This is only a preview of the July 1996 issue of Silicon Chip. You can view 25 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Build A VGA Digital Oscilloscope; Pt.1":
Articles in this series:
Items relevant to "Remote Control Extender For VCRs":
Items relevant to "Build A 2A SLA Battery Charger":
Items relevant to "A Three-Band Parametric Equaliser":
Articles in this series:
|
Build a three-band
parametric equaliser
If you are interested in musical instruments,
public address systems or any application
where you need fine control of the audio
spectrum, then this three-band parametric
equaliser could be just what you want. It is a
very quiet, low-distortion circuit that is easy
to use.
Design by BOB FLYNN
T
HERE ARE many audio appli-
cations where simple tone controls or graphic equalisers just
can’t do the job. For the most precise
control of the audio spectrum, a onethird octave graphic equaliser is the
best but it is a complex unit. Such a
graphic equaliser will have 30 or more
sliders to cover the full audio range
but its capabilities may be wasted in
many situations.
For example, you may only have
two or three troublesome peaks or
dips in the response and these could
possibly be fixed by nudging only three
of the sliders – all the rest would be
unnecessary.
70 Silicon Chip
By contrast, a three-band parametric
equaliser can do many of the tasks of
a graphic equaliser and it is a much
simpler unit with considerably less
active circuitry.
Our parametric equaliser has three
frequency bands, with their centre
frequency adjustable over the nominal
ranges from 40Hz to 160Hz, 320Hz to
1.3kHz and 2.2kHz to 5kHz. While
they do not overlap, these ranges have
been selected as a good compromise
between overall circuit complexity,
minimum interaction between ranges,
ease of use and audible effectiveness.
We could have added more bands
but since each band needs a minimum
of three potentiometers, the number
of knobs on the control panel rapidly
gets out of hand. With three bands we
end up with 10 controls in all. The
controls for each band are frequency,
boost/cut and Q. The frequency control is self-explanatory – it tunes the
centre frequency for each band; ie, the
frequency at which the boost or cut
setting is at a maximum.
The boost/cut control is same as
a bass or treble control. In its centre
setting the frequency response for the
band is flat; when rotated clockwise,
boost is applied and when rotated
anticlockwise, the frequencies are cut.
The third control is labelled “Q”
and this knob determines whether
the boost will be applied as a sharp
peak or over a much broader range of
frequencies. Similarly, when cut is
applied, the Q control determines
whether the cut will result in a deep
notch or a much broader “valley” in
the response.
Let’s look at a few examples to see
how the parametric equaliser works in
practice. Have a look at the response
curves in Fig.1. There are actually
three response curves, all with the
Q control set for maximum. The top
AUDIO PRECISION SCFREQRE AMPL(dBr) vs FREQ(Hz)
15.000
17 MAY 96 11:21:36
AUDIO PRECISION SCFREQRE AMPL(dBr) vs FREQ(Hz)
15.000
10.000
10.000
5.0000
5.0000
0.0
0.0
-5.000
-5.000
-10.00
-10.00
-15.00
17 MAY 96 12:37:47
-15.00
20
100
1k
10k
20k
20
100
1k
10k
20k
Fig.1: these boost and cut response curves were taken
with the Q control set for maximum. The top curve shows
the effect when maximum boost is applied in all three
bands. This results in three sharp peaks centred at about
64Hz, 490Hz and 3.3kHz. Each one of those peaks can be
moved back or forward within its respective frequency
band, by rotating the relevant frequency control.
Fig.2: this set of response curves was taken with the Q
controls set for a medium value; ie, with the control
centred. The first curve shows the low band set for
medium cut while the other two bands have medium
boost applied. The second is the reverse, with medium
boost applied in the low band and medium cut applied in
the middle and top bands.
