This is only a preview of the July 2021 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
Microphone Preamplifier (for Vocoder) – Part 3
Fig.22. The assembled microphone preamplifier PCB – note the links for NPN version.
IN+
IN GND
IN–
+
SK1
R29
R4
REX3
L2
R1
POT3
POT1-2
(link wiper and
CCW at pot)
SK2
D
3
TR1
c
C4
R6
e
C2
C
5
D
6
D5
D4
R8
R10
R14
R13
R16
C7
VR2
VC1
C22
C21
C
10
R11
R13
R22
L3
R15
R5
C3
C24
+
D2
D1
b
TR2
REX2
R3
R9
R7
b
c
Gain pot
C6
C23
e
L1
Most small components are mounted on
the PCB, which is a standard, easy-to-repair, double-sided plated-through-hole
design available from the PE PCB Service
C1
R2
R
28
Construction
(code AO-JUL21). Mike Grindle has
created a clear mirror-imaged symmetrical
design. The overlay is shown in Fig.21 and
the capacitors are dual outline, so either
standard radial or expensive axial types
can be fitted. Stuffing the board should
be in the following order: resistors, chip
sockets, small capacitors, preset, transistor
holders, big capacitors, Molex connectors, and finally the trimmer capacitor,
C8
C20
R20
R
17
IC1
R
18
C19
+
C
11
C9
SK3
C13
+
C
18
R19
IC2
R26
R25
R21
C
19
R27
C15
+48V
–15V
GND
+15V
Power
C12
R23
L5
Output
R24
C14
+
L4
C
16
SK4
Transformer out
Unbalanced out
GND
Transformer
return
+
Balanced
input
just go noisy. Reverse-polarised tantalum
capacitors may go short circuit, killing
connected components. So, pay attention
to polarity and component orientation!
+
to carry out the construction
and testing of our microphone
preamplifier. The circuit is mainly
direct coupled and has lots of polarised
components, so mistakes can cause
damage. For example, if D1 is reversed,
the transistor base will be connected to
the power rail and the transistor and op
amp may be destroyed or less obviously,
+
W
e’re now in a position
Fig.21. Microphone preamplifier PCB overlay – take care with links and orientations if using different transistors to the BFW16A.
Important – in red boxes only use one link: solid for NPN and dashed for PNP version of the microphone preamplifier.
50
Practical Electronics | July | 2021
n
Fig.23. Occasionally the input capacitors
C1/C2 (here C2, at bottom) may have
values 20% apart. In this case, an extra
2.2µF capacitor (C24, shown here above
C2) was added in parallel to get good
matching and hence high CMRR at 50Hz.
which has a delicate adjustment screw.
Several views of the completed PCB are
shown in Fig.22 to Fig.25.
The big, mechanically stressed components are all panel mounted and
hard-wired to the board using 7/0.1
stranded wire. No PCB-mounted pots
or external plug connectors here, which
often cause broken soldered joints when
used on a studio floor. I just love diecast boxes (see Fig.26), especially those
by Hammond and Eddystone and use
them for most of my small-batch builds
because they are easy to cut and drill
(much nicer than plastic or steel). Surrey
Electronics always used them for their
range of microphone preamplifiers and
broadcast products.
A couple of other assembly points:
XLR connectors have earthing tags which
are useful for enclosure earthing (see
Fig.27). Note how the output transformer
is mounted off the board using a capacitor clip shown in Fig.28.
Fig.24. Note the green 82µH inductor,
which looks like a strange resistor. The
odd things that look like back-up batteries
are Plessey Castanet tantalum capacitors.
Fig.25. A thermal link was tried to ensure
both transistors drifted together to minimise
DC offsets. It was not worth it in this case
(but is worthwhile on synthesisers).
Testing
away. A solution used at Kemo Filters is
to put 560Ω resistors in series with any
grounded input pins or those connected
to the output pin, such as in buffers, as
shown in Fig.29.
Once powered up, check all transistor
voltages and make sure the outputs of the
op amps are all at zero volts, apart from
the IC2 outputs which should be +0.2V
approximately (due to the input bias current flowing through the 680kΩ resistors
R11 and R12). If the DC conditions aren’t
right, the audio won’t be right.
DC conditions
At the testing stage – if possible – use a
tracking plus and minus dual supply. This
should be current limited to approximately 100mA. It’s not nice when expensive
components go up in smoke or eyeballs
are damaged from flying chips when errors
and high currents are combined.
The circuit runs at a standard supply of
±15V, with the transistors taking 4.2mA
each and the 5532s 9mA. The total current
consumption is 22 to 25mA. (The 5532s
and transistors can work on an absolute
maximum of ±22V supply, taking 30mA.)
An anomaly with the 5532 is that it can
latch up if the rails don’t rise together at
turn-on. This seems to be exacerbated when
capacitors are connected to the input pins,
which is the situation here. I was puzzled
because other op amps don’t do this.
When I switched the power supply from
independent to tracked, this issue went
Fig.26. A single-channel microphone preamplifier fits in this die-cast box, size 187 ×
120 × 55mm. (Note the PCB shown above is a slightly different prototype).
Practical Electronics | July | 2021
Power supply
Naturally, low noise can only be achieved if
the power supply is low noise. A standard
audio LM317/337 dual-rail circuit will do
the job. The 48V supply is more specialised
requiring a voltage doubler perched on top
of the positive rail. My LM317 circuit (see
Audio Out, PE March 2019) is suitable – it
has low noise and while you can certainly
build your own, I do still have some PCBs
– see my Audio Out Shop advert on p.43
for contact details. Alternatively, a TL783
regulator will work.
It’s best to keep the mains transformer in
a separate screened (earthed metal) box to
minimise mains buzz.
The circuit needs very low impedance
supplies, since the power supply rejection ratio of the input transistor stages
is very low. The emitters and collectors
are connected to the power rails only
Fig.27. Metal cases can be earthed off XLR
sockets via an earthing tag on the rear of
the socket connected to the PSU 0V.
51
–
5 5 3 4/ 2
+
R
R
IN
F
–
5 5 3 4/ 2
+
0 V
Fig.28. The output transformer is neatly
clamped in a 45mm capacitor clip.
Fig.29. Inserting 560Ω resistors to
prevent latch-up with 5532/4 op amps.
by resistors, providing an easy entry
route for supply ripple. The best way
to ensure a low-impedance supply is
by connecting directly to the outputs of
the regulators. In one design, I had 22Ω
decoupling resistors feeding the board
and this increased the distortion into a
600Ω load by 0.5% at low frequencies.
both sections wired in parallel), and finally
74dB (×5000) with the gain control fully
clockwise. In the middle of rotation, it was
39dB. Check the clipping is symmetrical
and the response is flat from 20Hz to 20kHz.
Noise check
Selection of the fittest has always been
part of audio evolution, especially input
transistors, which is why I recommend
using sockets for circuits where transistors need to be selected for noise level to
screen out poor devices (see Fig.33). Use
a ‘scope to look at both outputs of IC2
with the gain control (VR1) set to maximum, that way you can see the noise
level of each transistor and pick the best
ones by comparison. Do turn the power
off when swapping the transistors and
leave the input unloaded to achieve maximum noise for this test.
Gain check
Check the overall preamplifier gain by
feeding a signal generator into the input
via a small transformer or other balancing
method. I use the video isolation transformer shown in Fig.30. The gain should
be around 15dB (×5.6) with no Rg and
20dB (×10) with Rg = 3.69kΩ (actual value of the Blore Edwards 5kΩ dual pot with
Square wave check
Applying a 1kHz square wave is always
a good idea to check for ringing. This is
necessary to optimise the Zobel network
on any transformers used (See Fig.41 next
month). I also found a bit of ringing on the
input chokes L1 and L2; putting 1.3kΩ resistors across them on the spare inductor
holes as shown in Fig.31 damped it out.
These resistors are labelled R28 and R29
on the latest board but are not shown on
the circuit diagram (Fig.18).
The noise should also be checked by
putting two terminating resistors of 75Ω on
the input pins joined together (pins 2 and
3 on the XLR socket) as shown in Fig.32.
This will give a source resistance of 150Ω,
mimicking a low impedance moving-coil
microphone. At full gain, the noise on the
output should be around 12mVpk-pk, with
the low-cut filter in. (The low frequency
noise is inaudible and makes it harder to
read as the trace jumps about). With a metal
housing, battery power and careful selection of transistors it should be possible to
get the noise level lower. I’ve built various
‘resistor terminating units’ into male XLR
plugs. I always recommend using metal
XLR plugs to provide screening.
Fig.30. When testing balanced-input
circuits it is necessary to use an isolating/
balancing transformer between the signal
generator output and preamplifier input.
Common-mode rejection ratio
(CMRR ) check
The CMRR can be checked by driving
both inputs together in phase to see what
comes out of the output, ideally nothing
(see Fig.32). Unfortunately, the CMRR varies with the gain setting, a disadvantage
of transformerless techniques. Because of
this, I suggest checking the CMRR with the
gain at around 30-40dB (two o’clock on the
gain control) with an input of around 1V
to 3Vpk-pk. Because the gain control (Rg)
only boosts gain on common-mode signals,
both outputs from IC1 will remain almost
unchanged at around 5Vpk-pk until the last
bit of the maximum gain region. This is
why the middle gain range is suggested
for CMRR adjustment. If the unit is to be
used at a fixed gain, say in a broadcast application, the CMRR should be optimised
for this. First adjust trimmer VR2 at 1kHz
for minimum output, then do a null at
9kHz with the trimmer capacitor VC1. The
low-frequency CMRR at 50Hz can be optimised by putting 220nF to 2.2µF padder
capacitors (C23 and C24) alternately across
input capacitors C1 and C2 until a null is
achieved (see Fig.23). I suggest these frequencies because most of the noise that’s
a problem in the field is mains related,
I np u t f r o m si gnal
gener ato r f o r
C M R R test .
Si gnal gener ato r
s o u r ce i mp edance
1%
1%
3
1
2
M al e X L R p l u g
(r ear vi ew )
Fig.31. To reduce a minor amount of ringing
on square waves, the input filter chokes L1
and L2 can be damped by wiring 1.3kΩ
resistors across them (R28 and R29).
52
Fig.32. XLR termination unit provides
the correct source impedance of 150Ω
and enables a signal generator to be
connected for CMRR testing.
Fig.33. Transistors should be mounted
in sockets so noisy ones can be easily
removed. (It’s also good for finding
unknown gems in one’s parts box)
Transistors from Hitachi, Toshiba, Sanyo
and Rohm seem to be the best. Only home
constructors have time for this, so it is
possible to improve on ready-made units.
Practical Electronics | July | 2021
mainly switching noise from power supplies and lighting controllers.
Surprisingly, I found the adjustment
range for the trimmer was better if C13
(the trimmer parallel capacitor) was reduced to 120pF when the circuit was
transferred to PCB. (There may be some
stray capacitance associated with the
board layout, but I doubt it’d be over
100pF. More investigation is needed!)
On my boards I used C13 values ranging
from 68pF to 180pF to keep the null of
VC1 in the middle position. In the parts
list I specified an 80pF trimmer, but I
now recommend using a larger value,
such as 250pF.
Mods and embellishments
Vinyl virtues
This circuit would make an effective
moving-coil (MC) pick-up amplifier with
RIAA equalisation incorporated, but I
would have to convert my record deck
to balanced output. This would make
it non-standard in the Hi-Fi world, but
ideally, all audio equipment should be
balanced – even electric guitars! The
board could be used for an unbalanced
stereo moving-coil pickup head amplifier
by just using the first stages and omitting
IC2 and its associated components.
+ 48 V
+ 48 V O N
1N 4148
B S170
B al anced
i np u t
470 nF
10 nF
0 V
Fig.34. Switching on phantom power
can produce ear-splitting cracks unless
the rise time of the voltage is slowed.
This can be done with a 100Ω resistor
and 470µF capacitor or the capacitor
multiplier circuit shown above.
‘standard’ 15% C reverse-log-law pot give
too abrupt a jump in gain at the limit of
the rotation, and is therefore not optimal.
A better law is the 8% RD law produced
by Alps. (se Fig.35) That Corporation decided 2.5% was best, avoiding the sudden
up-lift in gain. Such a pot would have
to be specially made and be expensive.
I should point out here that cermet and
wire-wound pots have bad rotational
Phantom power
noise, the wire-wound type sounds like
Switching phantom power can make a loud
undoing a zip! If you want to read up in
noise. It is best to slug the rise with a big
more detail on potentiometer law details,
electrolytic capacitor or MOSFET capacithen see my articles from Audio Out, Notance-multiplier circuit, as shown in Fig.34.
vember 2015 and April 2018.
The alternative to pots is switched gain,
Gain switch
which enables each step to be optimised
Instrumentation-type amplifiers (as used
and allows the use of low-noise metal-film
in this design) use reverse-log-law gain
resistors. Cadac mixers use a 21-way switch
control pots. Unfortunately, they are difand Neve preamplifiers often use mulficult to source. – but it gets worse! The
tiway switches. The table
opposite shows resistor values required for 2dB steps for
the biggest 29-way switches
available and Fig.36 shows
its construction. If you have
fewer steps, it’s best to have
the smallest increments on
the highest gain settings.
That Corporation makes
a special SMT chip for use
with their 1570 microphone
amplifier chip that includes
electronic switching and a
servo to control DC clicks,
called the 5171. This has to
be connected to a microcontroller, but it does give 1dB
steps. Anyone fancy designing a rotary encoder circuit
for it? Do note, however,
Fig.35. Alps reverse log pot tapers. This shows
that both chips are expenvoltage/resistance vs rotation. At 50% rotation, a linear
sive (1570, £6 / 5171, £11
potentiometer split point would be would be 50% / 50%,
from Mouser). That said, I
taper C is 82% / 18% and RD is 92% / 8%.
Practical Electronics | July | 2021
Fig.36. Test your soldering skills: a 29-way
Painton switch with resistors attached, all
different values. Note the ring of 22swg
tinned copper wire that forms one terminal
of the gain control; the other is the wiper. I
often repurpose these high-quality specialist
switches from old, scrapped test equipment.
do worry about using single-sourced specialist chips, but anyone can make up a
switch with a load of resistors on it.
Resistor values for stepped
gain switch
Step
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Gain
(dB)
15
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58
60
62
64
66
68
70
72
74
Ideal
Resistance
open
3.69kΩ
2.7kΩ
1.88kΩ
1.6kΩ
1.17kΩ
790Ω
600Ω
470Ω
390Ω
320Ω
220Ω
178Ω
138Ω
109Ω
85Ω
66Ω
51Ω
40.4Ω
31.9Ω
23.6Ω
18Ω
13.2Ω
9.8Ω
7.1Ω
5.1Ω
2.7Ω
1Ω
short
E24
value
open
3.6kΩ
2.7kΩ
1.8kΩ
1.6kΩ
1.1kΩ
750Ω
620Ω
470Ω
390Ω
300Ω
220Ω
180Ω
130Ω
110Ω
82Ω
68Ω
51Ω
39Ω
30Ω
24Ω
18Ω
13Ω
10Ω
6.8Ω
5.1Ω
2.7Ω
1Ω
short
Next month
We’ll conclude in Part 4 with optional modifications, including the use of transformers.
53
|