Professional Documents
Culture Documents
Millennium Edition: Crescendo
Millennium Edition: Crescendo
Millennium Edition: Crescendo
AUDIO
Crescendo
Millennium Edition
remake of a famous amplifier
Design by T. Giesberts
Measured results (power supply as shown in Figure 3; quiescent current 200 mA)
– input sensitivity: 1 Vrms
– input impedance: 45 kΩ
– sine-wave power (0.1 % THD): 90 W/8 Ω, 137 W/4 Ω
– power bandwidth (80 W/8 Ω): 1.5 Hz – 300 kHz
– slew rate: 60 V/µs (rise time = 1µs)
– signal/noise ratio: 104 dB (A–weighted)
(with respect to 1 W/8 Ω) 96 dB (BW = 22 kHz, linear)
– harmonic distortion at 8 Ω: at 4 Ω:
(bandwidth 80 kHz): at 1 kHz: 0.002 % (1 W) 0.0026 % (1 W)
0.0017 (40 W) 0.004 % (80 W)
at 20 kHz: 0.028 % (40 W) 0.04 % (80 W)
– intermodulation distortion: 0.0017 % (1 W) 0.003 % (1 W)
(50 Hz : 7 kHz = 4 : 1) 0.004 % (40 W) 0.007 % (80 W)
– dynamic IM distortion: 0.0026 % (1 W) 0.003 % (1 W)
(3.15 kHz square wave with 0.0014 % (40 W) 0.0023 % (80 W)
15 kHz sine wave)
– damping factor (at 8 Ω): 460 (1 kHz) 1
a
330 (20 kHz) 0.5
– open–loop parameters:
gain: 4,000 0.2
%
40W
protection: 0.02
0.001
20 50 100 200 500 1k 2k 5k 10k 20k
Hz 010001-A
From the number of zeros after the decimal point, you can see in a 1
b
single glance that this is an exemplary set of results. You will not 0.5
often come across a better set of figures. The distortion is very 0.2
low, the damping factor is very good and the slew rate can even be
0.1
said to be remarkably good.
As you may expect, we have also measured a number of curves 0.05
20 kHz is minimal, but at the 50% power level (40 W is equivalent 0.001
to 70% of the maximum output amplitude) the effect of the non- 0.0006
linear input capacitance of the MOSFETs can be recognised. 1m 2m 5m 10m 20m 50m 100m 200m
W
500m 1 2 5 10 20 50
010001- B
100
8Ω
increase in the distortion can be seen above 10 W, but a level of 100
Figure C shows the maximum output power into 4-Ω and 8-Ω W
whose harmonics lie below –100 dB. The 2nd and 3rd harmonics lie 1
20 50 100 200 500 1k 2k 5k 10k 20k
at negligibly low levels (–118 dB and –115 dB, respectively). Hz 010001 - C
D1 R17 R8 R9 D3 R25
C6 C8 C12 C13
270Ω
1k
1k
33Ω
2SJ201 MJE340
100µ 100n 1000µ 2SK1530 2SK537
25V 220µ 25V 3V9 63V MJE350
T5 0W5 T7
R16
BC556B BC560C
22k
R18 R10 C4
8k2
330Ω R30
R24 E B
1n G S G S
22k
T12
10k
D D C
MJE350 T8
T1 T2
bias R5 R4 R32
1M8 1M8 2x
R29
220Ω * see text
120Ω
C3
R6 R7 tp1 * siehe Text
47Ω
R34 L1
0Ω22
100n 2SK537 tp3
C1
* LSP+
R3 R23 R37
470Ω 12k 1Ω
R28 R35 R36
2µ2 R22
270Ω
0Ω22
10Ω
R11 R12
470Ω
*
47Ω
47Ω
R1 R2
C2 tp2
1M
47k
CW
1n P1 C11
1k T13
T3 T4
R33 10n
* LSP–
2x 220Ω
R26
BC556B R31
10k
R21 R15 C5
8k2
R19 330Ω
1n
22k
T6 T9
BC546B
R20 R13 R14 BC550C
D4
C7 C9 C14 C15
270Ω
1k
1k
33Ω
– 45 ... 50V
010001 - 11
Figure 1. In the schematic diagram, the changes from the original version are hardly noticeable at first glance.
