AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
Theremin Audio Amplifier – Part 2
L
ast month, we introducedour
dedicated PE Theremin amplifier –
this month we will build it.
Construction
The PCB overlay is shown in Fig.10. It’s
an ideal beginner’s PCB, no surface-mount
technology, just well-spaced traditional
(‘jellybean’, as our American friends call
them) components. All the transistors are
TO92 centre-base devices. The numbering
is next to the component, not underneath,
for ease of checking. Note the transistors
annotation is ‘Q’ rather than ‘TR’. This is
the default in Eagle CAD.
As usual, solder the resistors in first, in
the same direction for easy reading. Next,
solder the transistors and pre-sets. Do the
middle wire first, then bend them so they
are straight, then solder the other two. Finally, insert the tall electrolytics. There’s
provision for bigger power transistors with
centre-pin collector packages for more
advanced constructors. These are on the
periphery of the board to allow for heat
sinking. Note the bias transistor, TR3 is
designed to be thermally coupled to one
of the output transistors TR4. Fig.11 shows
the completed PCB.
Parts list
(Low-power version only)
Resistors
All resitors are 0.25W 5% carbon-film or
1% metal-film for lower noise
R1
12kΩ
R2
100kΩ
R3
270kΩ
R4
150Ω
R5, R6
3.3kΩ
R7, R8
1kΩ
R9
620Ω
R10
68Ω
R11, R12 1Ω
R13
22Ω
R14
10kΩ
VR1 1kΩ TO5 outline pre-set Rapid 680044 Truohm
VR2 5kΩ TO5 outline pre-set Rapid 680288 Suntan
Alternatively, cheap 5/6mm semi-open
presets, such as Rapid Suntan 68-1574
can be fitted in the other holes.
Capacitors
C1
470nF any type. If using a polarised type make sure plus end goes
to R1 pointing into board.
C2
22µF 3V (minimum) radial
electrolytic or tantalum bead
C3
6.8µF 10V radial electrolytic or
tantalum bead
C4
100µF 10V radial electrolytic
C5
22µF 10V radial electrolytic or
tantalum bead
C6
22nF polyester 5mm
C7
220µF 10V radial electrolytic
C8
470µF 10V radial electrolytic
C9
15pF ceramic
C10 8.2pF ceramic
C11 10µF 6.3V(minimum) radial
electrolytic
Semiconductors
TR1 BC549C small-signal high Hfe NPN
TR2, 3 BC549C small-signal high Hfe NPN
TR4 BC337-40 medium-power NPN
TR5 BC327-40 medium-power PNP
D1
BAT86 or other small signal
Schottky diode
LED 1 standard 3mm red diode
Miscellaneous
PCB from PE PCB Service (AO-1220-01)
Loudspeaker: 25Ω 90mm EuroTec (available
from author: jrothman1962<at>gmail.com)
Testing
Fig.10. PCB overlay – note the power supply noding on the main decoupling capacitor, C8.
Always use some form of current limiting when testing power amplifiers. A
PP3 battery normally has a high enough
internal resistance to provide this. Exceptions are rechargeable batteries and
lithium smoke-alarm batteries, I saw a
student burn his tongue doing the ‘licktest’ on one! However, if a bench PSU is
used, set it to below 300mA because this
is the maximum collector current (IC) of
most small transistors. The DC bias preset PR1 should be set midway. Make sure
the quiescent current preset PR2 is set
fully anticlockwise for minimum current
before turning on. This is tweaked to remove crossover distortion caused by the
dead-band where one output transistor
46
Practical Electronics | December | 2020
Quiescent current
A 300Hz sinewave test set to give an output
of around 2.5Vpk-pk (peak to peak) across
the load is particularly revealing of crossover distortion when setting up by ear. The
preset is turned clockwise until the distortion just disappears and no more. The
current consumption must be monitored
with no signal. If it’s turned up too much,
thermal runaway may occur and cook the
output transistors. Crossover harmonics
are odd high-order, such as seventh and
ninth, and a higher frequency test signal
such as 1kHz will mask them. It’s interesting that the high low-order, second and
third, distortions of loudspeakers do not
mask the edgy crossover distortion generated by class-B amplifiers.
n
Mid-point bias
Fig.11. Completed PCB. Note how TR3 and TR4 are pressed together for thermal
coupling, to keep the quiescent current stable.
accurate with a scope. It’s a good idea to
listen as you look at the screen. It’s an essential part of one’s audio education to
correlate what one sees with what one hears.
