hi-fi noise filter-range expander

hi-fi noise filter-range expander
THE ARTICLE, “NOISE REDUCTION TECH-
niques,” that appeared in the January
and February 1981 issues of RadioElectronics, presented block diagrams
of commercially available dynamic
range expanders and noise filters. That
two-part article showed how, by improving the dynamic range of even the
best recorded musical signals, expanders and noise filters restore much of the
emotional impact that is lost during the
recording process. This two-part article
will describe the operation and construction of a combination dynamic
range expander/noise filter called the
ASRU (Audio Signal Restoration Unit).
This month, we will describe the
asic operation of the ASRU and proide an in-depth description of how the
xpander portion of the circuitry works.
ext month, we will discuss how the
noise-filter circuitry of the ASRU works
and provide the construction, installation, and operation details.
The expander-how it works
Like the expanders discussed in the
January 1981 issue of Radio-Electronics,
the expander section of the ASRU
makes the low-level signals softer and
the loud signals louder, thus providing
improved realism and reduced noise.
The expansion curve of the circuit
is shown in Fig. 1. Note that the total change in gain is about 8.5 dB; the
slope is very shallow. It requires over
40 dB of range to change from minimum
to maximum gain, for an average expansion rate of about 1.2: 1 (the ratio of
output-level change to input-level
change in dB). The curve shown provides expansion without unnatural
side-effects.
1 00µV
1mV
1OmV
1OOmV
1V
SIGNAL LEVEL
FIG. l-EXPANSION
a shallow slope.
of the ASRU shows
A block diagram of the expander portion of the ASRU is shown in Fig. 2.
The first stage sums the right- and
left-channel signals coming from the
noise filter so that both channels can be
controlled together, preventing the
stereo image from changing due to
variations in signal level in one channel
or the other.
The control-voltage filter takes the
output of the summing network and attentuates the high- and low-end frequencies to produce an audio signal
that approximates the response of the
human ear (see Fig. 3). That response is
shown by the well-known FletcherMunson curves (Fig. 4) that depict the
sensitivity of the ear for equal perceived
loudness at different frequencies. Note
that, at most levels, the ear is significantly more sensitive to midrange frequencies than to high- or low-end ones.
In fact, due to the resonance of the ear
canal, the ear is most sensitive to
sounds in the 4-kHz range.
That midrange sensitivity accounts
very strongly for our perception of the
loudness of a sound and the controlvoltage filter is designed to take advantage of that fact.
The attenuation of both ends of the
audio spectrum tends to reduce the effects of noises such as turntable rumble
and FM multiplex “hiss.”
Furthermore, the steep roll-off at low
frequencies prevents low-frequency
signals from causing rapid and unnaturalsounding gain changes. That is beneficial because sudden changes in gain
during the period of a signal can result
in harmonic distortion-something we
+lO
OdB
R-E TESTS IT
a -10
P
2 -20
CJ
-30
LEN FELDMAN
CONTRIBUTING HI-FI EDITOR
We tested a prototype of the Audio
Signal Restoration Unit in our laboratory, using static signals as well as
musical program material. As the
author suggests, setting up the
unit is a bit tricky. To some degree
(unless the expander section is
turned off altogether), there is some
audible interaction between the
various front-panel controls on the
unit. We found that the best setting
for the sensitivity control is such that
medium or average loudness-level
portions of the program source cause
sequential extinguishing of the indicator LED’s. The threshold control
should be set so that in the absence of
any signal, the lowest-level LED
flashes only occasionally.
With the expander switch to the ON
position, optimum setting of the
expander-sensitivity
control occurs when the right-hand LED flashes
only intermittently. Of course, it is
possible to use each section (noise
reduction, dynamic filter and expander) as required, to suit program
material, but we found that with the
controls set as described above, we
were able to improve reproduction of
most program sources without having
to make extensive readjustments
every time we altered program material
or content.
With the expander out of the circuit,
and with the unit set for widest bandwidth (no dynamic filtering or noise
reduction), overall frequency response of the unit measured flat within
±0.75 dB from 20 Hz to 20 kHz.
The unit has essentially unity gain,
but that may be varied by means of the
input sensitivity control. With 0.5
volt input, we measured a signal-tonoise ratio of 90 dB, IHF “A’‘-weighted.
