuration allows the external EQ to function without the need for a
separate microphone preamplifier or mix board, instead using the internal low
noise SSM2019 microphone preamplifier circuit.
(It has been determined by extensive field testing that the 540 series
“Audio Sculpting” provides superior performance and better level control than
earlier attempts to use external EQ products with the MAX Audio Processor.
The latest model 540
builds further on the EQ platform, replacing the three panel switches of
earlier versions (low cut, low boost, and pre-emphasis) with an internal
adjustable EQ tailored specifically for amateur voice communications on the HF
band with amplitude modulation. The
three new controls are continuously variable: a cut and boost low frequency
gain control, a variable depth sharp notch filter at about 300 Hz, and a
pre-emphasis control from flat response to optimal, then on to excessive
pre-emphasis levels. In addition to the
standard pair of inputs for microphone and line level, and the two discrete
balanced differential output circuits, the 540 includes an optional expansion
area (not populated) which the creative experimenter might employ to add
another dedicated balanced differential (high or low gain) custom input circuit,
and possibly another low impedance balanced differential output to connect
another transmitter or other device.
Finally, a subset of the
540 is the new 540L, or “Lite” version, which is identical to the 540 in
features, with the exceptions that it provides only one input, one output, and
no unpopulated expansion area. The
single input still employs the SSM-2019, therefore the gain may be set to
either low-level microphone or high-level line mode via a single jumper. This feature retains the high CMR performance
when using line level inputs, as opposed to the typical noise prevalent “attenuate-then-over-amplify”
method typically used to feed line level program material into a microphone
level stage. (Note that the selection of
mic or line level must be set via an on-board jumper, and because there is only
one input preamplifier, the front panel “Mic/Line” switch must remain in the
Mic position. The 540L printed circuit
board is sized to fit within the much smaller desktop enclosure, only 10” wide,
7” deep, and 2” high. If this
combination is selected, an external wall transformer must be used for power,
as there is insufficient space for an internal transformer.
Note: While many of the above versions are no
longer available, the details are listed to reflect the additional features and
incremental improvements provided with each update. (Assembly instructions and other resources
will remain on the web for some time to support users of these earlier versions
of the MAX Audio Processor). The
cost of current versions 540 and 540L (lite) has been reduced due to savings
resulting from volume discounts realized on recent component purchases. The MAX Audio Processor has always been
provided “at cost”, with minimal overhead, but never at a profit.
The Evolution of the MAX Audio Processor
As
more and more folks are becoming aware of the MAX Audio Processor and its
capabilities, it seems like a good time to share a bit of history, the goals,
and current mission with regard to the kit preparation and availability. Its
developers never expected to see the project listed in reviews and other
recommendations, and for that they are very pleased and most thankful!
It all started back about six years ago, when
W8KHK and another ham friend were attempting to hand-assemble a bandwidth
filter and compressor-limiter chip on a solderable perf board, all to test a
theory that might provide better modulation quality when operating a Johnson
Viking II transmitter. Four or five decades ago, it was relatively easy for
them to solder hair-fine wire leads between massive arrays of memory chips.
Jump forward years later, as more tiny wires were added to each assembly,
neatness suffered, patience wore thin, and before either was half-finished,
both simultaneously ended up in the trash bin. There had to be a better way!
A short time later, he was introduced to a
small project enabled by an EDA tool to bridge the gap from an idea and a
schematic diagram straight to the generation of the Gerber files necessary to
direct the fabrication of a printed circuit board, with neatly etched fine
traces, completely eliminating the need to hand solder so many fine wires to
the closely-spaced integrated circuit pins. Immediately he commenced to study
the toolset, and quickly became reasonably adept at designing and creating some
rudimentary printed circuits. A working filter was finally created, but since
the fab facilities policy required a minimum order quantity of ten units,
several boards were left over. Some of these were given to other hams, and some
were made available with the components required to complete the filter
project.
Soon word about the filter spread on amateur
radio via HF AM, and N1BCG immediately became interested in combining the
variable bandwidth filter with his ideas to create a processor for improved AM
audio on the HF ham bands. He contacted W8KHK, and the rest is history. While
these two individuals have never met in person, a collaborative effort, via
email and telephone, quickly ensued. The design has evolved significantly over
a four-year period. During that time, incredibly, W8KHK happened to work Gary
Blau, W3AM (now an SK) during one of the annual "AM Transmitter
Rally" events, who suggested and approved the inclusion of his active op
amp design to implement the "all-pass filter", (also known as a phase
rotator), originally invented as a passive device by Kahn for controlling
asymmetry in broadcast applications. The end result of many years and multiple
prototypes is now a complete "audio chain in a box", from microphone
to modulator, which includes other features such as pre-emphasis, low frequency
management, compression, "Dynamic Audio Processing", the adjustable
bandwidth filter, peak limiting, a distribution amplifier, and the required
ancillary circuits to display performance status and provide guidance in the
proper control settings via LED indicators.