There are surely not very many circuit protective circuitry to be a major between the schematic diagrams of
designs that continue to enjoy such a high weakness. the old and new versions are mini-
level of interest more than ten years after Consequently, in honour of our mal. The design still consists of an
their original appearance, as does the anniversary, we decided to take input stage with dual differential
Crescendo power amplifier from 1984. In part, another look at the original design. amplifiers and current sources, a
this is due to its completely symmetrical Our objective was to update the cascode driver stage and a MOSFET
design, which was in fact an unusual feature design of the amplifier in a way that output stage. That may have been a
at that time, but unquestionably it is mainly would eliminate the sources of criti- rather sophisticated design in 1984,
due to the use of power MOSFETs in the out- cism without sacrificing the good but nowadays it would more likely
put stage. A lot of people happen to be fer- characteristics of the original design. be described as a ‘minimal design’.
vent fans of these devices. Even people who This objective has been quite suc- There’s nothing wrong with this, by
swear by valve amplifiers and are allergic to cessfully achieved. In addition, we the way, since attempting to keep
anything with ‘semiconductor’ in its name were able to obtain such a generous the signal path as short as possible
often have a weakness for MOSFETs, and are level of output power using a new is certainly not a mistaken endeav-
thus prepared to make an exception for them. pair of MOSFETs that it is not neces- our in an amplifier design — but we
Sadly enough, most of the problems with the sary to split the new Crescendo into don’t need to dwell on this point.
amplifier in question had to do with the MOS- ‘light’ and ‘heavy’ versions. Since the basic concept of the origi-
FETs. The original types have long since gone nal design has been retained, any-
obsolete and become unavailable, and suit- The same concept one who compares the schematic
able replacements are hard to find. However, diagram of the new version (see Fig-
there were also other difficulties. The stabil- Since we have intentionally tried to ure 1) with that of the old version
ity of the amplifier sometimes gave cause for change the old amplifier design as (May 1984) will first have to try to
concern, and users considered the absence of little as possible, the differences find the differences. Of course, there
are indeed differences, and it seems 1.5 Hz. C1 is also needed to isolate too much for the transistor types used for T8
like a good idea to list the most the DC bias of the input stage. The and T10 in the old version, they have been
important changes before diving into combination of R3 and C2 forms a replaced by the somewhat more robust types
a detailed description of the low-pass filter that is dimensioned MJE340 and MJE350.
schematic diagram. for a frequency of more than 300 kHz. Now we come to the output stage. In contrast
The most evident change is naturally This helps prevent TIM (transient to the MOSFETs used in the old version, the
the new pair of MOSFETs in the out- intermodulation) distortion and elim- 2SK1530 and 2SJ201 devices used here have a
put stage. The Toshiba 2SK1530 and inates possible RF interference. positive temperature coefficient. This means
2SJ201 are readily available, and fur- The dual differential amplifier that with a constant gate-source voltage, the
thermore they can dissipate so much (T1–T4) has been designed to work drain current increases with increasing tem-
more power than the original with a bias current that is approxi- perature. This made it necessary to use a dif-
devices that we were able to boost mately three times a great as that of ferent design for the quiescent- current cir-
the output power of the old ‘Mini- the original design, on account of the cuit. Here the MOSFET T11, which is
Crescendo’ by a factor of nearly two increased output power. The current mounted on the same heat sink as T8/T10
(90 W into 8 Ω in place of 50 W) sources that regulate this setting, T5 and T12/T13, provides the necessary com-
using only a single pair of transis- and T6, now use LEDs as references pensation.
tors. (D1 and D2), since this results in less Finally, there are a couple of other significant
As a consequence of the increased noise than using Zener diodes. In the items.
power level, the bias currents of the interest of the thermal stability of the Insiders will notice that the none-too-attrac-
various stages must be modified and DC setting, D1/T5 and D2/T6 are tive bipolar electrolytic capacitor has been
different transistors must be used in thermally coupled, as are the tran- eliminated from the reverse feedback network
the cascode stage, as will be seen sistor pairs T1/T2 and T3/T4. (R22/R23), which means that DC coupling is
later on. The bias currents of the cascode used here. To get rid of the resulting output
The next change is the addition of stages T7/T8 and T9/T10 are also offset, we have provided an automatic com-
the networks R10/C4, R15/C5 and significantly greater than in the orig- pensation circuit that is located on the pro-
R30/R31, which represent the results inal design. Since this would be a bit tection circuit board. We anyhow intended to
of measures that have
been taken to optimise
the stability of the
amplifier. A very impor-
tant final item is that
the amplifier has been
provided with reliable
protection circuitry and
automatic offset com-
pensation, by means of
an extra printed circuit
board.