With both the expander and the noise
filter on, total harmonic distortion
for a 1-kHz input signal at the 0.5volt
level measured 0.17%. With the expander turned fully off (the threshold
control at its minimum position) but
the noise filter on, distortion decreased
to less than 0.1% for the same test
signal.
A series of composite spectrum
analyzer sweep photos for the expander/filter/noise-reduction unit is
shown in Fig. 1. In both the upper and
lower series of sweeps, the expander
is on and degree of expansion is
varied, as are the noise reduction
and filtering action. Note that greater
expansion occurs at the higher signallevel (upper traces) and that regardless of the level at which the tests
were made, no expansion is evident
at the low-bass frequencies.
Figure 2 shows the expander action
alone (without any noise reduction or
band-filtering action). With the expander turned off, response is flat
from 20 Hz to 20 kHz; but with the
42
-40
-50
20
100
10K
1K
FREQUENCY-Hz
20K
FIG. 3-FREQUENCY RESPONSE of controlvoltage filter matches that of ear.
LOUDNESSLEVEL
-PHONS
COMPOSITE spectrum-analyzer sweep
photos for the ASRU.
100
20
0
-10
20
50
100 200
500 1K
2K
FREQUENCY-Hz
SPECTRUM-ANALYZER
sweep photos for
the expander alone.
expander turned on, the degree of
expansion for louder passages, less
for moderate passages and, in the
lower traces, even a bit of downward
expansion for quietest passages.
The Audio Signal Restoration
Unit operates with very few side
effects once it is properly adjusted. By
not allowing expansion to take place
at the bass frequencies, the designer
has overcome some of the pumping
and breathing effects common to
other linear expanders. The 1.2:1
ratio of expansion is quite moderate,
compared with some other commercially available expanders, but
nevertheless is sufficient to add a
measure of realism to most program
material that has been compressed
during recording.
As for the variable-bandwidth
filters: if used to excess, they can
create some undesirable audible
effects; but it is possible to benefit
from them without suffering such
effects if adjustment of threshold and
bandwidth is carefully done while
listening to program - material. We
did not find the indicator LED’s to be
as helpful in setting up the unit as the
author had suggested; but we did
find that, with a little practice, we
were able to use the “ASRU” with just
about any component system that is
equipped with an ordinary tape-out/
tape-play monitor loop. The tapemonitor loop on the amplifier that is
used to connect this unit is duplicated
on the unit itself, so owners of cassette or open-reel tape decks need not
worry about losing it.
5K 10K 20
K
FIG. 4-FLETCHER-MUNSON curves show that
the ear is most sensitive to midrange frequencies.
can do without.
The audio from the filter is passed
through a precision full-wave rectifier
that generates a current used to produce
the control signal.
The logarithmic curve-shaping an
attack/delay circuits convert that cur
rent into a control voltage that is ap
proximately proportional to the logarithm
of the current and that section of the
expander provides attack and decay
times that adjust themselves to the rate
of change in signal strength.
Finally, the control voltage is sup
plied to the voltage-controlled attenua
tar/amplifier where it is used to modify
the qualities of the original audio signal
The ASRU’s expander does not ex
pand signals in the low-bass region as
much as it does in others. There are
two reasons for that.
First, consider Fig. 5-a, showing a
warp or rumble (very-low-frequency)
waveform along with a toneburst. As
can be seen in Fig. 5-b, at the moment
the toneburst is added, the level of the
warp signal will increase because the
expander will increase the gain and a
“thump” will be evident, even though
the warp noise alone was inaudible.
Figure 5-c shows what happens when
the ASRU is used--the “thump”
doesn’t occur because the action at
very low frequencies is minimal.
Second, although the ear is relatively
insensitive to very low frequenciesrefer to the Fletcher-Munson curves in
Fig. 4-once their level is above the
threshold of hearing, a 2-dB increase
appears as great as a 5-dB increase in
the midrange area.
For both of those reasons, as well as
to keep distortion to a minimum, the
ASRU’s expander does not expand the
low-bass as much as it does the midrange. Figure 6 shows the ASRU’s gain
vs. frequency response at varying control-voltage levels. Note how well that
matches the changes in gain sensitivity
shown in Fig. 4.
