This is only a preview of the February 1992 issue of Silicon Chip. You can view 45 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Build A Stereo Power Amplifier":
Articles in this series:
Articles in this series:
Articles in this series:
Articles in this series:
|
New stereo module delivers up to 160 watts
into 4-ohm loads
By LEO SIMPSON & BOB FLYNN
The last time we published a power
module using bipolar transistors was
back in December 198 7. That single
channel design used 2N2955/2N3055
TO-3 metal-pack transistors to deliver
either 50 watts into an or 100 watts
into 4Q, depending on whether two or
four output transistors were used.
Later, in February 1988, we upgraded
the module by substituting the more
rugged (and more expensive) MJ15003/4 TO-3 power transistors.
Both designs have been very popular and are still available but recently
we have seen the need for a more
compact, multi-purpose amplifier
module which would drive 4Q or 8Q
loads without having to change the
design. In fact, the real stimulus for
the design was that we wanted to
produce a new integrated stereo amplifier which would fit into a midisized chassis; ie, about 340mm wide.
We had a target of 50 watts per
34
SILICON CHIP
Say goodbye to tin lid
transistors and hello
to f antas tic plastic.
This new stereo power
module uses four big
plastic Darlington
transistors in each
channel, making a
rugged and compact
design incorporating
full protection.
channel for the new design and initially we intended to base it on one of
the Japanese-made stereo modules.
These are used in very large numbers
in today's lower cost stereo amplifiers
and particularly in the all-in-one rack
systems.
In fact, we went ahead with a design based on such a module , capable
of delivering 50 watts per channel.
But after building several prototypes
we had to give the game away. There
were just too many compromises in
the design. Those modules looked
great ,on paper but, in practice, they
have drawbacks which cannot be
cured, since the core circuit is contained in moulded black plastic.
So it was back to the drawing board.
OK then, having spent a great deal of
development time which had so far
come to nought, what was to be the
next approach? We did not want to
use TO-3 power transistors (hence our
into 4Q loads, has low distortion and very low residual noise.
The total power output
will depend to a great extent on the regulation of the
power supply. We will have
more to say about this later.
Protection
reference to tin lid transistors at the
start of this article). Sure, TO-3 power
transistors give plenty of power for
their size (particularly the MJ15003/
4s) but after 30 years or more, they are
a bit old hat and are becoming more
expensive as time goes by.
By contrast, plastic encapsulated
power transistors are becoming more
rugged and cheaper. And they are
much easier to design into PC boards
and have single screw mounting.
So the transistors we have selected
for the new design are the TIP142
(NPN) and TIP147 (PNP) plastic
Darlingtons. Made by Philips, Motorola and a number of other companies, they are housed in the so-called
plastic TO-3 encapsulation (ie, TO3P) although Philips list it as the SOT93 pack and Motorola as the TO-218.
Either way, these new Darlingtons
(first listed by Philips in July 1988)
are quite rugged, with the following
ratings: power dissipation 125 watts
(TMB = 25°C); collector-emitter voltage 100 volts (V CEO - open base); collector current 10 amps DC, 15 amps
peak; DC current gain (hFEl >1000<at> 5
amps; and a maximum junction temperature of 150°C. In fact, they compare very favourably with the old favourite 2N3055s which have a power
dissipation of 115 watts, 15A maximum collector current and VCER of 70
volts. They also compare quite well
in the critical area of "second breakdown".
Performance
The performance of this new stereo
power module is very respectable and
certainly better than the vast majority
of the low to midrange amplifiers. In
particular, it delivers lots of power
This aspect is most important in any medium to
high power amplifier design and we have followed
the same approach as we
have used in our designs of
December 1987 and February 1988 - fuses in the positive and negative supply rails and a
Polyswitch PTC (positive temperature
coefficient) thermistor in series with
the output.
While some readers may regard PTC
thermistors as a needless option they
will bless them if ever they are called
into operation. They are much cheaper
than having to replace the drivers in
your loudspeakers and they are excellent insurance against loudspeaker
fires which can happen in some cases
of amplifier failure.