It should be emphasized that this was never
intended to be a product for profit, but instead a "Labor of Love",
addressing a common interest by two individuals. As it evolved, however, it appeared
that the design had the potential to provide the average AM aficionado with the
means to enjoy a clean, loud, densely modulated signal, without resorting to
expensive professional or broadcast audio processing equipment. Much of the
evolution was stimulated by on-air experimentation resulting in many
suggestions and constructive criticism. Incremental revisions were implemented,
some to improve performance or add capabilities, while others addressed ease of
assembly. Initially, fabrication required the hand-wiring of many individual
circuits between the main circuit board and the switches and LED indicators on
the control panel, leading in a circle back to square one, where yet another
half-finished prototype found its way to the trash bin. It should be noted that
this attempt elicited significant negative feedback! Extensive rework of the
board layout finally eliminated the majority of hand wiring, first by employing
prefabricated ribbon cables connecting the main board to a new interface board,
where all the switches and LEDs on the control panel were soldered. Finally,
the ribbon cables were also eliminated, and the main board then directly
interfaced to the control panel board. Remaining hand-wiring requirements were
reduced to a pair of power wires, and a few more leads for input and output
balanced audio, which could be either XLR or TRS connectors. With the recent
refinements, it became vividly obvious the resulting device had the potential
to enable others to "Re-Live the “Heathkit
Experience" while personally assembling their own processor! And now that
the "MAX" evolution appears to be complete, W8KHK, already retired
for many years, may now focus his free time on the many other
partially-completed amateur radio projects that clutter the ham shack!
Since there have always been extra boards, due
to the minimum order requirement, the developers have made them available, with
the associated components, at approximately the acquisition cost. There has
never been, and likely never will be, any intent at making any profit, as this
is a hobby - amateur radio, and the entire project is looked at in a fashion
similar to open source software development and collaboration; the only
difference being that with software, many electrons are temporarily inconvenienced
in order to arrange individual bits into the bytes and words needed to code an
elegant solution to a problem, whereas with a hardware collaborative effort,
the implementation requires the repositioning of atoms and molecules, from the
component vendor location to the consumer, in order to realize a useful device.
It is obvious that it becomes impractical,
almost immediately, to acquire and organize a large list of individual
components to assemble one processor, due not only to the cost of parts in unit
quantity, but also the handling and shipping costs for such orders. Since the
developers of the MAX realized, early on, that several iterations would be
required to achieve a practical design, a large variety of components was
ordered in a manner that quantity discounts and combined shipping and handling
benefits significantly reduced the overall cost of the individual items. Such
is the cost of jumping into this sort of hobby! The savings realized continue
to be passed on to folks who show an interest in having their own "MAX"
experience, building and using an Audio Processor at their own station. Since
the processor incorporates some features tailored directly to the needs of
amateur two-way communication, it has been found to be a better fit, or solution,
compared to even some very expensive processors designed primarily for
broadcast applications. It is our hope that the cost per individual unit will
not be prohibitive, such that many may enjoy the experience as the developers
have over the past few years. Incidentally, it should be mentioned that the
name of the project is in no way related to one of the developers, but was
coined by the other, as he felt the processor provided the "MAXimum utility and performance for a minimum of effort and
investment.
Now that a bit of the history and evolution of
the MAX has been revealed, the developers wish to reiterate the ongoing mission
is to make the benefits of the development efforts available to all that wish
to have a similar processor, whether they are each able to assemble the kit of
parts, or need to enlist the services of others who seek to assemble the kits
for others as a pleasurable and rewarding pastime. As many of the unique
components and many of the earlier circuit boards will never be applicable to
the current design, there has been considerable overhead and waste, but as the
parts complement has been reduced to just those needed devices, it is expected
that the cost to share the kit of parts will be further reduced in the future.
It would be nice to just break even between the costs of parts acquisition and
the price of each kit, but even that is not a concern, as there is no desire to
drive this as any sort of business, with all the associated costs of marketing,
advertising, distribution, and of course the bean counter activities of keeping
expense and income logs. Suffice it to say the developers take great pride and
pleasure in hearing more stations on the amateur bands enjoying better, more
listenable and readable audio without the great expense of professional or
broadcast quality audio processing equipment. With that, they wish to thank
everyone who has contributed time, ideas, criticism, complements, etc. A few
years back, this project would have either been impossible, or would have taken
much longer, but with the available resources of the internet, email, special
interest forums, and web sites, the resulting communication has enabled many
folks with similar interest to participate in a pleasurable and rewarding
endeavor, and it is hoped that many may more may continue to benefit directly
from these efforts.