This pretty well covers
the most important
changes.
Schematic
details
Now that we’ve seen
the global picture, it’s
time to take a more
detailed look at the cir-
cuit diagram. Let’s start
at the beginning, which
is of course the input
stage.
The design of the input
filter is more or less
standard. R2 (with R1 in
parallel) determines the
input impedance, and
in combination with C1
it forms a high-pass fil-
ter that blocks frequen-
cies below around
V+
V+
R3 R8 R18 R22 R23
C1 C5
18k
5k6
12k
15k
680Ω
150p T3 T7 T9
47µ
63V
R7 T8
BD140
1k
BC556B BC516
R12 R17 R19 R21
330Ω
1M
330k
220k
T1 Re1, Re2 = G2R-1-E
tp1 R1 BC556B
390Ω
D1 Re1
R2 Amp.
MJE340
120Ω
R20
220k
T5
tp3 R11 R13 1N4148
10k 1M
R5 C3 D2 Re2 LSP
39k
150k
120Ω
R15 R14
T6
22µ 63V
T2 C4 V+
tp2 R4
1N4148
390Ω 2x
22µ 63V BC546B R28
470k
MJE350
R6 R9 R16
D7 35V
18k
1k
220k
T10
R30 R31
1N4004
T4 IC2 10k 10k
D8 35V
5 6 1
R29
R10 BC546B C9
1N4004
3k3
C2
5k6
BC546B
1µ
150p 63V
4 2
V– CNY17-3
V- R24 R26
+20V
V+ +20V
1M
4k7
C6
D3
C7
2µ2
1 220µ
7 20V 0W5 25V
2
D5 D6 8 R25 bias
6
IC1 10k
2x 3 D4
C8
BAT85 4
OP77 220µ
R27 20V 0W5 25V
V- 4k7 –20V
–20V
010001 - 12
Figure 2. Schematic diagram of the added protection circuitry and offset compensation circuit.
use the compensation circuit to correct for the R10/C4 and R15/C5, the modification information to the protection cir-
offset caused by the unavoidable asymmetry made using R30/R31 ensures that cuitry: tp1 and tp2 convey the volt-
of the input stage. The necessary compensa- the amplifier is unconditionally sta- age across the emitter resistors,
tion circuit consists of nothing more than an ble, so much so that the standard while tp3 conveys the output volt-
opamp wired as an integrator, which mea- Boucherot network (R36/C11) can age. The actual protection takes
sures the output voltage of the amplifier and even be omitted. place with the help of two relays
provides the proper amount of reverse current (Re1 and Re2), whose switching con-
feedback to the (bias) input. Thanks to the Protection tacts are connected in parallel in
very high values of R4 and R5 and the decou- order to keep the insertion resis-
pling provided by C3, this correction has The protection circuitry (Figure 2), tance as low as possible. The relay
absolutely no effect on the audio signal. which is located on a separate contacts are wired in series with the
Another essential detail is that the open-loop printed circuit board, includes over- amplifier output via the terminals
gain has been made independent of the load load protection, DC protection, a ‘Amp’ and ‘LSP’.
by the addition of R30 and R31. These resis- switch-on delay for the output relay The supply voltage for the protection
tors together determine the output imped- and a voltage detector that directly circuitry is tapped off from the sup-
ance of the voltage amplifier, and as a result disables the output relay when the ply points on the amplifier board.
the source followers T12 and T13 now oper- power is switched off or any of the The supply voltage for the integrator
ate purely as buffers in the audio range. With- transformer voltages is absent. The is simply derived from the amplifier
out these resistors, the behaviour of the integrator for the offset compensa- supply voltages using a pair of Zener
amplifier is directly dependent on the con- tion is also located on this circuit diodes (D3 and D4).
nected load, which is not the way things are board. The overload protection circuit is
supposed to be. There are three terminals on the constructed in a ‘classic’ manner
Together with the compensation networks power amplifier board that provide using a voltage divider and a tran-
R1 R2 5W
470k 470k R3
input protection and is short-circuit proof.
C1 220Ω R5
10Ω
B1
K1
zie tekst
* see text 5W it was already noted that the power supply is
* voir texte R7 one of the most important components of a
10Ω
V-
V+
tp2
tp3
tp1
1-100010 ROTKELE )C(
D6 COMPONENTS LIST
C8
T7
bias
H1
H2
H2 H1
D5
T5
R17
C6
Amplifier board
R25
T8
R8
R9
R16 R24
D3
R11
R6
R4
R5
R2
R1
R3
T1
T2
R25
R24 C6
Resistors:
D1
C12
R32
R5
+
R18 C2 C1
C4
bias
R10 R1 = 1MΩ
C4
R9
R7
R2 = 47kΩ
R10
LSP+
C13
C3
R8
IC1
R6 R3,R22 = 470Ω
T12
R4
R7
T4
T3
R30
R4,R5 = 1MΩ8
R16
R15
R14
R13
R12
T1
R27
T2
D4
T
D3
tp3
tp1
P1 R8,R9,R13,R14 = 1kΩ
R17
R18
R28 T5
R26
R3
C1
R10,R15 = 330Ω
T6
T7
R37
C7
R28
T10
R16,R19,R30,R31 = 22kΩ
R31
R29
R20 R19
C11
C5
R2 R17,R20,R28 = 270Ω
R1 L1 IC2 R21
R23
R22
R18,R21 = 8kΩ2
T8
T11 R22
C8
T9
R36 R23
LSP-
R23 = 12kΩ
C10
C2
R24,R26 = 10kΩ
T T3
T4
R29
R30
R31
C9
RE1 R25,R27 = 33Ω
D1
R35
R12
0
D7 R32,R33 = 220Ω
0
C15
-
T13
R21 RE2
D2
T6
R26
~
R36 = 10Ω / 1W *
R13
R14
R27
D4
Amp.
LSP
R19 0
D2
T9
R37 = 1Ω / 5W
C7
T10
R20
C9
PC7
P1 = 1kΩ preset H
H3
H3
H4 H4
010001-1
PC9
PC8
Capacitors:
C1 = 2µF2, MKT (Siemens), lead
pitch 5mm or 7.5mm
C2,C4,C5 = 1nF
C3 = 180nF
C6,C7 = 100µF 25V radial
C8,C9 = 220µF 25V radial
C10,C12,C14 = 100nF
C11 = 10nF *
C13,C15 = 1000µF 63V radial
Inductors:
L1 = 9 turns 1.5 mm dia. ECW
around R37, inside diameter
8 mm
Semiconductors:
D1,D2 = rectangular face, red
D3,D4 = zener diode 3V9 / 0.5W
T1,T2,T6 = BC546B
T3,T4,T5 = BC556B
T7 = BC560C
T8 = MJE350
T9 = BC550C
T10 = MJE340
T11 = 2SK537 (Toshiba)
010001-1 (C) ELEKTOR T12 = 2SK1530 (Toshiba)
T13 = 2SJ201 (Toshiba)
Figure 5. The printed circuit boards for the amplifier and the protection circuitry are deliv- Miscellaneous:
ered as a single board and must be sawn apart. 5 off M3 spade terminals, PCB
mount
3 off ceramic (or mica) isolating
washer for voor T8/T10/T11
ings, we hasten to point out that we are talk- a toroidal transformer is used. This 2 off mica isolating washer for
ing about a monaural version here, so for a circuit does exactly what its name
stereo amplifier you will have to build two of suggests, and it ensures that exces-
these supplies! sive current surges do not occur
The ‘mains switch-on delay’ shown inside when the mains voltage is switched ics; the most recent one can be found
the dotted box in Figure 3 is not mandatory, on. Such circuits have frequently in the Summer Circuits issue of 1997,
but it is highly recommended — especially if been described in Elektor Electron- and we have reproduced its
*) may be omitted
Protection board
Resistors:
R1,R4 = 390Ω
R2,R5 = 120Ω
R3,R6 = 18kΩ
R7,R9 = 1kΩ / 5W
R8,R10 = 5kΩ6
R11,R25,R30,R31 = 10kΩ
R12,R13,R24 = 1MΩ
R14 = 150kΩ
R15 = 39kΩ
R16,R20,R21 = 220kΩ
R17 = 330kΩ
R18 = 12kΩ
R19 = 330Ω
R22 = 15kΩ
R23 = 680Ω
R26,R27 = 4kΩ7
R28 = 470kΩ
R29 = 3kΩ3
Figure 6. This is how the finished circuit board should appear. Don’t forget the insulating
Capacitors: washers for transistors T8 and T10–T13!
C1,C2 = 150pF
C3,C4 = 22µF 63V radial
C5 = 47µF 63V radial
C6 = 2µF2 MKT (Siemens), lead pitch limited by R4-R7 immediately after On the amplifier board, five wire bridges
5mm or 7.5mm switch-on. After the expiry of a time must be inserted, and it is a good idea to do
C7,C8 = 220µF 25V radial delay determined by C2 and C3, this at the beginning. In addition, there are
C9 = 1µF 63V radial these resistors are bridged over by two items on the amplifier board that could
the relay and the full current flows be considered to be somewhat difficult: the
Semiconductors:
between K1 and K2. The relay used thermal coupling and the output coil L1.
D1,D2 = 1N4148
D3,D4 = zender diode 20V / 0.5W here is a type that can switch For the thermal coupling between the D1/T5
D5,D6 = BAT85 2000 VA. The supply voltage for the and D2/T6 pairs, it is sufficient to mount the
D7,D8 = 1N4004 relay is taken directly from the mains LED so that it is in contact with the flat face of
T1 = MJE340 circuit via C1, R3 and B1, so this cir- the transistor. In the case of the T1/T2 and
T2 = MJE350 cuit is dangerous to the touch! T3/T4 transistor pairs, it is recommended to
T3,T8 = BC556B clamp a small metal ring around each pair.
T4,T5,T6,T10 = BC546B Soldering Incidentally, we have discovered that suitable
T7 = BC516 rings can be made by sawing them from a
T9 = BD140 The printed circuit board layouts for piece of copper water pipe and then bending
IC1 = OP77GP (Analog Devices)
the amplifier and protection circuitry them into a suitable shape.
IC2 = CNY17-3
are shown in Figure 5. These circuit Coil L1 can be easily wound on an 8-mm drill
Miscellaneous: boards are supplied as a single bit. After this you can insert R37 into the coil
Re1,Re2 = relay, type G2R-1-E piece, so the must be (carefully) and then solder both components to the cir-
(Omron), 16A / 24V / 1100 ohm) sawn apart. Experienced electronics cuit board, after having first removed the lac-
3 off M3 spade terminal, PCB mount types will not need very much quer from the two ends of the coil with the
advise with regard to the construc- aid of a knife.
tion of the circuit boards, since the Transistors T8 and T10–T13 are intentionally
component layout overlay and the placed along one edge of the circuit board so
schematic diagram in Figure 4. Its components list speak for them- that they can easily be screwed to a single
operation is simple, and is based on selves. Still, we would like to make a common heat sink. Naturally, the transistors
the fact that the current is initially few practical remarks. must be mounted using insulating washers,
C1
H4
the insulator before mounting the transistor.
1-870479
974078-1
The thermal resistance of the heat sink
F1
should be less than 0.5 K/W. Figure 6 shows
R1
one of the fully assembled prototype ampli-
R3
fier circuit boards with attached heat sink.
B1
K1
There isn’t much to say about the protection
R4
C3
~
circuit board. You should pay attention to the
diameter of the electrolytic capacitor C5,
~
which must be no more than 8 mm. If you
R5
cannot obtain a suitable type, a 40-V type can
974078-1
also be used.
RE1
R6
~
For the sake of completeness, the printed cir-
cuit board layout of the previously mentioned
K2
C2
OUT
R7
mains switch-on delay circuit is shown in
H2
H1
Wiring and set-up between these points must naturally sured the supply voltages, it’s nearly
be made using heavy-gauge wiring. time to power up the amplifier.
Once you have finished building the amplifier We used 2.5-mm2 electrical wire for Before doing this, however, you must
and protection logic boards (or sets of boards) this purpose. The contacts of relays turn trimpot P1 fully to the left
and have carefully checked them against the Re1 and Re2 on the protection board (counter clockwise). Otherwise you
components list, it is time to start looking for a are simply connected in series with run the risk that the quiescent cur-
suitable enclosure. The first decision to be the amplifier output by connecting rent will immediately rise to a very
made is whether you want to build the ampli- the output terminal ‘LSP+’ to the high level, which is not what we
fier as a monophonic building block or as a relay input terminal ‘Amp’ and the want.
stereo version. We chose the latter option for ‘LSP’ terminal of the protection After switching on the unit, first
our prototype, which means that what we board to the positive output socket check the amplifier output (test point
actually did was to build two mono blocks (banana socket). The other (nega- tp3) to verify that the voltage is zero.
into a single enclosure, each with its own tive) banana socket is connected An offset of a few millivolts is
power supply and mains switch-on delay. The directly to the ‘LSP–‘ terminal. acceptable, but if you measure 0.1 V
only shared item is the mains switch. For the The necessary connection between or more you will have to carefully
enclosure, we chose a Monacor (in some coun- the circuit ground of the amplifier reinspect the whole assembly, since
tries: Monarch) box that provides a generous and the metallic enclosure can best there is something wrong.
amount of room for everything, and then be realised by fitting the Cinch Following this, you can set the quies-
mounted hefty heat sinks (bigger than actu- (a.k.a. RCA or ‘line’) input sockets in cent current to the proper value. The
ally required) on opposite sides of the box. a ‘normal’ (non-insulated) manner. ideal value for this amplifier is 200 to
Since there are several circuit boards Take care that there is not any other 250 mA. To adjust the quiescent cur-
involved, the wiring of the complete amplifier unintentional connection between rent, connect a voltmeter across R34
includes quite a few interconnections – which the signal ground and the enclosure (test points tp1 and tp3) and turn P1
is why we have made a separate wiring dia- ground, since this will create an slowly until the measured voltage is
gram, as shown in Figure 8. Connect the V+, earth loop that can cause stubborn between 0.044 and 0.055 V. Then let
V–, earth, tp1, tp2, tp3 and bias points on the hum problems. the amplifier warm up for half an
protection board to the corresponding points It goes without saying that a well- hour, and again adjust the current to
on the amplifier board using ordinary insu- insulated cable, a robust mains the same value using P1.
lated stranded wire. The ‘∼35 V’ points switch and an equally robust mains
should be connected directly to the outer entrance must be used for the con- Listening
ends of the transformer windings, and point nection to the 230-V mains circuit.
‘0’ should be connected to the junction of the Pay attention to the electrical safety Readers who have already taken a
filter capacitors in the power supply. of the overall assembly, and attach peek at the measurement results
Use lengths of screened audio cable to make an identification label that lists the shown in the separate box will have
the connections between the input sockets specified values of the supply volt- quickly concluded that the Crescendo
(Cinch sockets) and the input points on the age (230 V) and fuse to the outside scores very well as far as the num-
amplifier boards. of the enclosure. bers are concerned. However, we
Flat tab connectors (automotive connectors) know from experience that amplifiers
are used for the output and supply connec- Once you have again thoroughly with practically identical specifica-
tions on the circuit boards. The connections checked everything and re-mea- tions can sound quite different.
PC8
PC9
Amp.
T10
C9
H4 H4
H3
H3
R20
PC7
LSP
types of music. Of course, the
C7
T9
D2
R19 0
R27
R14
R13
D4
~
relative differences between
R33
R26
D8
C14
T6
D2
R21 RE2
tp2
T13
C15
good amplifiers are always
R15
~
0
D7
C5
very subtle, but the Crescendo
R11
0
R35
R12
D1
RE1
C9
R31
R30
R29
clearly revealed itself to be an
T4
T3
T
C10
LSP-
C2
T9
amplifier with a pleasantly
R36 R23
C8
T11
R22
R22
T8
R23 R1
L1 R21 IC2
warm-blooded character.
C11
R2
R29
C5
R19 R20
R31
T10
R28
R37
C7
After listening to the amplifier
T7
T6
C1
R3
R28 R26
R18
R17
T5
tp3
P1 D3
tp1
for a while, we developed a
T
D4
C3
R34
T2
R12
R13
R14
R15
R16
R27
T1
certain understanding of the R30
T3
T4
R7
T12
R4
LSP+
R6
IC1
C13
R10
R8
C3
preferences of fervent MOS-
C4
R7
R9
bias
R10
C4
FET fans, since the sound pro-
R18 C1 C2
R32
C12
+
R5
D1
R24 C6
R25
duced by the amplifier is just a
R11
T2
T1
R3
R1
R2
R5
R4
R6
R25
D3
R16 R24
R9
R8
T8
bias
C6
R17
T5
D5
T7
C8
H2
D6
tp1
tp3
tp2
V+
V-
than that produced by a typi-
cal amplifier with bipolar tran-
sistors in the output stage. An
amplifier such as the ‘Compact
AF Power Amplifier’, which
was published in May 1997
C C
(and which is one of our 22000µ 22000µ
favourites), offers reproduction 63V 63V
that (according to our convic-
tions) can hardly be surpassed
in terms of natural fidelity and
detailing, but it still misses
that slight trace of warmth
that is so typical of the B = 200V / 35A
Crescendo. Can we say that B
one of the two is the better F1 = 1A T
amplifier? No, that would be R2
H3
R1
nouncement, and anyhow
R3
B1
R6
OUT
R7
H2
H1