The ASRU’s shallow expansionslope, midrange-emphasized controlsignal and minimized low-bass expansion explain why it is so cleansounding, while allowing 8.5 dB of
effective expansion.
The noise filter-how it works
The heart of the noise-reduction system is a triple-output, voltage-controlled filter, the block diagram of
which is shown in Fig. 7. It is a statevariable filter, which means that certain of its characteristics can be modified while others are maintained.
The integrators process the signal
for use by later stages of the noisereduction system (see Fig. 8). For
sinewaves, the output is reduced by a
factor of two (6 dB) for every octave
increase in input frequency. By varying the gain or time-constant of those
integrators, or the amount of feedback
around them, the comer frequency (the
frequency at which the amount of attenuation reaches 3 dB) can be changed
without changing the shape of the
filter.
Refer to Fig. 8 as we discuss the
ASRU noise-reduction system.
If no signal is present, the control
voltage sets the comer frequency of
the triple-output filter to 1.2 kHz.
Figure 9 shows the frequency response
of each output of the triple-output
filter with the comer frequency set at
1.2 kHz. The overall output of the
noise filter is taken from the low-pass
output via a buffer. Thus, with no input signal present, any noise will be
greatly attenuated.
If a 5-kHz tone is suddenly applied
to the input, it will appear unattenuated at the high-pass output and will
be greatly attenuated at the low-pass
and bandpass
outputs. The AC-DC
converter connected to the high-pass
output will provide a strong signal
that will rapidly pass through the
attack/delay element and cause the
control voltage to increase. As the control voltage increases, the comer frequency of the filter will also increase
until it exceeds 5 kHz.
Soon there will be a stronger signal
in the bandpass
section than in the
high-pass section. That is converted to
DC and will be fed back and reduce
the control voltage. In the case of a
steady tone, that action will serve as a
feedback loop that forces the bandwidth of the filter to “catch” the input
frequency, allowing it to go through
the low-pass filter to the output, while
the noise above that frequency is
filtered out.
Music, of course, is more than just
simple tones. The ASRU noise filter
will track the highest significant frequency of a complex signal. During
a transient-a short, but intense, increase in high-frequency energy-the
comer frequency will overshoot slightly.
That is desirable, since transients mask
noise very well.
If the signal is extremely strong,
I
43
lvblR@sw
-Anrsdr;tors%~896unleaa~
~
Sanicondudor?
R5L
-rpadtbd
RI, R@, RIO, RlOl, RlOB, IWO, R221,
R521,-100,000
ohm8
R2435, RlOIk-RN&, R211, R2lh R214,
R3U5, R3@,‘R3OB, R3lQ, R405, s
’ R4u8, R409, RQf$l, R5lB, Fq33,.
R53!$-14000
oJl’ms
R8, RlO6, R23G R507, R50&, Rs;M,
~*uoo ~rnsa
R7, Rt2, F?lW, Rtl2-2#X3 ohms 1
88, Rl4, RI& RlO& Rlr4, Rlj5, R205
>~,~~Rs2B-22uocjh~
’
.* IA
RI?, Rlll, R213-W~ahm~
R13, R113, R201, R2B& :feu% &2?& ^
c (A
Fm5, R22u, 4%?.g2$ R229* R!M!i*~
R522,~7OO~ms
1 4 I !