AUDIO PRECISION SCFREQRE AMPL(dBr) vs FREQ(Hz)
15.000
AUDIO PRECISION SCFREQRE AMPL(dBr) vs FREQ(Hz)
15.000
17 MAY 96 11:29:54
10.000
10.000
5.0000
5.0000
0.0
0.0
-5.000
-5.000
-10.00
-10.00
-15.00
17 MAY 96 11:33:38
-15.00
20
100
1k
10k
20k
Fig.3: one of these curves shows the low and top bands
boosted while the centre channel is cut. The other curve
shows the low and top bands cut and the centre channel
boosted.
curve shows the effect when maximum
boost is applied in all three bands.
This results in three sharp peaks as
you can see, centred at about 64Hz,
490Hz and 3.3kHz.
Each one of those peaks could be
moved back or forward within its respective frequency band, by rotating
the relevant frequency control.
The bottom curve shows the same
frequency and Q settings as for the
top curve except that the boost/cut
control is now set to maximum cut.
Meanwhile, the third curve which
is between the top and bottom traces
shows the overall flatness of response
20
100
1k
10k
20k
Fig.4: this pair of frequency plots shows the low band set
for a flat response, while the centre and top bands have
either modest boost or cut.
when the boost/cut controls are all
centred. The response is less than 1dB
down at 20Hz and 20kHz.
As shown by the above curves, the
maximum boost and cut which can be
obtained at any frequency within the
band ranges is ±10dB. Note that you
can have any combination of boost &
cut, frequency and Q settings so the
number of response curves you could
obtain is virtually infinite. It means
you can compen
sate or “equalise”
the frequency response for many “real
world” applications.
Fig.2 gives another set of response
curves, this time with the Q controls
set for a medium value; ie, with the
controls centred. The first curve shows
the low band set for medium cut while
the other two bands have medium
boost applied. The second is the reverse, with medium boost applied in
the low band and medium cut applied
in the middle and top bands.
Fig.3 is another variation on the
theme, this time with the low and
top bands boosted while the centre
channel is cut and then with the low
and top bands cut while the centre
channel is boosted.
Finally, Fig.4 is a pair of frequency
plots with the low band flat while the
July 1996 71
AUDIO PRECISION SCTHD-HZ THD+N(%) vs FREQ(Hz)
5
17 MAY 96 12:42:53
1
AUDIO PRECISION SCTHD-HZ THD+N(%) vs FREQ(Hz)
5
17 MAY 96 13:36:10
1
0.1
0.1
0.010
0.010
0.001
0.001
.0005
.0005
20
100
1k
10k
20k
Fig.5: total harmonic distortion versus frequency with all
the boost/cut controls centred (ie, with a flat response), at
a level of 1.5V RMS.
AUDIO PRECISION SCFREQRE AMPL(dBr) vs FREQ(Hz)
15.000
20
100
1k
10k
20k
Fig.6: total harmonic distortion versus frequency, taken
with the three bands set for maximum boost and high Q,
as in Fig.7.
17 MAY 96 12:54:11
Performance
Frequency response ............... (see graphs)
10.000
Signal-to-noise ratio ............... 99dB unweighted
(22Hz to 22kHz); -103dB A-weighted, with respect to
1V RMS (with boost/cut controls centred)
5.0000
0.0
Harmonic distortion ................ see graphs
-5.000
Maximum output level ............ 9.3V RMS
Maximum boost & cut ............. ±10dB
-10.00
Range of Q ............................. 0.45 to 5
-15.00
20
100
1k
10k
20k
Fig.7: response curve with all bands boosted; this is the
test condition for the distortion measurement of Fig.6.
other two bands have modest boost
or cut.
Fig.5 is a plot of total harmonic distortion versus frequency with all the
boost/cut controls centred (ie, with a
flat response), at a level of 1.5V RMS.
As can be seen the distortion is very
low, averaging about .002%.
Fig.6 is another plot of total harmonic distortion but this time with
the three bands set for maximum boost
and high Q, as in Fig.7. This time the
distortion is somewhat higher but still
satisfactory for the applications in
which the circuit is likely to be used.
Circuit description
Fig.8 shows the complete circuit of
the three band parametric equaliser.
It is based on three “state variable”
filters, one for each of the bands. Each
of the state variable filters is identical
72 Silicon Chip
Supply current ........................ 30mA (typical) at ±15V
apart from the capacitors which determine their frequency ranges. All the op
amps are LM833 dual low noise types.