C 4
+
4 7 0 µ F
2 5 V
R 5
3.3kΩ
R 1 8
12Ω
C 1 0
8 . 2 pF
+
C 1
4 7 0 nF
6 V
T ant
T R 1
B C 5 4 9 C
R 1
12kΩ
T R 2
B C 3 3 7
* * T R 3 / 6 and T R 4 / 7
cl ose therm al tracki ng
R 7
1kΩ
D 2
R ed
D 1
1 N 4 1 4 8
T R 3 /6 * *
B D 1 3 5
C 1 2
4 7 nF
+
R 3
270kΩ
C 1 1
1 0 µ F
1 0 V
R 1 2
0.39Ω
V R 2
5kΩ
I q set
C 9 *
1 5 pF
1 3 m A
R 1 6
330Ω
R 2
100kΩ
* R 1 4 /C 9
not used
R 4
150Ω
V R 1
1kΩ
C 2
2 2 µ F
+ 6 V
+ 1 V
C 7
1 0 0 0 µ F
2 5 V
R 1 3
10Ω
T R 5 /8 *
B D 1 3 6
T R 9
U 1 8 9 8
R 8
180Ω
R 9
75Ω
+ 8 .1 V
C 6
1 0 0 nF
+ 3 .5 V
R 1 7
1MΩ
R 1 1
0.39Ω
1 3 m A
m odulated
cu rrent sink
L S 1
1 2 . 4 V pk- pk
8Ω output
2 .4 W
+
V R 1 : D C
m id- point
adj ust
* T R 4 /7 , T R 5 /8
W ith sm all heatsink
+ 4 .8 V
C lip
R 1 4 *
10kΩ
T R 4 /7 *
B D 1 3 5
+ 9 .4 V
A udio
input
V +
1 5 V
C 1 3
1 0 0 µ F
2 5 V
R 1 5
56Ω
C 1 4
1 0 pF
I q = 2 0 m A to 3 0 m A
+
C 3
2 2 µ F
1 6 V
R 6
1.6kΩ
1 6 0 m V
+
C 8
+
1 0 0 0 µ F
2 5 V
R 1 0
10Ω
+
turns off just before the other output transistor turns on during the output cycle.
These adjustments can be done by
ear with a signal generator, but it’s more
Mid-point bias is not necessarily exactly
half the supply voltage because of battery
voltage droop, circuit asymmetries and
speaker impedance. This is best done
with a scope at 1kHz to get equal clip, top
and bottom of the sinewave. If it clips one
side before the other, maximum power before gross distortion sets in is reduced. Of
course, this can be done by ear, just tune
for maximum output without distortion.
C 5
1 0 0 µ F
2 5 V
0 V
Fig.12. The addition of a modulated JFET current sink and extra bootstrapping enable the amplifier to be scaled up to 2.4W
into 8Ω. The supply voltage is 15V, and Iq total is 30mA. Note the extra 10pF high-frequency stability capacitor, C14. Also note
that feedback components C9 and R14 have been removed.
Practical Electronics | December | 2020
47
www.poscope.com/epe
Fig.13. Modifications added to the PCB to give higher power and investigate
bootstrapped current sink. Note the bigger output transistors with heatsinks. Oh dear,
we need a new PCB design now –the price of tweaking!
Improvements and variations
(experimenter’s corner)
- USB
- Ethernet
- Web server
- Modbus
- CNC (Mach3/4)
- IO
- PWM
- Encoders
- LCD
- Analog inputs
- Compact PLC
Resistor R8 can be replaced with a current
source which allows the emitter-follower
operating current to be reduced further.
This is because the resistor itself is no
longer soaking up useable audio power. Discrete fixed-value current sources,
such as current-regulator diodes (CRDs),
are another component that is becoming
scarce and expensive. Rapid Electronics
still have some left. I have new old stock.
Don’t even bother looking on other mainstream distributors for them, the ‘generic
pharmaceutical business model’ means
they now cost £1.50 each. Ten years ago
- up to 256
- up to 32
microsteps
microsteps
- 50 V / 6 A
- 30 V / 2.5 A
- USB configuration
- Isolated
PoScope Mega1+
PoScope Mega50
- up to 50MS/s
- resolution up to 12bit
- Lowest power consumption
- Smallest and lightest
- 7 in 1: Oscilloscope, FFT, X/Y,
Recorder, Logic Analyzer, Protocol
decoder, Signal generator
48
Fig.14. Distortion vs frequency curve of the PE Theremin Amplifier in Fig.9 at 6V pk-pk
into 24Ω (190mW). Typical of small 1970s discrete amps.
Fig.15. Distortion curve of the scaled-up circuit in Fig.12 at 7Vpk-pk into 8Ω (760mW).
‘Warm sounding’ like the Mullard Hi-Fi amps in their applications book, Transistor Audio
and Radio Circuits (1972).
Practical Electronics | December | 2020
they were cheap. There are still variable
current sources, such as the LM334, which
are around 50p. Also, JFETs can be used,
although wide tolerances on Idss can give
a two-to-one variation in current. Mouser
still have lots of reasonably priced JFETs.
These are a bit tricky to fit to the PCB since
the devices have three leads and need a
resistor or two.
Upping the power
The low output power may be insufficient
for some people, so a circuit for experimenters using bigger output transistors
and a JFET current sink is given in Fig.12.
Resistor values are also reduced, increasFig.16, Frequency response of the PE Theremin Amplifier. The use of low-value
ing the currents to drive a lower speaker
capacitors gives a bit of bass loss. No problem given the small speaker used in Fig.1.
impedance of 8Ω. Of course, when the resistances are reduced, the capacitors have
to be increased to avoid bass loss. The extra positions will accommodate TO220/
TO126 outline transistors on the board.
They have extra numbers TR7 and TR8 for
the output, and TR6 for the Vbe bias transistor which can be bolted to the heatsink for
TR7. I used 40-year-old design ten-penny
BD135/6 devices, but driver currents can
be reduced by using some of the newer
high current-gain lighting/converter bipolar transistors, such as the Zetex series
and 2SA2039/2SC5706 types. Further improvements are a modulated JFET current
Fig.17. Frequency response of the higher-power version of the PE Theremin Amplifier in source, shown in the circuit; this further
minimises the current in the driver stage.
Fig.12. 1dB down at 20Hz and 20kHz – typical Hi-Fi response.
Also, another bootstrap is employed on the
top of the emitter-follower TR2 for greater voltage swing, using
components R15 and C13. Of course, adding all these extra bits
can be a bit messy, as shown in Fig.13! I hooked this amp up
to an LS3/5A speaker and it sounded transparent and possibly
‘warm’ (a subjective audio word, meaning low-order harmonic
distortion increasing in the low-frequency end).
Your best bet since MAPLIN
Chock-a-Block with Stock
Visit: www.cricklewoodelectronics.com
Or phone our friendly knowledgeable staff on 020 8452 0161
Components • Audio • Video • Connectors • Cables
Arduino • Test Equipment etc, etc
Visit our Shop, Call or Buy online at:
www.cricklewoodelectronics.com
020 8452 0161
Visit our shop at:
40-42 Cricklewood Broadway
London NW2 3ET
Practical Electronics | December | 2020
Distortion measurements
I’ve recently bought an Audio Precision SYS2712 analyser
from Stuart of Reading for £1500. Repair, calibration and the
USB interface added another £1300. This sounds horrendous,
but it’s a tenth of its cost back in 2004. Some people hanker after Apple computers or Mercedes cars, but I’ve always
wanted an AP. This is money well spent to obtaining quantitative measurements of the total harmonic distortion and
noise (THD+N), allowing the effects of circuit changes to be
seen instantly. This instrument will greatly enhance the circuits I lovingly design for the readers of PE and discriminate
the ‘audiofool’ from audiophile components.
Fig.14. shows the relatively high distortion of the low-power amp. It is of no consequence with the PE Theremin and
the small loudspeakers used, both of which have a THD+N
of around 10%. Fig.15 shows the higher-power version, still
technically ‘bad’ but not subjectively noticeable. Spectral
analysis will probably show a lot of second harmonic present because of the asymmetry of the circuit. The frequency
responses are shown in Fig.16 and Fig.17 respectively.
Germanium transistors
Since this is a minimum transistor design it was decided to
investigate the use of some old (now expensive) germanium
transistors. They are supposed to have increased voltage swing
and softer distortion. I will reveal all next month!
49