fm* Rl18, R2u2, f?2p3, two, l%pu@- I :,
lU,MIO ohm, 309& stkb #xz%+&m~< ‘l)_
D201-D204, D206-0210, D5OW513,
051%05170lN4148
’
D2u5-3.~volt 2!bner
~14-4.7%3tt 2&w
D5Ql-D504-1~1
tED201-&ED2U4, LED501, LED502mini-LED (TL2OB or @quivabnt)
Q2ulQ2u3, Q501, Q502-2N3BO4
Q2u4Q2u5-~25u
icl, Hz?, lc4, IC7, ICB, ICIO, ICl lRC4135 quad op-amp
IQ, G64U49 CMOS hex’invwter
tC5, -73B dual audio preamplifier
lCl2, tCl5-78ll?A 12-v&t positive,
vottagr, rtPgulator
L201, t-202-5.8 ml-l coil
Tl-13.5 VAC, 350 mA, wall-plug
tf8nsfc?fm~r (Dwmey8r PSl4204 or
equiv&ent)
JPd4, JlOl-JlO4-RCA-type
phorp jeclq
Sl, 43lti+opD;T toggle or slide switch
B 12-conductor ribbon
cab&k, fC sockets, cbsis and and
~#Ms, salder, wire, hardware, etc,
f9412, R514, FWH, F@3!&~~~ ” c~ ‘_
j ^ I ~“” ( _ ’
*OOQ &#p A
* ”
@4228*R51\~27u&&~:~~
;G,R4(f3-
writa few -
27Qi~43Mw
-
R232* .R%‘k4, b34U4-4200 ohms
R2Q R5lF ohms
TRIPI
EsflllTPIlT
un#rN&-~~
emlytig
-7-Q.t #ki &amic disc
OFF
rl
-Allprrkrrr
i
the clamp in the bandpass section will
allow the bandwidth to extend all the
way to 25 kHz. The attack/decay circuitry is designed so the bandwidth of
the filter can be expanded rapidly, but
takes longer to decrease than it did to
increase. Because of the large amount
of feedback used to control the bandwidth, that nonlinear response does
not affect the steady-state (constantlevel) response, but becomes very
important in the case of transients.
As pointed out in the “Noise Reduction Techniques” article in the
February 1981 Radio-Electronics, one
of the advantages of a filter/expander
combination is that each section can
be adjusted to keep side-effects to a
continued on page 110
minimum. A further advantage provided by multiple notch-filter techniques, is that the worst possible adjustment is limited to between 8 dB and 10
dB of attenuation. If a sliding cut-off
filter were used, that might not be the
limit and the error induced could be intolerable.
With the ASRU, the maximum permissible error can be set by the user
and the possibility of obtaining an
unnatural response-where a lowfrequency band might be attenuated
more than a higher-frequency one-is
avoided. The noise-reduction control
and the spectral-tracking concept with
feedback are two features that make
the operation of the ASRU so effective, yet free of side effects.
Next month, we will provide the circuit details for both the noise reduction
and dynamic range expander portions
of the ASRU. The construction details
will also be given.
R-E
Expander-circuit
description
Figure 10 is a schematic of the right
channel and control circuits of the
expander portion of the ASRU. The
signal is applied through R301 to IC6-a
a digital logic gate acting as a voltagecontrolled resistor. Figure 11 shows
the circuit of one of the six inverters
that comprise the CD4049 IC and will
help explain that unusual, but very
effective and economical, circuit. The
CD4049 CMOS hex inverter differs
from most CMOS IC’s because protection diode Dz is intentionally omitted
to allow driving the input from lo-volt
logic signals while using a 5-volt power
supply. In the ASRU, the “+” supply is connected to ground. That insures that the P-channel FET’s, QA,
are always off. That leaves us with
QB active. As a result, we are left with
six matched (see Fig. 11) N-Channel
FET’s because they are all fabricated
on the same chip. The FET’s are enhancement-mode devices, which means
that after applying about 1.5volts
to the gate, the FET begins to turn on
and the resistance between the source
and drain decreases as the gate voltage
is increased.
That causes IC6-a and R301 to form a
voltage-controlled attenuator. The resistance of IC6-a varies between 55 and
90 ohms. At higher input levels, the
gate voltage is reduced by the control
_
.
.
”
,
circuits to reduce the attenuation. The
output of the attenuator goes through
C302 to operational amplifiers IC4 and
IC5. Section IC6-f and its associated
circuitry vary the amount of feedback
around that op-amp combination and
thus control the gain. Resistor R302
and capacitors C301 and C302, along
with other parts in the feedback path,
make sure that, at low audio frequencies,
the signal does not go through the attenuators and is not expanded (see Fig.
6) of Part 1.
High attenuation in the early stages
of the circuit reduces distortion by reducing signal levels, but makes noise a
potential problem. The high gain and
low noise required make the double opamp combination (IC5-a and IC4-b)
necessary. The front-end of that, opamp ICS. is exceptionally quiet
(4/n V @) and is used as an inputdifferential pair of transistors.
Integrated circuit IC7-b serves to add
1.2 times the drain voltages onto the
gates of IC6’s FET’s in a compensation
scheme that further reduces distortion.