Eleven op amps out the total of 12 are
used and IC2b is unused.
To simplify the discussion of the
state variable filters, let’s confine ourselves to band 1, the low frequency
band. It employs IC1a, IC1b and IC2a.
The latter two op amps are integrators
with their frequency cutoff determined
by the 0.12µF ca
pacitors and their
tuning controlled by the 25kΩ dualganged pot VR3a & VR3b.
State variable filters have three useable outputs: high-pass, low-pass and
bandpass (ie, low-pass and high-pass
in combination). The bandpass output
is the one we want and this is taken
from the output of IC1b, via the 6.8µF
non-polarised (NP) capacitor.
The Q of the filter is controlled by
IC1a, in conjunction with the 100kΩ
dual-ganged pot VR2a & VR2b. VR2a
is in the input to IC1a while VR2b is
in the feedback loop from IC1b to IC1a.
Both pot sections are wired as variable
resistors. Notice that the wipers of
VR2a & VR2b are shown with an arrow
to show clockwise rotation of the knob;
maximum clockwise rotation gives
maximum resistance for VR2a & VR2b
and this corresponds to the maximum
Q condition.
The three state variable filters are
Fig.8 (right): the parametric equaliser
is based on three “state variable”
filters, one for each of the bands. Each
of the state variable filters is identical
apart from the capacitors which
determine their frequency ranges.
July 1996 73
Fig.9: follow this layout diagram when installing the parts
on the PC board. In particular, check that the ICs are
correctly oriented and don’t get the pot values confused.
74 Silicon Chip
Fig.10: check your board carefully for etching
defects before installing any of the parts by
comparing it against this full-size pattern.
There are quite a few links on the board and these should be installed before
any other components are soldered in. Take care to ensure that all polarised
parts are correctly oriented and note that the ICs all face in the same direction.
effectively in parallel and connected
into the feedback network of op amp
IC6b on the input side and into the
input circuit of op amp IC4b on the
output side. When all the boost/cut
controls are centred, the gain of the
circuit is unity over the whole audio
frequency range. When one of the
boost/cut controls is set to boost, the
signal from the accompanying state
variable filter is increased to IC4b,
while the feedback to IC6b is reduced.
Hence, the gain is boosted for that
particular band.
VR1 provides an input volume control for the whole circuit. We assume
that for most applications it will be
set for maximum input signal to the
circuit and thereby give an overall
gain of unity; ie, 1V in gives 1V out.
The circuit is designed to run from
TABLE 1: CAPACITOR CODES
❏
❏
❏
❏
❏
Value
IEC Code
EIA Code
0.12µF 120n 124
0.1µF 100n 104
.015µF 15n 153
.0022µF 2n2 222
±15V supply rails and these will normally be supplied by 3-terminal 15V
regulators in the main amplifier or
mixer. The rails are heavily bypassed
with 100µF and 0.1µF capacitors to
ensure good stability.
Assembly
We are presenting this project as a
PC board which can be installed in a
case together with a suitable power
supply or incorporated into a larger
piece of equipment. The PC board
measures 230 x 72mm and is coded
01107961.
To make the board size manageable
it has been designed around 16mm
diameter pots. Actually, we could have
made the board a good deal smaller
but in practice, the knobs need to be
spaced so that typical male fingers can
operate them comfortably.
By itself, the PC board is difficult to use unless you also have the
control panel; otherwise you don’t
know where the pots are set. We have
designed a control panel which measures 249 x 59mm. The completed PC
board and control panel have been
designed to fit neatly into a plastic
TABLE 2: RESISTOR COLOUR CODES
❏
❏
❏
❏
❏
❏
❏
No.
1
6
10
6
3
1
Value
100kΩ
20kΩ
10kΩ
8.2kΩ
4.3kΩ
100Ω
4-Band Code (1%)
brown black yellow brown
red black orange brown
brown black orange brown
grey red red brown
yellow orange red brown
brown black brown brown
PARTS LIST
1 PC board, code 01107961,
72mm x 230mm
1 front panel label, 249 x 59mm
10 knobs to suit 16mm pots,
15mm in diameter
2 metres 0.71mm dia. tinned
copper wire (for links)
7 PC pins
6 LM833 low noise dual op amps
(IC1-1C6)
Potentiometers (all 16mm dia.)
3 100kΩ linear dual-ganged pots
(VR2,5,8)
3 25kΩ linear dual-ganged pots
(VR3,6,9)
3 10kΩ linear pots (VR4,7,10)
1 50kΩ logarithmic pot (VR1)
Capacitors
4 100µF 16VW electrolytic
4 6.8µF non-polarised
electrolytic
1 2.2µF non-polarised
electrolytic
2 0.12µF 63V MKT polyester
6 0.1µF 63V MKT polyester
2 .015 63V MKT polyester
2 .0022 63V MKT polyester
Resistors (0.25W, 1%)
1 100kΩ
6 8.2kΩ
6 20kΩ
3 4.3kΩ
10 10kΩ
1 100Ω
5-Band Code (1%)
brown black black orange brown
red black black red brown
brown black black red brown
grey red black brown brown
yellow orange black brown brown
brown black black black brown
July 1996 75
Running The Circuit From A 12V Supply
Fig.11: this full-size artwork can be used as a drilling template for the front panel.
Fig.12: use this power supply arrangement if you wish to run
the parametric equaliser from the 12V supply in a car.
76 Silicon Chip
While the parametric equaliser has been specifically de
signed to run from balanced
±15V rails, it is also possible
to run the whole circuit from a
single 12V supply, as would be
the case if the unit was used
in a car. The distortion, signal
han
dling and signal-to-noise
ratio will not be as good but
for car applications its performance would still be more than
adequate.
To run from 12V it will be
necessary to split the supply to
effectively give ±6V rails. This
can be done by wiring two 4.7kΩ
instrument case measuring 259 x
65 x 180mm (W x H x D). This has
space for a power supply and is
available from Jaycar Electronics
with plastic front and rear panels
(Cat. HB-5974) or with aluminium
panels (Cat. HB-5984).
The full wiring details for the
PC board are shown in Fig.9.
Start construction by checking
the PC board against Fig.10. Fix
any shorts or broken tracks that
may be evident. There should not
be any of these faults but if they
are present it is better to fix them
before any parts are soldered in.
There are quite a few links
shown in Fig.9 and these should
all be installed before the other
components. This done, fit the
resistors. Table 2 shows the colour
codes for all the resistor values
specified. Use your multimeter to
check the resistor values if you are
not sure of the colour codes.
Next, fit all the capacitors, making sure that the electrolytics are
correctly polarised; ie, connected
the right way around. Now fit all
resistors across the 12V supply,
as shown in Fig.12. However, the
input and output signal earths
will no longer be tied to the
centre rail; instead, they go to
the 0V rail.
This means that input earth,
the grounded side of the input
pot VR1 and the output earth
must all be isolated from the
earth system (supply centre tap)
and connected instead to the 0V
line of the incoming 12V supply.
If this is not done correctly,
there will be a short across the
-6V rail and the circuit will malfunction.
six ICs; they are all oriented in the
same direction.
Last, fit the pots and make sure
you don’t get the 25kΩ and 100kΩ
pots swapped around. Check your
work carefully against the wiring
diagram when you are finished.
Power up
When the board is complete,
connect a DC supply set to ±15V
and check the voltages. +15V
should be present at pin 8 of each
LM833 while -15V should be at
pin 4 of each IC. Then, if you check
the output of each op amp, pins 1
or 7, the voltage should be close to
0V. The exception is pin 7 of IC2b
(unused) which is likely to be at
-15V; this does not matter.
Further testing cannot be done
until you make input and output
connections to the board via
shielded cable. You can then use
an audio oscillator and an oscilloscope (or an AC millivoltmeter
or DVM with a wide frequency
response) to check the effect of
SC
each control.
|