AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
Theremin Audio Amplifier – Part 4
I
n Part 3 last month, we looked
at some germanium transistor
versions of the PE Theremin
Amplifier. This month, we’ll detail the
component options and see how the
different design strategies performed.
Heatsinks
For the silicon amplifier, standard clipon heatsinks are adequate. Germanium
devices, however, need much more heatsinking. They should run barely warm
to obtain a long life. Bolting the devices
to the base of a metal box or chassis is
sensible. The completed boards and their
heatsinking arrangements are shown in
Fig.12 and Fig.13.
Specifications
The gain of both amps was 16.7 (24dB).
This is about right for running a radio tuner. For CD use, R1 should be
increased to 27kΩ. The germanium amplifier worked down to a supply of 5V
while it was about 6.5V for the silicon.
Power output
As expected, the germanium amplifier gave more power, providing 880mW
RMS (7.5Vpk-pk). The silicon circuit gave
720mW (6.5Vpk-pk). This was less than the
Siemens circuits because of the addition
of an emitter follower (TR2). This stage
does allow considerable mismatching
of the output transistors due to the low
impedance drive. The class-A emitter
follower is run at a high current of 17mA
to allow for the modest Hfe of germanium and old TO5 silicon devices, which
can often be as low as 40. Thus, the total
quiescent current is around 20 to 30mA,
which precludes most battery operation.
The maximum current just clipping was
190mA for the silicon circuit and 220mA
for the germanium one.
Using a germanium output stage, but
with TR1 and TR2 being silicon, gave
810mW (7.2Vpk-pk). As an experiment,
the germanium amplifier was found to
give 1.32W (6.5Vpk-pk) into 4Ω, but only
with good heatsinks.
42
Fig.12. Completed silicon amplifier. Note how bias transistor TR3 is tied to the heatsink of
output transistor TR4, providing a degree of thermal stabilisation of the quiescent current.
Distortion
The crossover glitch was softer and shallower with the germanium output stage
and less quiescent current was needed
(3mA) to smooth it out. The silicon devices
needed around 10mA. With no negative
feedback the germanium output stage
would in theory produce less distortion.
However, the lower open-loop gain of the
germanium circuit meant there was less
negative feedback factor to reduce distortion. This also gave the effect of softer
Fig.13. The completed germanium amplifier. The output transistors must be bolted to
sufficient metal to keep them cool, almost at ambient. If this is done, the bias transistor
can be left unbolted to the heat sink to track the ambient temperature.
Practical Electronics | February | 2021
into this deeper. Interestingly, guitarists
are prepared to pay £700 for small germanium amplifiers such as the Deacy
amps. I suspect it’s because the lack of
clarity exhibited by imperfect amplifiers
hides their mistakes!
Mix it up
n
Fig.14. Distortion for the silicon amp: this is at 3Vpk-pk into 8Ω.
The best result was obtained with a BC557B
silicon input transistor for TR1, while keeping the rest germanium (TR2, 2SA12). The
distortion curve for this set-up is shown
in Fig.17. Making TR2 silicon made the
mid-band distortion a bit worse and gave
less output. Leak did the right thing when
they replaced the OC44 input transistor
with a BC153 in their Stereo 30 amplifier.
Bass response
The capacitors limit the amplifier’s low-frequency −3dB point to around 65Hz. This is
fine for the small paper-coned loudspeakers that are used in low-power applications
(guitar practice amplifiers) that typically
cut-off around 100Hz. For a true vintage
guitar/radio sound I suggest using the
Celestion and Philips 4072 units shown
in Fig.18. The frequency response of the
silicon and germanium amplifiers are
shown in Fig.19 and Fig.20.
Fig.15. Distortion for the germanium amplifier under the same conditions. It can
be seen this is much higher than the silicon amplifier. This is using slow NKT214
transistors for TR1 and TR2.
Noise
The noise on both circuits was 1mVpk-pk
with the input shorted to ground. There
was more high-frequency noise in the silicon design and more low-frequency noise
in the germanium one. The odd germanium device was very noisy and had to be
swapped out from the input stage. Both
units worked well as headphone amplifiers.
Overload protection
The total forward-voltage drop of the
short-circuit protection diode network
(D1 and D2) needed to be 1V. This was
obtained by combining the forward voltage drop of a Zener diode (0.75V) with a
germanium diode (0.25V).
Fig.16. Using fast 2SA12 RF transistors for TR1 and TR2 improved the distortion of the
germanium amplifier. This was a trick Leak used in their Stereo 30 amplifier design.
clipping. The loop gain was mainly reduced by the lower value of R2 used on
the germanium amplifier to reduce the
effect of leakage currents. The result was
that the total harmonic distortion was
2.5-times higher in the mid-band (500Hz)
and 10-times higher at high frequencies
(8kHz) in the germanium amplifier.
The audible effect of these differences could be described as the germanium
amplifier sounding ‘softer and fuzzier’,
with the silicon being ‘clearer but harsher’
when clipped. The germanium amplifier was better for guitar practice and the
silicon version for Hi-Fi. The distortion
curves are shown in Fig.14 and Fig.15.
Practical Electronics | February | 2021
In Fig.16 you can see the effect of using
fast 2SA12 transistors for TR1 and TR2,
dropping the 10kHz distortion from 1.7%
to 0.35%. Technically, the germanium
amplifier was ‘less linear’ than the silicon. However, a silicon amplifier with
the same level of total harmonic distortion would sound worse because of the
higher proportion of high harmonics.
In the end, the physics of bipolar transistor action is the same, an exponential
voltage-to-current conversion process,
whether you use germanium or silicon.
It’s the second-order differences which
affect the overall results. When I get the
spectrum analyser to work we will look
Parts list
PCB – this is the same board used in the
November/December 2020 Audio Out, part
AO-1220-01, available from the PE PCB
Service at: www.electronpublishing.com
Resistors
All 0.25W 5% carbon film (silicon
amplifier resistors in brackets)
R1
R2
R3
R4, R9
R5, R6
R7
R8
4.7kΩ
15kΩ (47kΩ for silicon TR1,
such as BC557B)
100kΩ
47Ω
1.5kΩ
220Ω (1kΩ)
160Ω
43
R10, R13 10Ω
R11, R12 0.39Ω
R14
not used
VR1
VR2
500Ω TO5 outline preset
5kΩ TO5 outline preset
Capacitors
All 10V or higher radial electrolytic unless otherwise stated
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
C11
10µF
100µF
22µF
470µF
100µF
100nF ceramic or polyester
470µF
1000µF
not used
33pF ceramic (Not normally used,
see text.)
10µF
Semiconductors
TR1,2 For Ge amp, use small-signal
high-frequency PNP germanium
OC44, 2SA12, GET872 or audio
devices such as OC71, OC75,
NKT214; for Si amp, use BC549
TR3
For Ge amp, use AC153, AC128
or any medium-power low-frequency germanium device. A
Brimar/Thorn/Mazda AC169 bias
transistor can also be used; for Si
amp, use BC337
TR4
For Ge amp, use AC153K, AC188
PNP germanium power transistor;
for Si amp, use BC138, BC140,
BFY51 TO5 medium-power silicon NPN transistors
TR5
For Ge amp, use AC176K, AC187
NPN germanium power transistor;
for Si amp, use BC143, 2N2905 TO5
medium-power PNP transistors
D1
For Ge amp use low-voltage Zener
3.3V to 12V, 400mW; eg, BZY88C5V6; for Si amp use standard
high-brightness red LED
D2
For Ge amp use germanium
small-signal diode; eg, CG92,
OA91; for Si amp use BAT86
small Schottky diode
Fig.17. A further improvement to the design was to use a BC557B silicon transistor
for the input stage TR1 while keeping the rest germanium. TR2 was a 2SA12 device
and R2 is 47kΩ. This configuration was the best, producing just 0.045% mid-band
distortion. Again, Leak did this as a later mod to the Stereo 30 amplifier.
Fig.18. Suitable speakers (available from the author) to complement the germanium
transistor amplifier. They are particularly suitable for guitar.
Fig.19. Frequency response of the germanium amplifier with slow input transistors
(NKT214) and C10 connected.
Germanium parts available from the author:
Tel
01597 829102
Email jrothman1962<at>gmail.com
Why do it?
For the audio historian and component
obsessive it’s definitely worth the work.
It shows 1970s components did the job,
and five decades on they are still viable.
Fig.20. Frequency response of the silicon
amplifier. The germanium amplifiers using
fast transistors for TR1 and TR2 gave the
same response.
44
Practical Electronics | February | 2021