The net result is an expansion block
that is both quiet and undistorted. To
derive the control signal, the two channels are summed by resistors R501 and
R502. The gain of the control channel is
varied with sensitivity control R503.
Capacitors C506 and C502 reduce the
low-frequency gain, and C501 rolls off
the higher frequencies to obtain the
curve shown in Fig. 3 of Part 1. The
gain-stage of that filter is ICI l-d. That
IC and its associated components form
a full-wave rectifier, such that the sum
of the current through R509 and R5 10 is
proportional to the absolute value of the
filtered waveform. That makes the ex-
pansion polarity-independent so that all
signals are expanded properly, regardless of their phase.
The currents in R509 and R510 and a
bias current through R5 11 are converted
by IClI-b, R512-R514, and D503-D505
to a voltage that is approximately proportional to the logarithm of the current.
It is the non-linear impedance of the resistor-diode network that forms the
gradual slope shown in Fig. 2 of Part 1.
A rapid upward change in signal level
causes D507 and D508 to turn on, and
C504 and C505 can be rapidly charged.
A slower rise will turn on only D507, so
that C505 is charged more slowly
through R517; that dual-mode reduces
distortion for steady-state signals. For
falling signal amplitudes, C504 and
C505 must discharge slowly through
R515 and R517.
The voltage on C505 is amplified by
ICI l-a. If the control voltage (IC 11, pin
3) is below 4 volts, D515 is off and the
signal is attenuated by R525-R527, to
control FET gate-voltage deriver IC7d. Figure 12 shows a block diagram of
the gate-voltage deriver. Varying either
the input voltage or drain current (I,)
will cause the op-amp to change the
FET gate-voltage so that V,, = I, x
R,,,. In that case, increasing signal
levels will increase V,, and thus R,,,.
Since IC7-d drives the gates of both
IC6-a and b, their resistances will track
its output.
In summary, increased signal level
will increase the control voltage, an
thus the resistance of IC6-a, reducing
the attenuation. As the control voltag
gets even higher, diode D515 turns on
and the increasing current from R524
flows into the drain of IC6-c. That
causes the IC7-c gate-voltage deriver to
reduce the resistance of IC6-c and IC6-f,
increasing the gain of the op-amp combination, IC4-IC5.
The op-amps driving LED501 (-6 dB)
and LED502 (+ 1 dB) are used as comparators and absorb the current from
Q501 through their respective LED’s if
the gain is either less than -6 dB or
more than + 1 dB. Transistors 0501 and
502 act as a constant 12 mA current
source that is used for the display of the
noise filter part of the ASRU. Zener
diode D514 makes sure that the current
source can still operate when both
LED501 and LED502 are off.
Even the power supply (Fig. 13) is
unusual. To maintain the exceptional
signal-to-noise ratio desirable in a noisereduction accessory, without requiring
extensive magnetic shielding or coaxial
wiring, the transformer is of the wallplug type, and is physically separate
from the ASRU. This type of transformer has a single, untapped secondary.
To get both plus- and minus-12-volt
supplies using economical positive
voltage regulators, the circuit operates
as follows. On positive half-cycles of
the transformer output, C
6
0
1is charged
up through D60 I and D602. A standard
IC regulator, IC12, is used. On negative
half-cycles D603 and D604 charge up
02, which powers regulator IC13, except in this case, the regulator’s output
is grounded, and the terminal that is
normally grounded supplies - 12-volts.
Noise-reduction circuit description
The noise-reduction-section schematic is shown in Fig. 14. One of the
variable integrators is formed by resistor R6, IC2-a, Cl, R8 and IC3-b. With
just R6 as an input resistor, and Cl as a
feedback capacitor, IC2-a would be a
fixed-gain integrator. The attenuatorconsisting of R8 and IC3-b-in the feedback loop allows the gain to that integrator to be varied.
Capacitor C2 provides AC coupling
to prevent DC errors (such as voltageoffsets) from causing “thumps” when
the resistance of IC3-b changes. Resistor R7 provides bandwith compensation for the op-amp.
Capacitor C3 and resistor R9 do two
things: They filter the control signal to
keep it from acting too quickly and also
feed back drain voltage to the gate of
this section of the IC to cancel some of
the FET’s distortion.
The summing network (see Fig. 7 in
Part 1) is made up of ICl-b and R2-R5.