The trouble with today's power
amplifiers is that they use big power
supplies which can deliver a great
deal of current. If just one of the transistors in the amplifier fails, the result
can be that the circuit applies the full
positive or negative supply rail to the
loudspeaker. Typically, an 8Q loudspeaker will have a voice coil resistance of about 5.5Q. When that cops
the full positive supply of this amplifier, it will have to dissipate around
150 watts or more. Now maybe the
fuses will blow and save it but the
most likely result is that the voice coil
will be burnt out.
That's not all. In some cases, the
red hot voice coil sets the loudspeaker
on fire which then generates huge
quantities of acrid smoke from the
acetate filling material in the cabinet.
There have been documented cases of
this happening - and big insurance
payouts for smoke and fire damage to
homes. So the PTC thermistor in this
amplifier is highly desirable.
Note that you cannot rely on the
fuses to give protection to the loudspeaker. They are selected to protect
the amplifier and its supply more than
for output protection. And in the case
of the fault condition outlined in the
preceding paragraphs, they may not
What Is A Darlington Transistor?
The TIP142/147 plastic power
transistors specified in this new
amplifier design are referred to as
Darlingtons, after S. Darlington, who
first proposed the Darlington transistor pair in the early 1960s. Effectively, they are a compound transistor pair, with the emitter of the first
transistor connected to the base of
the second. Hence, the current gain
of the pair is approximately equal to
the product of the two transistor
gains.
The internal circuit configuration
of the TIP142 is shown in Fig.1. It
includes a reverse diode from collector to emitter of the second transistor. This diode is very handy for
reverse voltage protection in amplifier and switching circuits.
These days, these integrated transistors are shown on circuit diagrams
by the conventional transistor symbol, with nothing to distinguish them
from ordinary discrete transistors.
NPN
TIP140
TIP141
TIP142
COLLECTOR
r--_
- -_
-- - 7I
_.......,
1
I
I
BASE
I
I
I
I
I
L ___ - - - -
__ _j
EMITTER
Fig.1: the equivalent circuit of the
TIP142 NPN Darlington which has
inbuilt base-emitter shunt resistors
and a reverse protection diode.
The collector is connected to the
metal tab of the plastic package.
However, while Darlington transistors have the advantage of high
current gain and space saving on
printed circuit boards, they are at a
disadvantage in switching circuits
where their saturation voltage
(V CEsat) is usually not as low as can
be achieved with discrete transistor
switching circuits.
FEBRUARY1992
35
+38.5V
01
f1
0.22!
5A
02
0.22!
010
03
l
BC556
C
.,.
.033
_j
L1
6.8uH
D.4m
3W
0.1
1
B
o--1
INPUT
OUTPUT
22k
i
2.2pF
0.15!
,.
o.4m
3W
1k
1.,.
2xTIP147
+
47
16VW+
8
68pf
011
4.7k
1
2.2V 4.7k
C
05
04
8
2xBF469
37.2V
E
F2
SA
0.22J
2.2V
1000
i
1.6V
-38.SV
0.221
A
PLASTIC
SIDE
YELLOW
8
+
240VAC
+38.SV
4700
N
+
sovw
YELLOW
ELJc
VIEWED FROM
BELOW
~
ECB
BF-,BD139
~
BCE
TIP-
GND
.
4700
+
sovw
-38.SV
TWIN SOW POWER AMPLIFIER
Fig.2: the voltages shown on the circuit diagram are nominal values & are what
can be expected if you have a 240V mains supply. If your mains supply is
higher, you can expect most of the voltages to be higher in proportion. Note that
the DC voltage on the output should be within ±30mV.
blow soon enough, if at all.
Note that this loudspeaker (and fire)
hazard is common to all modern amplifier designs , commercial and do-ityourself, not just the design under
discussion here.
PTC protection thermistors were
first used in loudspeakers from the
UK about seven or eight years ago. To
our knowledge, we were the first to
incorporate them in the output of an
amplifier design although they are also
used in the high voltage rails of amplifiers such as the NAD which have
very high music power output.
Normally, PTC thermistors have a
36
SILICON CHIP
very low resistance, a mere 0. H1 or
less in the case of the ones specified
for this amplifier. They stay that way
while ever the current through them
is below their cutoff rating (around 5
amps in this case). If the current rises
above this value, the PTC thermistor's
resistance suddenly rises to a high
value, around several hundred ohms,
effectively disconnecting the amplifier from the load and thus protecting
it and the loudspeaker.
Thus, the PTC thermistor is very
effective in protecting the loudspeaker
against overdrive from the amplifier
or worse, a catastrophic failure in the
amplifier. When the fault condition is
removed, the PTC thermistor's resistance gradually returns to normal although this may take several minutes
or more to fully recover.
The PTC thermistor will also protect the amplifier against short circuit
loads1although in this respect the supply fuses give backup protection.
The circuit
Now let's have a look at the circuit
of Fig.2. This is very similar to the
design featured in December 1987 except that we have substituted the
TIP142/147 Darlingtons for the
2N2955/2N3055 output transistors
and their MJE340/350 driver transistors.
The input signal is coupled via a
lµF metallised polyester capacitor and
1.8kQ resistor to the base of Ql which
together with Q2 makes up a differential pair. Q3 is a "constant current
tail" which sets the current though
Ql and Q2 and thus renders the amplifier largely insensitive to variations
in its supply rails (known as power
supply rejection).
Diodes D1 and D2 provide a voltage
reference of about 1.2V for Q3 so that
it applies a constant voltage to its
680Q emitter resistor. This sets the
current through Q3 to close to lmA.
This means that Ql and Q2 each operate with a collector current of about
0.5 milliamps.
Signals from the collectors of Ql
and Q2 drive another differential pair,
Q4 and Q5, which have a "current
mirror" as their load. The main advantage of the current mirror, D3 and
Q6, is that it makes the second differential pair highly linear and therefore
low in distortion.
The output of Q5 drives the classAB output stage consisting of Darlingtons Q8-Q11. By class-AB we mean
an amplifier which is essentially classB (ie, each half of the output stage
conducts for only half the signal) but
which has a small current bias to minimise cross-over distortion.
PARTS LIST
1 PC board, code SC01102921,
80 x 233mm
1 60 x 60 x 290mm 3mm-thick
angle aluminium
19 PC stakes
8 SOT-93 (T0-218) transistor
mounting kits
2 SOT-32 (T0-126) transistor
mounting kits
8 20mm fuse clips
4 5A M205 20mm fuses
2 Philips 4322-021-30330 or
Neosid 60-601-72 coil formers
2 RDE245A polyswitches
2.5 metres 0.8mm enamelled
copper wire
1 power transformer 2 x 25V,
160VA (from Harbuch, Altronics
or Torrtech)
2 500Q horizontal mount trimpots
2 560Q 5W resistoris (for setting
quiescent current)
Semiconductors
4 TIP142 NPN Darlington
transistors (08 ,010)
4 TIP147 PNP Darlington
transistors (09,011)
4 BF469 NPN transistors (04,05)
2 BF470 PNP transistors (06)
2 BO139 NPN transistors (07)
6 BC556 PNP transistors
(01,02,03)
6 1N4148 signal diodes (01 ,02,03)
1 P04 6A bridge rectifier
Capacitors
2 4700µF 50VW electrolytics
2 47µF 16VW electrolytics
2 1µ,F 63VW 5mm pitch metallised
polyester
8 0.22µF 63VW 5mm pitch
metallised polyester
2 0.15µF 10% 100VDC 10mm pitch
metallised polycarbonate (Philips
2222 344 21154)
2 0.1µF 63VW 5mm pitch
metallised polyester
2 .033µF 100VW 5mm pitch
metallised polyester
2 820pF 50V ceramic
2 68pF 100V ceramic
2 2.2pF 50V ceramic
Resistors (0.25W, 5%)
6 22kQ
4 680Q
215kQ 0.5W
4180Q
44.7kQ
101000
24.7k01W
2 6.8Q 1W
2 1.8k0
8 0.47Q 3W 10%
21kQ
Vbe multiplier
The current bias in the output stage
is controlled by transistor Q7 and trimpot VR1 . Q7 is a Vbe multiplier, so
called because the voltage between
its base and emitter is multiplied by
the ratio of the resistors between base
and collector and base and emitter,
respectively. VR1 adjusts this voltage
to give a voltage between the collector
and emitter of Q7 of about 2 volts. In
practice, it is adjusted to give an output stage current of 40 milliamps.
Q7 is a BD139, a transistor normally used in amplifier driver stages
and video circuits. It has a dissipation
rating of 8 watts and so is only doing
light duty. However, it is specified
here because it gives better Vbe tracking with the output stage transistors
and therefore better stability for the
quiescent current.
The Darlington transistors Q8 and
Q10 and Q9 and Ql 1 are connected as
parallel pairs to share the output current. Each Darlington has a 0.47Q emitter resistor which helps ensure equal
current sharing. As well, the emitter
resistors improve the output stage
bandwidth and the stability of the
quiescent current. The value is a compromise though; bigger emitter resistors would give better stability and
current sharing but would reduce the
maximum output power capability.
The 100Q base resistors for the
Darlingtons serve a number of functions. First, they reduce any tendency
for the output stage to oscillate; always a possibility with emitter follower stages. Second, they limit the
base current in the event of a short
circuit in the output and thereby reduce the possibility of damage to the
Darlingtons.
Negative feedback is applied from
the output stage back to the base ofQ2
via a 22kQ resistor. This resistor, and
the lkQ resistor also connected to the
base of Q2, sets the voltage gain to 23.
The low frequency rolloff of the voltage gain is set by the 47µF capacitor
in series with the lkQ resistor. This
sets the -3dB point at about 3Hz. However, the lµF input capacitor is the
main factor in the low frequency response of the amplifier and sets a
-3dB point at 7Hz. The overall effect
of the two time constants is a -3dB
point at 10Hz.
The 820pF capacitor and the 1.8kQ
input resistor feeding Ql form a low
pass filter which rolls off frequencies
above l00kHz. This filter is a little
more savage than we have used in
previous designs but we have done
this to give a greater margin of safety
in the output in case the preamplifier
stages have any tendency to high frequency instability.
The 68pF capacitor between base
and collector of Q5 and the 2.2pF
capacitor between base and collector
ofQ2 are used to roll off the open loop
gain to ensure stability with feedback
applied. We have also used our standard RLC network in the output stage.
A configuration originally proposed
by Australian engineer Neville Thiele,
it uses a 6.8µH air-cored choke, a 6.8Q
resistor and a 0.15µF capacitor.
FEBRUARY1992
37
The beauty of this network is that it effectively isolates
the amplifier output stage from any nasty impedance
dips which may occur at high frequencies and which
could cause the amplifier to be unstable. It also has
another favourable effect because it kills any RF signal
pickup by long speaker leads.
Power supply
The power supply for the amplifier is shown in Fig.2.
This uses a 160VA transformer with a centre-tapped 50V
winding feeding a 6-amp bridge rectifier and two 4700µF
50VW electrolytic capacitors.
PC board design
The PC board for this stereo amplifier has been designed so that it can be built as two separate modules.
The stereo pair can be built with an onboard power
supply which will also feed a preamp stage. Alternatively, if you want to use a bigger bank of filter capacitors,
a higher rated bridge rectifier and the capacitors would
be mounted off the board.
Do not substitute a transformer with a higher secondary voltage. If you do so, you run the risk of blowing the
Darlington transistors, particularly if you are driving 4Q
loudspeakers.
