This is only a preview of the April 2024 issue of Practical Electronics. You can view 0 of the 72 pages in the full issue. Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
AUDIO
OUT
AUDIO OUT
L
R
By Jake Rothman
Back to the buffers – Part 3
F
ollowing the stripboard
*Add these parts for
48V single-rail operation
R4*
33kΩ
0V
VIN
R2
620Ω
0.92mA
C4
1nF
4.4V
R3*
47kΩ
C1
470nF
R1
100kΩ
R7
4.7kΩ
R8
390Ω
TR2
BC556B
ZD1
3.7V 3.9V
R9
15Ω +
C6
2.2µF
50V
TR1
BC550C
44V
TR3
MPSA29
1.13V
R10
560Ω
0.58V
4.8mA
C8
100µF
25V
R11
47Ω
10.4mA
0V
47µF
35V
0V
R5
10kΩ
–0.8V
C2
100pF
+25V
+ C9
R14
100kΩ
R6
47kΩ
C5 +
10µF
10V
VO
LED1
Red
high-efficiency
1.7V
R13
47Ω
–25V
C10
47µF
35V
0V
+
C3* +
2.2µF
25V
R12
47Ω
16.2mA
+
designs for a discrete buffer in
last month’s Audio Out I have
produced a PCB version; this board
will be available from the PE PCB
Service from next month. All the
versions of the circuit given in the
last issue can be built on this PCB,
but it is most likely to be used for the
high-voltage Darlington version given
originally in Fig.14 (shown again
here in Fig.24 in case you missed it).
Unlike the discrete op amp PCB, the
components are more spaced-out,
reflecting the simplicity of the circuit.
This helps with experimentation,
allows for big audiophile capacitors
and is shown in Fig.25. The single-rail
version is shown in Fig.26.
I’ve also had a few more useful buffering circuit design ideas, which will be
the topic of this and next month’s pieces.
JFET input buffer
I’m sorry, but I just can’t leave anything
Fig.24. Buffer circuit repeated from Fig.14 in Part Two (PE, March 2024).
alone! A new version of the JFET buffer
circuit (originally Fig.22) is shown
0.3% at 6Vrms). Playing with this design
also removed the anti-spike components
in Fig.27. This ‘re-optimisation’ often
reminded me that one of the best things
R9 and ZD1. This was done because the
happens when I receive a PCB because
about discrete circuits is the way the
later stages in most audio systems clip
it’s easier to change parts and test comresistor values can be optimised for any
first. To run the buffer at ±25V, (matchpared to original prototype stripboard
given operating condition.
ing the discrete op amp power rail) the
versions. I found I got better results with
I also increased the rails to ±18V, the
decoupling resistors R12 and R13 can be
the Fairchild/OnSemi J113 JFET than
maximum allowable, since the JFET has
increased to 470Ω.
the BF244A. They are cheaper as well,
a voltage limit of 35V Vds (drain-source
I was surprised to see noise curves for
around 13p each (when buying 100 from
the J111 to J113 JFETs on the OnSemi
voltage). Of course, I then had to change
Mouser). The distortion at 1Vrms into
data sheet, I always thought these devices
all the resistor values. Note that some
were intended for switching rather than
components are omitted, such as the
600Ω was 0.0008% with R8 changed
audio designs. At an operating current of
single-rail biasing parts, R3, R4 and C3. I
to 300Ω. (At 4Vrms it was 0.006%, and
Fig.25. The assembled discrete buffer amplifier.
62
Fig.26. Single-rail version of discrete buffer amplifier. Extra biasing
parts: R3, R4, C3 and SGL supply link installed. C10 is omitted.