The reason there are two “-” inputs
and one “+” input is because the design of the integrators causes the signal
to be inverted. The input impedance is
kept high by using ICl-a as an input
buffer. The output buffer, IClO-d, acts
to keep the output impedance low. Resistor R16 is used to vary the amount of
noise reduction.
The highpass and bandpass signals
are added by R14 and R114, and by R15
and R115, respectively. Two 19-kHz
filters to remove any residual pilot tone
from FM multiplex signals-preventing
interference with the action of the noise
filter-are made up of C201 and L201,
and C204 and L202.
It is important that strong fundamental tones below 1 kHz and nonmusical signals above 25 kHz not be
allowed to affect the control-voltagedetermining circuits. High sensitivity,
though, is necessary to allow the filter
to respond to low-level, high-frequency
signals such as those produced by
triangles (the percussion instrument
that goes “ding”). The A739--IC8provides that high sensitivity at a very
low noise figure.
That IC is run in an open loop configuration (without feedback) at the
frequencies where most musical activity
takes place. Resistors R206 and R207,
and capacitors C2 10 and C211 provide
feedback to control the biasing of the
op-amp and to roll off the low-frequency gain. Resistors R204 and R205
reduce the open-loop gain, and capacitors C206-C209, along with C203, roll
off the high-frequency gain. Low-frequency gain is also rolled-off by C202,
C205, C212 and C213.
Capacitors C212 and C213 also couple
the highpass and bandpass signals into
precision rectifiers IC9-b and IC9-c.
The combined gain of those two IC’s is
about 2000. By obtaining that gain
through two stages-IC8, whose output
is a current, and IC9 (together with
R208 and R209), that changes that current to a voltage- it is possible to get a
gain with rectification that is normally
very difficult to obtain due to stray
capacitive feedback.
The weighted-difference blocks (Fig.
8, Part 1) are made from resistors R212R214. they subtract, rather than add,
because IC9-c is a positive-output rectifier, while IC9-b has a negative output.
The bandpass channel is clamped by
diode D205 that turns on at about three
volts, allowing the energy in the highpass section to push the bandwidth all
the way out at high levels. The action of
the bandpass
channel is slowed by
C2 14. That helps cause filter comer-frequency overshoot during transients.
The fast-attack/slow-decay circuit
consists of IC9-d and its associated
components. The attack time is 4 mS;
the decay time 80 mS. Low-impedance
drive for the exponential curve-shaping
and display network, R224-R228 and
Q204-Q206, is provided by IC9-a. When
low-level control voltages are present
at C216, all the transistors are off, and
the resistors act as a simple voltage
divider.
As the voltage at the capacitor increases to a few tenths of a volt, Q202
turns off, Q201 turns on, and LED201
goes out. At higher voltages, Q204
turns on, and LED202 goes out. Since
the base-emitter voltage of Q204 cannot
be much greater than 0.6 volt, R224 is
eliminated from the voltage divider, increasing its output dramatically. In
turn, Q205 and Q206 turn on with the
same effect. As the voltage on R228 approaches 1.2 volts, D210 and Q203 turn
on and clamp the voltage at C216 in a
feedback path to prevent driving the
voltage-controlled filter too hard. Thus,
until the limit is reached, attenuation
constantly declines exponentially.
The voltage on R228 is offset slightly
by R230 so that, even with no input
signal, the filter bandwidth can never
go below 1.2 kHz. As was explained
previously, IC4-a, IC3-c a n d IC3-d,
together with their associated parts,
form a gate-voltage deriving circuit.
Control-circuit gain is varied by R202
and R203. That sensitivity control
determines the ASRU’s response to
transients. Resistor R210, the THRESHOLD control, supplies a bias current to
the attack/decay circuit; if there is
enough noise to change the 1.2-kHz
quiescent comer frequency, that current can keep that from happening.
Construction
Because of the complexity of this
project, the use of PC boards is recommended. A single foil pattern, Fig. 15,
can be etched on one piece of copperclad material that can then be cut in two
to provide both the front and rear singlesided ASRU boards.
Parts-placement diagrams for the
front and rear boards are shown in
Figs. 16 and 17. Use those, together
with Fig. 18, to help you in stuffing the
boards.
