AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
PE Mini-organ – Part 1
Student projects
All educational establishments are closed
now, so I thought I would get a nice pile
of student projects finished for when everything restarts. An eternal favourite is
the Stylophone, but all previous designs
have had problems. My version presented here has a big Mike-Grindle-designed
wide-spaced through-hole PCB. There are
only two (optional) surface-mount components: a mini-USB (everybody needs
to learn how to solder these since they
regularly break) and an output transistor
option. There are three traditional analogue stages and a 4×AA battery-pack,
which lasts for days. The keyboard resistor-chain is a nice challenge for those
learning the resistor colour code. (Top
tip, do check every resistor with a good
DMM – if you get just one resistor wrong
then it will upset every note.)
Basic design
The PE Mini-organ
I
’m writing this during the
All that really works is FM radio. This
has brought home the fragility of our
interconnected global society and
dependence on tenuous component
supply chains. Even my solar panels
are useless because
they need power from
T une
the grid to run the
Output
inverter. What a mess,
stage
but at least there’s still
S peaker
Cadbury’s Chocolate
7 5 5 5
L ow- freq uency
S q uare- wave
and AA cells.
osci llator
osci llator
Looking on the bright
side, I don’t feel guilty
for being a hoarder any
V ibrato
V ibrato
V olum e
more. Glancing smugfreq uency
depth
ly at my vast piles of
K eyboard
stylus
dust-covered components and bur nt
d a t a - s h e e t s , i t ’s a
great time to do some
Fig.1. The basic system consists of a CMOS 7555 square-wave low-power analogue circuit design.
oscillator, a vibrato oscillator and a unique output stage.
pandemic but also still suffering
from February’s storm damage, which
means no Internet, TV or landline – the
mobile works in just one room, badly.
54
This PE Mini-organ is based on a provisional design I did for Audio Out (EPE, August
2018 Fig.11). This was later refined to produce a design for Dubreq, which became
their new Analogue Stylophone, an ultra
low-cost surface-mount design. (This will
of course be cheaper to buy than building the instrument described here when
it becomes available. However, that’s not
the point, building your own is always
so much more than just the final result).
The analogue version is cheaper to make
and uses around a fifth of the power of
the Stylophone S1 digital design. It also
sounds nicer, having no aliasing and other high frequency whistles. Incidentally,
this ‘analogue renaissance’ in electronic
music is now an established marketing
angle to boosts sales. Of course, the main
disadvantage with analogue is tuning accuracy. There is always more drift with a
capacitor as the heart of a timing element
compared to a numerical divided-down
quartz or ceramic resonator. There’s no
need to worry too much about this for a
‘toy’ instrument. I don’t think a Stylophone
has ever been used in an orchestra; it’s pinnacle was probably David Bowie’s Space
Oddity. (https://youtu.be/iYYRH4apXDo).
Practical Electronics | June | 2020
4
R
2 /3 × V C C
5
T uning/ vi brato
co ntrol vo ltage
–
6
M aster reset
R eset
+
T hreshold
tim ing ca p
R
3
Output
Q
C om parators
C
2
–
T rigger
F req uency
set resistor
( keyboard
resistor
ch ain)
G round
S et
+
Q
7
D isch arge
( not used)
1 /3 × V C C
R
5kΩ
1
C harge/ disch arge cu rrent
Fig.2. Internal circuit of a CMOS 7555. Is it analogue or digital? I would say both.
Block diagram
oscillator to provide vibrato. The output
wave is fed into a MOSFET speaker driver.
That’s a 20p chip with two 5p transistors,
about right for a student project – see Fig.1.
The basic system is a 555 square-wave oscillator controlled by the keyboard. This is
modulated by a phase-shift low-frequency
+
n
8
I make no apologies for using the good
old 555-timer chip for the oscillator. It was
used in the original Stylophone designs
from the late 1970s. Internally, the chip
uses a chain of three resistors to define
the top and bottom comparator-switching
thresholds (see Fig.2). Since these thresholds are ratio-based, the output frequency
does not vary much with supply voltage
or temperature. It is stable enough for a
Mini-organ run off batteries with no voltage regulator. In its CMOS version (called
the 7555), as used here, the power consumption is less than 0.2mA. This chip
could be considered an analogue/digital
hybrid. Are comparators analogue or digital? Here, it is used in an analogue way
with a capacitor that charges up.
The circuit diagram is shown in Fig.3.
It looks challenging for a beginner, but it’s
not when broken down into its functional
blocks. The internal operation of the 555
has been well covered before. Pin 5 allows
a degree of voltage control of the frequency. The output of the tuning control (VR1)
and the vibrato oscillator are summed here.
+ V C C
–
B attery
P ower
D 1
S B 4 0
2 . 1 m m power socke t
with battery bypass
R 7
470kΩ
S 1
D 2
S B 4 0
C entre pin
V B U S
D –
U S B -A
D +
G N D
N C
C 1
1 5 0 nF
N C
C 2
1 5 0 nF
C 9
4 7 0 µ F
1 6 V
+
V R 1
0kΩ
R 4
30kΩ
C 3
1 5 0 nF
U S B -B
M ini
D +
G N D
D 3
1 N 4 1 4 8
8
5
D epth
6
V R 2
220kΩ
2
C 1 2
1 nF
C 1 0
1 0 nF
C 8
1 0 nF
C 5
1 0 nF
C 6
1 0 nF
C 1 3
4 7 0 pF
C ON
R E S
T H R E
D IS
T R IG
OU T
S peed
S 3 b
N C
N C
N C
S 3 a
N C
* T R 2 ca n also
be an S M D :
F D C 6 3 4 P
6 .3 m m
output
j ack
socke t
4
7
N C
T R 2 *
Z V P 2 1 0 6 A
3
R 1 1
47kΩ
C 7
4 7 0 nF
G N D
1
S 3 c
IC 1
7 5 5 5
+ V C C
V R 3
00kΩ
T R 1
B C 5 4 9 C
R 2
22kΩ
+ –
C W
N C
C W
R 9
330kΩ
C 4
4 7 0 nF
R 3
4.7MΩ
N C
R 1
8kΩ
C 1 1
4 7 0 nF
R 5
2kΩ
R 6
00kΩ
V B U S
D –
P itch
C W
L S 1
35 80Ω
(4 .5 -9 V )
R 8
2.2kΩ
C W
R 1 0
2.2MΩ
d
V R 4
4.7kΩ
A - log
S 3 d
N C
s
g
V olum e
R 1 2
2.2kΩ
K eyboard
stylus
N C
R 3 8
5kΩ
R 3 7
3kΩ
R 3 6
3kΩ
R 3 5
2kΩ
R 3 4
2kΩ
R 3 3
kΩ
R 3 2
0kΩ
R 3 1
0kΩ
R 3 0
. kΩ
R 2 9
R 2 8
R 2 7
R 2 6
. kΩ 8.2kΩ 7.5kΩ 7.5kΩ
R 2 5
.8kΩ
R 2 4
.2kΩ
R 2 3
.2kΩ
R 2 2
R 2 1
R 2 0
R 1 9
R 1 8
R 1 7
R 1 6
R 1 5
.2kΩ 5. kΩ 5. kΩ 5. kΩ 4.3kΩ 4.3kΩ 3. kΩ 3. kΩ
R 1 4
2kΩ
R 6 3 *
R 6 2 *
R 6 1 *
R 6 0 *
R 5 9 *
R 5 8 *
R 5 7 *
R 5 6 *
R 5 5 *
R 5 4 *
R 5 0 *
R 4 9 *
R 4 8 *
R 4 7 *
R 7 0
.0MΩ
R 5 3 *
R 5 2 *
R 5 1 *
R 4 6 *
R 4 5 *
R 4 4 *
R 4 3 *
R 4 2 *
R 4 1 *
R 4 0 *
T op C
highest
note
Fig.3. Full circuit of the PE Mini-organ.
Practical Electronics | June | 2020
55
Fig.4. PE Mini-organ output waveform
viewed across the speaker. The ringing
is the resonant back electromotive force
voltage (EMF) from the speaker.
Power supply
The PE Mini-organ operates from 3.5
to 12V. Current consumption is 0.2mA
with no note playing. Average current
when played is 12mA, and a continuous
note at full volume draws 40mA. The
current consumption is minimised by
having an uneven mark-to-space ratio on
the output waveform – off more than it
is on. The waveform is shown in Fig.4.
Power supply connections options are
provided via a 2.1mm DC connector, a
standard USB socket and a mini-USB version. Schottky diodes D1 and D2 block
back current passing from the battery.
(I don’t know what happens if you put
9V into a USB socket on a laptop and I
don’t want to test it). The 7555 goes up
in smoke if the power is connected the
wrong way and these diodes also provide reverse-polarity protection. The
current from standard batteries is limited
which avoids most damage, but power
supplies will generally supply enough
current to start a fire.
Note that the 2.1mm connector follows the ‘guitar-pedal standard’, which
is centre-pin negative. Watch out for this
when using ‘off-the-internet’ ‘wall-warts’
that are generally the other way round.
Octave switching
An octave interval is simply a doubling or
halving of frequency. In a digital system,
to go down an octave, a divide-by-two
stage (eg, a flip-flop) would be used. Since
the PE Mini-organ is an all-analogue design, we are just going to double or halve
the value of the timing capacitor.
Three octave ranges are available and
the total capacitance for each range is
10nF, 20nF and 40nF respectively, selected by a rotary switch. Although very
simple in principle, there are problems
with this approach because capacitors
have wide tolerances, causing the octaves to not be exact. ±5% is the best
specification for cheap capacitors, such
as ceramic NP0 types. We could use 1%
polystyrene or silvered-mica versions but
they are scarce and pricey. However, it
56
is the ratios between the capacitors that
matter, rather than the absolute value.
I’ve have found 5% capacitors taken
from adjacent positions of the same reel
are accurate enough because there is
generally little variation between each
one in the same production batch. One
batch I had were all around 9.6nF with
a maximum of 9.65 and a minimum of
9.42nF. So what I have done here is use
parallel capacitors from the same batch
to create accurate doubling of capacitor
values. C6 provides the first (top) octave.
For the second octave, C5 is switched in
parallel. For the final (lowest) octave, C10
and C8 are added in parallel.
The rotary switch specified can be a
pain to wire up, but it’s nice and easy
when soldered directly to a PCB where
all the ‘wiring’ is done for you. Mike
Grindle even paralleled-up the unused
sections, a standard technique which
improves reliability. You’ve already paid
for those contacts, so why not use them?
All oscillators have a tendency to go
a bit flat as they go higher in frequency
due to their finite switching time. To
compensate for this, the upper octave
capacitor (C6) has to be a bit smaller
than the lower ones (C5, C8/C10). If you
have access to a capacitance meter and
you find you have a variation between
them, the lowest value should be used
for C6 and the highest for C10. (Optional) padder capacitors C12 and C13 are
provided to deal with this.
Keyboard
The big problem in musical electronics
is that most oscillators are linear and
musical scales are exponential. This is
because human senses are logarithmic;
it’s the way biological organisms handle
the huge range of intensity of the various
stimuli encountered in nature. The way
round this is to make the keypard resistor value increments non-linear, so that
the notes increase in the correct musical steps (the twelfth root of two if you
need to know). The editor and I nearly
had breakdowns typing the calculations
and ratios for the keyboard resistors last
time (in Audio Out, EPE, August 2018
p.56) so we’re not going to go through
it again! The problem here is to get the
right value resistors in the right holes.
For those who are obsessed with getting
the ratio exactly right (1.0595) there are
spare positions for parallel resistors and
or presets. Most people are happy with
the standard 1% E24 series resistors. Resistor R14 is critical to the scale of the
whole keyboard from the top octave to
the bottom. If you find the top ‘C’ is flat
compared to the bottom ‘C’ it will be
necessary to tweak it a bit. I had to add
a 1MΩ resistor (R70) in parallel with R14
to bring it into line.
Modulation oscillator
This is a standard phase-shift sinewave
oscillator consisting of three capacitors
(C1 to C3) and two resistors. This network is placed in a negative feedback
loop around a standard common-emitter stage. At the frequency where the
network phase lag hits 180°, the feedback becomes positive, and oscillation
commences. On the original design the
vibrato modulation level and frequency
were fixed. Since we are less limited by
size and cost in home construction, pots
are provided for level/depth (VR3) and
frequency (VR2). Note that as usual in
audio RC oscillators, the frequency potentiometer must be anti-logarithmic to
give a smooth adjustment range.
Output stage
Since the waveform is basically a square
wave (see Fig.4) a linear audio amplifier, such as an LM386, is not required. A
simple ‘switch’ will do. A bipolar common-emitter stage could be used, but
the CMOS 7555’s output drive current
capability is too low, so a MOSFET with
its high input impedance has been used.
This stage is wired ‘upside down’ so the
speaker goes to the ground rail using a
P-channel device rather than the more
common N-channel with the speaker going to the power rail. This approach is
needed so that the output is ground referenced, allowing external amplifiers to
be connected via a jack socket.
The volume control is a bit unusual
in that it is simply a variable resistor
I nput
I nput
–
+
+
–
R ear vi ew
of switch
Out- ofphase
I n- phase
–
+
Output
T oggle next
to output is
in- phase
+
–
Output
Fig.5. Try flipping the phase on the
speaker with this specially wired DPDT
toggle switch. Sometimes it sounds
better in one position.
Practical Electronics | June | 2020
resistor to prevent oscillation. R10 and
R9 form the biasing network for the
FET to make sure it is normally biased
off. These resistors may have to be adjusted if different devices are used. D3
is a clamping diode that prevents the
waveform developing an additional
DC voltage across the coupling capacitor as the notes are pulsed on and off.
If the diode is omitted, an interesting
pulse-width modulation effect occurs
as the bias changes. The diode D3 and
coupling capacitor C7 could be avoided by DC coupling the output device,
but then it would be difficult to bias
correctly and there is always the possibility of it latching-up hard on. The
speaker output is muted when a jack is
inserted into the socket. A DC load on
the TR2 is maintained by R12.
Fig.6. A suitable high-impedance speaker
for the PE Mini-organ – note its position on
the bench edge to give a baffle effect.
placed in series with the speaker to limit the current. It is placed at the output
rather than the more normal input position, so that the MOSFET is always fully
driven. I was surprised such a primitive
arrangement worked so well. Since the
resistance of the pot has to decrease as
it is rotated clockwise, it has to be anti-logarithmic. Normally, a logarithmic
type is used in potential divider mode
where the volume increases as the resistance gets bigger going clockwise.
However, here the resistance needs to
decrease because the potentiometer is
used a variable resistor.
The best output device to use if you
want to run the instrument on low-voltage power sources, such as USB, is the
Fairchild FDC634P. Unfortunately this
is only available in surface-mount, but
there is provision on the board for both
a through-hole and SMD MOSFET. A
similar MOSFET is available in a TO92
or Zetex E-line package, the ZVP2106A.
This costs a bit more and has a higher
turn-on voltage. It is best suited for 6-9V
operation, but it will work with the USB
supply, although it will give slightly
reduced output. R11 is a gate-stopper
Loudspeaker
For musical instruments, never use Mylar
cone speaker. They’re designed for alarms
and have poor tone quality. Instead, use
a large lightweight-paper-coned speaker,
but not an 8Ω one because peak currents
could reach 1.1A – not good for small batteries! High impedance speakers can be
hard to obtain, so I have made provision
for a supply of new old stock (NOS) 5 ×
3-inch paper speakers originally made
for old tube style TVs and then left in
a warehouse for 20 years. For low voltages (4V to 6V) use a 50Ω speaker. For
9V, use 80Ω.
One strange observation I have made
is that speakers fed with asymmetrical
waveforms often sound better connected one way compared to the other. So
it’s worth quickly flipping the plus and
minus terminals on the speaker just to
check. I suspect this is because the cone
motion of the speaker is also asymmetrical. A quick phase-flip switch,
shown in Fig.5, is an essential bit
+ 5 V
(+ 9 V )
R 9
330kΩ
C 7
4 7 0 nF
R 1 1
47kΩ
If you want to use a standard 8Ω or 4Ω
speakers, then an output transformer can
be used to match the impedance. Even
these components are now difficult to get.
Luckily, Mouser offer several types from
Xicon, also J Birkett’s, a small electronics
shop sell Eagle brand transformers (write
to: J. Birkett, 25 The Strait, Lincoln LN2
1JF). There is a degree of leakage inductance with transformers, so it is possible
to use this to resonate with a capacitor
(C14), producing a peaking low-pass filter, which gives a ‘warm’ tone. This trick
was often used in low-power valve amplifiers. Since transformers are still used
extensively in expensive ‘Neve’ style
audio processors, this is a low-cost way
of introducing their principles to audio
engineering students. Fig.7 shows the
transformer circuit and Fig.8 shows a
photo of the transformer. One of the interesting points of inductive/transformer
loading is that very little of the power-rail
voltage is lost, unlike the high impedance speaker. This means the transformer
should present a load of at least 200Ω (for
5V) and 500Ω for 9V to give the same
power consumption and power output
as a high-impedance loudspeaker.
Next month
In Part 2 next month we will discuss
compnents, including sourcing them,
and how to assemble the project.
T R 2
s Z V P 2 1 0 6 A
g
d
V R 4
4.7kΩ
A - log
C W
V olum e
* T 1 has
500Ω primary
T 1 *
im pedance
S tart L T 7 2 6
‘ T one’
ca p
C T
C 1 4 *
3 3 0 nF
0 V
Transformer output
D 3
1 N 4 1 4 8
I nput
R 1 0
2.2MΩ
of gear in every audio designer’s tool kit.
The 50Ω ITT speaker specified sounded
best in phase, with the ‘+’ on the speaker
going to the ‘+’ on the board.
A loudspeaker is an air pump and it
needs a baffle or box to give good bass
response. For testing, placing the speaker halfway face-down across the bench
(see Fig.6), which provides an adequate
degree of baffling for testing purposes.
* W ired acr oss
transform er
F inish
8Ω
L S 1
8Ω
4Ω
C onnect 0 V to ‘ finish’
for 9 V operation
0 V
Fig.7. Adding a transformer to the PE Mini-organ
output stage enables a standard low-impedance
speaker to be used. Tuning it with a capacitor
can improve the tone over direct drive.
Practical Electronics | June | 2020
Fig.8. Output transformer for impedance matching. It’s too big to mount on the
board, so should be mounted in a secure way.
57