Practical Electronics | April | 2024
+17.8V
C4
270pF
3V
1.4mA
VIN
R2
1kΩ
C1
100nF
TR1
J113
R7
2.2kΩ
R8
300Ω
TR2
BC556B
BC143*
Ib
R1
1MΩ
C2
47pF
0V
J113
Top view
Interchangeable
symmetric JFET
D
S
G
2.4V
R5
6.8kΩ
C6 +
22µF
35V
31.4V
IC
8mA
4.8mA
+0.8V
TR3
BC546B
BC141*
0V
*TO5 transistor
for higher IC
R10*
100Ω
+25V
+ C9
0.95V
9.5mA
100µF
25V
0V
C7
220nF
VO
R14
100kΩ
R6
2.2kΩ
LED1
1.8V
Orange
Low-bri
1.8V
–17.8V
15.3mA
*Sets IC
C8
100µF
25V
R11
33Ω
+
IS
R12
470Ω
15.3mA
R13
470Ω
Fig.28. JFET buffer amplifier PCB. Note
the cross-legged mounting of JFET TR1
and that the output transistors TR2 and
TR3 have been upgraded to TO5 devices.
–25V
C10
100µF
25V
0V
+
Fig.27. JFET buffer circuit – the supply rails are ±18V and resistors R12 and R13 allow
it to be used on ±25V supplies. (Overlay diagram will be provided next month.)
1.35mA, the noise voltage of the JFET is
LED1 orange 3mm (or similar with a Vf
2.5nV/√Hz, better than the NE5534 and
of 1.8V at 5mA)
equal to most expensive audio JFET op
ZD1 omitted
amps. By increasing the current up to
10mA, the noise can be reduced to just
Capacitors
1.3nV/√Hz, not bad at all.
C1 100nF 5mm polyester 10%
C2 47pF 2.5mm ceramic 10% NP0
C3 omitted
JFET buffer assembly
C4 270pF 2.5mm ceramic 10% NP0
The JFET buffer assembled on the PCB
C5 10µF 10V 2.5mm pitch radial
is shown in Fig.28. Note the link for R9
electrolytic
and the omitted components. Sadly, the
C6 22µF 35V 2.5mm pitch radial
pinout of the J113 JFET was different,
electrolytic
not centre-gate, so a ‘leg wiggle’ was
C7 220nF 2.5mm ceramic 20% X7R
needed. Note also that the board can
C8 100µF 35V 5mm pitch radial
accommodate TO5-cased transistors for
electrolytic, non-polarised preTR2 and TR3, as shown. This allows a
ferred. Nichicon UEP1E101MPD
higher output stage current if desired.
from Mouser.
It’s also much easier to fit heatsink clips
C9, C10 100µF 25V 3mm pitch two off
if needed.
JFET buffer component list
Semiconductors
TR1 J113 N-channel JFET
TR2 BC556 PNP small-signal bipolar
TR3 BC546 NPN small-signal bipolar
High impedance
does not load
volume control
VIN
Resistors
R1
1MΩ
R2
1kΩ
R3, R4 not used
R5
6.8kΩ
R6, R7 2.2kΩ
Low input impedance
that also varies with
frequency and
control setting
Baxandall
tone control
Buffer
VO
CW
Balanced
input
3
1
Now that we have a good discrete op
amp and buffer circuit, we can start
combining them into useful audio systems. The classic use of a buffer in Hi-Fi
pre-amplifiers is to isolate the low and
changing input impedance of a Baxandall
tone control from the volume control, as
shown in Fig.29.
I mentioned in the last issue that
another pro audio use is to buffer the
inputs of a differential op amp circuit
to provide equal high input impedance for good CMRR. This is shown in
Fig.30. The op amp resistors have to be
kept low for low noise, resulting in a
difficult-to-drive low input impedance
that needs buffering. Another problem
is that the input resistances of the differential op amp circuit are unequal if
they are the same value resistors on both
inputs. I changed the values to equalise
this, giving equal loading to the buffers.
This ensures the even-order distortion
harmonics produced by the buffers are
equal. These harmonics cancel out in
0V
Buffer
Treble
Bass
Tone controls
Fig.29. A common use of a buffer in audio is driving a filter
network such as a Baxandall tone control. Filters need to be
driven from a low impedance and log volume potentiometers
need to be loaded by a high impedance.
Practical Electronics | April | 2024
R is a low value for
low noise ≤1kΩ
R
R1 = R2 = high
R2
input impedance,
typically 10kΩ to 1MΩ
R
R
–
Low impedance
Unequal impedance
Low output
impedance
0V
Volume control
typically 10-50kΩ
Balanced buffer
Buffer
–
+ R1
2
0V
R8
300Ω
R9
link
R10
100Ω
R11
33Ω
R12, R13 470Ω
R14
100Ω
+
VO
Differential
op amp circuit
R
0V
0V
Fig.30. Instrumentation amplifier arrangement of a buffered
differential amplifier. This gives inputs that are high impedance and
equal impedance. The buffers also allow the resistors to have low
values, minimising noise.
63
Negative resistance?
While making these changes I was surprised
to find that the input resistance on the inverting input of the op amp was less than
the input resistor of 1kΩ. How could this
be? The answer was that the same voltage
present on the non-inverting pin of the
op amp was also present on the inverting
pin, which is always the case with linear
op amp circuits with negative feedback.
This voltage is 180° out of phase compared
to the inverting input terminal voltage.
This increases the overall voltage across
the resistor, giving it effectively a lower
value (the opposite of bootstrapping) as
shown in Fig.32. I couldn’t find this effect
in differential amplifiers mentioned in
any textbook, but like many odd things in
analogue design, it’s still just Ohm’s law.
Fig.31. Distortion curve of differential amplifier with two JFET buffers on the inputs.
Input 1Vrms output 3.2Vrms (9.1Vpk-pk) into 600Ω.
Voltage across this resistor is 1.5Vpk-pk
Hence, I = 1.5mA
Resistance for 1.5mA with 1V = 667Ω not 1kΩ
1Vpk-pk
ZIN = 667Ω
VIN–
VIN+
1Vpk-pk
ZIN = 2kΩ
Balanced
input
3
1kΩ
1kΩ
1kΩ
0.5Vpk-pk
–
0.5Vpk-pk
+
1kΩ
1
VO
Op amp wired as
differential amplifier
0V
0V
–
+
2
JFET
buffer
ZIN = 667Ω
3.3kΩ
R20
1kΩ
–
DC-coupled
outputs
R19
130Ω
JFET
buffer
ZIN = 650Ω
(with CMRR
preset in
middle position)
In R8
position
VO
+
470Ω
100Ω
Discrete op
amp PCB
Offset
trim
CMRR
trim
0V
Fig.32. There is an interesting effect related to the input
impedance of the differential amplifier; both input signals are
equal and anti-phase. The 1kΩ resistor on the inverting input is
reduced to an effective value of 667Ω or 2/3 of its value. This
assumes all differeerntial amplifer resistors are equal.
Fig.33. In comparison with the circuit in Fig.32, the actual resistor
values used take into account the resistance-reduction effect
mentioned in the text. There is also a gain of 3.2, and the input
impedances have been equalised. Note the preset for CMRR
adjustment and the resistor values are low for noise minimisation.
the differential op amp. I also decided to add a CMRR trimmer (PR3) in the network. Looking at the final distortion
curves for the whole system in Fig.31, I think all this fiddly
tweaking and experimenting was worth it. The cancellation
effect achieved was 8dB reduction on the second harmonic.
DC coupling
The buffers can be DC coupled to the op amp inputs if desired
since their +0.81V offset can be easily be rejected by the differential op amp. The offset adjust preset PR1 will need to be
tweaked to bring the output of the whole system to zero. To
do this, omit the bipolar DC blocking
capacitor C8 from both buffer boards.
By omitting these coupling capacitors,
the CMRR will be maintained at low
frequencies since electrolytics have
poor tolerances, resulting in possible
mismatching. However, there is the safety
issue of a hard DC offset being transmitted to the output if there is a wiring
error or fault in one of the buffers. This
could then damage what is connected to
the output. Of course, the output could
be AC coupled, or have a DC detection
circuit and muting relay put in. The final
circuit is shown in Fig.33.
The assembly of the three PCBs is
shown in Fig.34. They will go in a nice
box one day to make a top-notch balanced
headphone amp.
Next month
Fig.34. Connecting two buffer boards with a differential op amp to make an
instrumentation or headphone balanced input amplifier. I hope to tidy this up and
build it into a decent enclosure.
64
We will conclude our discrete buffer
journey next month by looking at an op
amp buffer circuit and an interesting
design called the ‘diamond buffer’.
Practical Electronics | April | 2024
|