The use of IC sockets is recom-
mended. Install those first, followed, in
sequence, by the resistors, diodes,
capacitors, transistors, and other components. Make sure that all 24 jumpers
(12 on each board) are accounted for.
In the prototype (Fig. 18), the LED’s
were mounted in a 16-pin IC socket that
had been cut in half, lengthwise. That
allowed them to protrude far enough
forward to reach the holes in the front
panel.
The front and the rear boards are
mounted back to back. Use two short
pieces of 12-conductor
ribbon cable to
connect the two boards by means of the
holes located at the ends of the boards.
Hole 1 on the rear board is connected
to hole 1 on the front board, hole 2 to
hole 2, etc.
Be particularly careful when installing
IC3 and IC6-they’re static-sensitive.
Keep them in their protective foam or
tubes until you need them. Before handling them ground both yourself and
the PC board to discharge any static
electricity that may be present.
The two power-supply capacitors,
01 and C602, are mounted on the
back (foil side) of the rear board and
secured with plastic cable clamps. The
clamps are attached through a spacer to
the hole between jacks Jl and J6.
Make sure, of course, that every
polarized component is correctly
oriented.
The two PC boards are supported by
wooden end-panels and a metal cover
may be added for appearance’s sake
and to protect the boards. Since the
power transformer is of the wall-plug
type and is isolated by distance from
the ASRU, hum is not a problem.
Installation
Like most other signal processors,
the ASRU is connected to a receiver or
amplifier using the tape-monitor loop.
Its input is connected to the TAPE
RECORDER
or TAPE OUTPUT jacks, and
the output to the TAPE MONITOR, TAPE
PLAY or TAPE IN jacks.
Used that way, the TAPE MONITOR
switch can be used to bypass the ASRU
or to bring it into the circuit. (The EXPANDER
and NOISEREDUCTION controls, if
turned to the OFF position, will also
take the ASRU out of the circuit, and
the TAPE MONITOR switch can always be
left in the ON position.)
A tape deck can be connected to the
ASRU just as it had been connected to
your receiver or amplifier before the
ASRU took over that unit’s tape-monitor jacks.
Operation
It takes some effort to learn how to
set up and use the ASRU properly at
first. Things become easier with practice
though. The LED’s give you an idea of
what’s happening in the system, and
the controls are not very critical (because of the spectral-tracking loop
continued on page 104
noise filter and shallow-slope expander) once they are set for your system.
The noise-filter sensitivity should be
adjusted so that there is an adequate
change in bandwidth during musical
transients. That is the most difficult
part; you should experiment by watching the display and listening carefully
with the NOISE REDUCTION control at its
maximum setting and the THRESHOLD
control at its minimum setting. If you
set the sensitivity too high, you will
hear noise come and go during “unspectacular” musical passages-a sign
that the ASRU is working too hard.
If the sensitivity is set too low, you’ll
hear normal signals being rolled off too
much. Don’t be fooled by the apparent
lack of treble-it’s there, but people
who habitually listen to recorded music
often feel it’s reduced when listening to
a system with noise reduction.
It’s possible to set the sensitivity of
the unit high enough to make the noise
in the signal cause the bandwidth to
open up too easily. If that happens, advance the THRESHOLD control until,
during a silent passage, the lowest LED
of the display flickers occasionally.
The NOISE REDUCTION control setting
is largely a matter of personal taste.
Start with it turned about three-quarters
"I
C11b
way
up,
cul"113L~l1~a1~~ully
Wllbll
the bandwidth is reduced.
The expander’s SENSITIVITY control
should be set so that the right-hand
(highest) LED flickers during peaks.
After everything has been adjusted,
you may still have to reset the expander’s sensitivity from time to time for
use with different sources (e.g., if the
tuner’s output level is higher than the
phono’s). The THRESHOLD setting may
also have to be changed, depending on
the amount of noise in your program
material. For very noisy material, reduce the noise-filter sensitivity and increase the noise reduction to maximum.
You now have a top-notch signal processor that will greatly enhance your
listening pleasure. Use it well!
Acknowledgement
The concept of the spectral-tracking
loop for noise reduction was invented
by Fred Ives of Hewlett-Packard Company while at MIT. Pat Bosshart of
MIT also worked on the concept and
introduced it to me.
R-E
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising