This is only a preview of the August 2020 issue of Practical Electronics. You can view 0 of the 72 pages in the full issue. Articles in this series:
|
AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
Low-noise Theremin Power Supply – Part 1
(CE, known cynically as ‘Chinese
Export’) don’t care about interference
emitted below 1MHz. This is one reason
there are so many disillusioned audio,
and low-frequency radio engineers.
Low-noise design
Fig.1. The completed Theremin power supply board, delivering +9V and +15V.
O
h no, not another power
the pitch. It’s even worse if the PSU is a
switch-mode design. Many SMPSs have
4.7nF Y capacitors connected from the
mains input to 0V, guaranteed to make
Theremins have an absolute fit. For most
electronic equipment – such as PCs, TVs
or printers –
PSU RF (radiofrequency)
emissions
don’t matter.
The European
regulations
supply unit!, I hear you say.
‘You can buy them online for
a fiver and I’ve got 20 in the kitchen
drawer’. True, if it’s noisy, un-earthed
black plastic ‘wall-warts’ destined for
landfill that you want. Plug such a PSU
into a Theremin, AM radio or Hi-Fi preamplifier and their performance will be
degraded. Theremins generate a horrid
50Hz warble sound as the mains-related
electromagnetic interference (EMI)
emissions from cheap PSUs modulate
There is one way out of this modulation misery and that is to build your
own power supplies. In the quiet of the
present lock-down, the uniquely lownoise of this design shown in Fig.1 will
be even more noticeable.
The simplest way to reduce EMI-affecting circuits is distance, because fields
obey the inverse-square law (double
the distance, the intensity goes down
four times (ie, two squared)). Although
I dislike external PSUs, there is no substitute for distance to reduce noise. If
you are building expensive audio/music
gear, where an internal PSU is expected, then the use of a screened toroidal
transformer is almost mandatory. For a
Theremin housed in a big or long box,
it’s often possible to position the power supply out of the way.
Down to earth
To deliver maximum playing range, a
Theremin should be earthed to complete
Mains transformer
L
* snubber capacitors
typically 10nF to 100nF
Mains
input
*
*
–
+
*
2200µF
* 25V
Bridge
rectifier
V+
+
N
Smoothing
capacitor
0V
Fig.2. Snubber capacitors connected across the diodes in a
bridge rectifier suppress switching spikes.
48
Fig.3. Example of snubber capacitors from an old Robert’s radio –
these are essential for hum-free reception on AM (long/medium wave).
Practical Electronics | August | 2020
n
T o scope
Scope
probe
0V earth
D iode
under
test
L
23 0V
A C
15V
A C
+
Ω
10W
load
resistor
3 3 00µF
25V
N
Fig.4. Simple half-wave power supply used
to compare rectifier diode switching spikes.
Scope
1nF
T o diode
anode
1kΩ
0V earth
Fig.5. High-pass filter on scope probe
blocks 50Hz AC (‘mains hum’).
the capacitive circuit between the hand
and the antenna. Most wall-warts have
no earth at all; often having a plastic pin
in the earth position on the plug, which
easily breaks off. In the UK, this prevents
it being plugged into the mains socket
resulting in more e-waste. My power supply is – of course! – earthed via a proper
three-pin socket and mains plug.
Rectifiers
In the old days of AM radios, the bridge
rectifier diodes in the power supply
A lmost ve rtical
spike w ith lots
of harmonics
Very narrow
lots of R F
–
always had snubber
17 V
capacitors of around
P ow er supply load
10nF to 100nF conL
nected across them
+
(Fig.2) to stop what
Mains
was known as ‘mod+
input
ulation hum’. These
are shown in the 1978
N
Robert’s RM30 table
11V
1nF
radio supply in Fig.3.
These were needed to
Scope
Spare
probe
suppress RF bursts
w inding
produced by a sharp
1kΩ
spike when the diode turned off. I’ve
always added these
T o scope
capacitors as a matter
of habit in all equipment. For this article, Fig.7. An isolation transformer is normally needed to view
I decided to do a bit of switching spikes on bridge circuits. However, if the transformer has
investigation. These an extra secondary winding this can provide an isolated output.
spikes are produced
I had to use an isolation transformby the stored charge from the recombier on the ‘scope input to avoid earth
nation of holes and electrons at the diode
currents with bridge rectifiers to look
junction. This effect is especially proat the noise. It’s easy if the transformer
nounced with old slow silicon rectifiers,
has dual secondary windings, since the
such as the good-old 1N4001.
scope can then be connected to a floating
This problem is especially bad with
unused winding. In this way the mains
half-wave rectification, since the transtransformer becomes its own isolation
former has no load on half cycles where
transformer, as shown in Fig.7.
it is free to ring undamped. To examine
Newer fast rectifiers, such as the
the spikes, initially I used the half-wave
UF4001 (the UF stands for ultra-fast),
rectifier circuit in Fig.4 since it only needs
soft recovery, and Schottky diodes give a
one rectifier. For experimentation, it is
four-times smaller spike and consequentsimpler to change one diode than the
ly less RF noise. Snubbing capacitance
four of a full-wave rectifier. Also, with
is still needed, albeit a reduced amount,
a full-wave rectifier the switch-off diode
to reduce emissions further.
is damped the moment the other diode
When it comes to suppressing these
turns on, so the effect lasts a shorter time.
spikes, surprisingly cheap ‘n’ nasty
It’s a good idea when looking at this sort
capacitors can be very effective beof thing to put a high-pass filter consistcause they have high losses. So those
ing of a 10nF capacitor and 1kΩ resistor
much-derided barrier-layer ceramic disc
on the scope probe to block 50Hz, as
capacitors from the Far East work very
shown in Fig.5. The narrow
well. X7R and Y5V multi-layer ceramic
spike with its low repetition
types are also good. Plastic-film capacfrequency defied photography
itors may need a series 2.2Ω to 100Ω
on an analogue oscilloscope,
resistor to provide a defined loss. The
so it is drawn in Fig.6.
50H z repetition
freq uency
2V
8 0% more
spike w ith
a 1N4 001
Schottky or
ultra- fast diode
1V
0. 7 V
0V off
Fig.6. Illustration of rectifier switching spike.
Practical Electronics | August | 2020
Fig.8. High-pass filtered waveform from a snubbed bridge rectifier circuit.
Notice the resonant bursts and spikes. 400mVpk-pk (100mV/div) 100Hz
repetition frequency .
49
Scope
probe
1nF
This Zobel circuit is optional, since
it is not strictly needed with the Theremin, but is worthwhile for studios
where multiple power supplies are in
use. A plastic film type is best, but a
non-polarised capacitor can be made
from two back-to-back electrolytic capacitors – Ca and Cb on Fig.11. Note the
series connection reduces their value
to half. Solid-aluminium types are the
best choice here, since the high ripple
current won’t dry them out over time.
A third capacitor C3, with six times the
value of the non-polarised capacitor acts,
as a capacitive potential divider. This
develops a few volts across it to drive
the LED. It also maintains the high frequency bypassing effect. A Zener diode
stabilises the voltage at 2.7V to drive the
LED via a current-limiting resistor. On
negative cycles the Zener acts as a normal diode, clamping the reverse voltage
to 0.7V. This means a normal polarised
capacitor can be used. Since most of the
current bypasses it, a tantalum type is
usable. R1 sets the damping and limits
any surge currents at switch-on.
T o scope
1kΩ
0V
Z obel netw ork
L
23 0V
input
15V, 1A
N
1W
Ω
Load should be floating
otherw ise an isolation
4 × 1N4 001
transformer is needed
4 × 100nF
ceramic disc on the scope lead
–
3 3 00µF
25V
+
Load
+
4 . 7 µF
Ω
10W
Fig.9. Adding a Zobel network across the transformer secondary.
Fig.10. The effect of adding the Zobel network – the resonant bursts are damped down
to 80mVpk-pk and the spikes are reduced.
capacitors should be wired across the
diodes to minimise inductance and radiation loop area.
New buzz on the block
Diode switching also excites the resonance produced by the transformer’s
leakage inductance and winding capacitance (Fig.8). This is at a much lower
frequency, typically 10kHz to 80kHz and
can easily be suppressed by a Zobel (series RC) network across the secondary,
shown in Fig.9. This does not affect radios and Theremins, but can affect audio if
multiple power supplies are being used,
causing a buzz due to beating between
their respective frequencies. Typical values would be 1Ω to 100Ω and 1µF to 10µF.
large. A non-polarised electrolytic can be
used, such as those used in loudspeaker crossover networks. Since the current
through this capacitor is quite large at
38mA, you might think this is a waste
of energy. However, because the current
is 90° out-of-phase with the voltage, the
real power loss is quite small. The current can be put to good use to drive an
LED, avoiding the heat produced from
the normal dropper resistor, by using
the circuit in Fig.11. Someone told me
recently that this was a typical mad Jake
circuit! I always like to include one in
every column, otherwise you might as
well be reading a textbook.
Adding spike-snubbing capacitors without
damping resistors boosts this transformer
resonance. The effective damping produced by adding a Zobel network can be
seen in Fig.10. On one transformer, I was
pleased to find adding the Zobel also reduced the high frequency content of the
mechanical hum. Using the transformer
specified here, the Zobel was effective for
both rails when put on one winding only.
This is because the windings are magnetically closely coupled. A network of 2.2Ω
and 10µF was found to be most effective
when wired across the lower voltage
11V secondary. The capacitor has to be
non-polarised and is therefore physically
50
W ire across 11V secondary
Whacky circuit
E q uiva lent to
Ω
Ω
0. 25W
Special offer transformer!
If there’s one thing I detest it’s the scrapping of perfectly sound components. I
bought a whole load of mains transformers
for scrap value that were being dumped
because they had no built-in thermal cutout. I throw more transformers in the bin
due to random thermal cut-out failures
(where they are often embedded in the
windings) than any other cause. I always
insist on an external cut-out, where it can
be replaced. The ‘Right to Repair’ movement is gaining ground and hopefully
practices like embedded cut-outs will
be banned. There will be no shortage of
these transformers, because PCB designer Mike Grindle and I have over 200 in
stock. Even if we do run out, Mike will
quickly edit the PCB for a new transformer.
The transformer used has two secondaries of different voltages, as shown in
Fig.12. Thus, the board has provision
for two separate power supply circuits. Alternatively, the two secondaries
can be wired in series to give a higher
+
C a
15µF*
16 V
C b +
15µF*
16 V
4 7 µF +
6 V
T ant
P rimary
L
Ω
0. 25W
Secondary 1
11V
I Max = 254 mA
( 2. 51W at 9 . 9 V)
Voltage
measured
off- load
23 0- 24 0V
2. 7 V
4 00mW
R ed
N
18 V
I Max = 23 5mA
( 3 . 7 5W at 16 V)
Secondary 2
Fig.11. A whacky circuit? Incorporating an
LED into the Zobel network. This avoids a
heat-dissipating resistor.
Fig.12. The transformer specified has two
secondaries.
Practical Electronics | August | 2020
Fig.13. These excellent
transformers were going
to be dumped – nothing
wrong with them and they
are ideal for our Theremin
power supply.
www.poscope.com/epe
voltage single supply. The transformer
is shown in Fig.13.
It is possible to connect the two positive
supplies to give a dual-rail plus-andminus supply with a centre ground, as
shown in Fig.14. This is not as good as
a proper dual-rail supply with a negative voltage regulator. The 0V reference
can bounce around if too much current
goes from the ‘negative rail’ into ground.
This is because the output impedance of
the regulator is in series with the ground
line. This was how early op amp circuitry was originally powered; we just
designed it so current was never dumped
into ground. It was always arranged to
go from one rail to the other. This is still
good practice today, unless your ground
is a superconductor.
Transformer specification
When using surplus components with
no written specification (such as this
one), it is important to take some measurements. The first aspect to consider
is the transformer’s physical size, which
will give some indication of the power
rating. This is usually specified in terms
of VA for transformers rather than watts
because the current is pulsed when feeding a capacitor-smoothed rectifier. Most
small mains transformers are built up
from standard lamination sizes, so by
looking at the Danbury and Vigortronix
catalogues I determined it was about
8VA. Small transformers like this generally have a regulation figure (how much
the voltage drops on full load current)
of 22%, so the current rating of the secondaries can be determined by loading
with big wire-wound resistors until the
expected voltage drop is reached. Finally, the transformer should be left on for
a long time at estimated full load for a
while to make sure it does not get too hot.
Above around 70°C is too hot; if a smell
of burning polyurethane varnish fills
the air, it is definitely time to turn it off.
Transformer measured specification
Size (mm)
47 × 36 × 40 (w × l × h)
Weight (g)
250
Off-load voltages 18V and 11V.
On-load voltages (at max current):
16V <at> 235mA, load 68Ω
9.9V 254mA, load 39Ω.
With the secondaries in series, 26V was
obtained with a load of 240mA, temp rise
was 45°C above ambient.
- PWM
- Encoders
- LCD
- Analog inputs
- Compact PLC
Next month
That wraps it up for this month. As you
can see, even simple power supplies can
be complicated, or rather they need care
and consideration at the design stage.
Next month, we’ll build the circuit.
+ I nput
+12V
regulator
+
L
- USB
- Ethernet
- Web server
- Modbus
- CNC (Mach3/4)
- IO
O utput
+
- up to 256
- up to 32
microsteps
microsteps
- 50 V / 6 A
- 30 V / 2.5 A
- USB configuration
- Isolated
PoScope Mega1+
PoScope Mega50
+12V
12V
–
23 0- 24 0V
0V
N
+ I nput
+
+12V
regulator
O utput
+
12V
–
E
– 12V
I f using the ‘ special offer’ transformer then use the 18 V secondary for the positive
rail because current draw is uaually higher for the positive rail in most synthesisers.
Fig.14. Connecting two positive regulators to make a plus/minus power supply.
Practical Electronics | August | 2020
- 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
51
|