The specified heatsink is a 3mm-thick aluminium
angle extrusion, 60 x 60 x 290mm long. This heatsink is
adequate where the amplifier is intended for normal
program material. If you envisage using it with a bigger
capacitor bank and more onerous signal conditions such
as a guitar amplifier, then a bigger heatsink or thermal
cutouts would be desirable.
For the remainder of this article though, we will assume that the reader is building a stereo module on the
specified heatsink.
Note that the parts list specifies all the components for
a stereo amplifier and makes reference to transistors such
as Ql, Q2 etc. Transistors Q1-Q11 are shown on the
circuit diagram (Fig. 2) and these are duplicated in the
second channel. The same goes for the diodes.
Assembling the board
Fig.3: the parts layout for a complete stereo amplifier
power module with on-board power supply. The 6.8µH
output inductors (Ll) are each wound on a Philips 4322021-30330 or Neosid 60-601-72 coil former using 24.5
turns of 0.8mm enamelled copper wire. Fig.2 shows the
pinout details for the transistors.
38
SILICON CHIP
We suggest that you mount the PC pins, resistors ,
diodes and wire links first , followed by the capacitors.
There are only two electrolytics on the board, apart from
those in the power supply. Make sure they are mounted
with correct polarity. Most of the remaining capacitors
are moulded metallised polyester capacitors which have
a standard lead spacing (pitch) of 5mm. We recommend
against greencaps as they won't fit.
The fuses specified are M205 20mm-long types as
widely used in commercial amplifiers. The main reason
we have specified them is that they take up less board
space than the larger 32mm 3AG fuses and cost no more.
All the TO-5 transistors (Q1-Q3 , etc) are mounted with
the flat side facing towards the front; ie, away from the
heatsink. Similarly, the TO-126 transistors (Q4-Q6, etc)
mount with the metal side facing to the front. The exception is Q7 which naturally mounts with its metal face to
the heatsink (with a mica washer, of course).
The 0.47Q 3-watt resistors are made by Philips and
again have been specified to save board space, being a lot
more compact than the common 5W cement "bathtub"
types. Mount them so that they clear
the board by about 3-4mm.
The 6.8µH output inductors are
each wound on a Philips 4322-02130330 or Neosid 60-601-72 coil former
using 24.5 turns of 0.8mm enamelled
copper wire. Clean and tin the ends of
the inductors before installing them
on the board.
Output transistors
There are several ways of mounting
the output transistors but the way we
did it is as follows. First, all the Darlingtons and the two BD139s were
mounted on the aluminium heatsink.
In each case, they were mounted using the specified mounting kit consisting of a mica ~asher and plastic
insulating bush for the screw. Heatsink compound is applied sparingly
to both sides of the mica washer before it is set between the transistor
and heatsink. The details are shown
in the diagram of Fig.4.
With all the transistors mounted on
the heatsink, set your multimeter to a
low ohms range and check that the
transistor collectors are isolated (ie,
infinite resistance) from the heatsink.
That done, set the heatsink upside
down on your workbench and fit the
PC board over the transistor leads.
You may need to adjust some of the
transistors so that their leads line up
with the board holes. Tack soider a
couple of Darlington transistor leads
at each end so that the top board surface is about 8mm from the bottom
edge of the heatsink.
You will also need to slightly crank
the leads of the BD139s (Q7) to line
them up with their respective PC
board holes. When you are satisfied
with the lining up of the board, soldE;Jr
Performance of Prototype
Output power .. .. .......... .. ..................... 55W into 8 ohms, 80 watts into 4
ohms (one channel driven)
Frequency response ......................... 15Hz - 35kHz ±1 dB
Input sensitivity .................................. 900mV (for clip point into 8 ohms)
Harmonic distortion ........................... typically less .05% from 20Hz to
20kHz)
Signal to noise ratio ........................... 105d8 unweighted; 11 ?dB Aweighted
Separation between channels ........... 84d8 or greater (1 00Hz - 10kHz)
Protection .......................................... 5A fuses plus RDE245A Polyswitch
Damping factor .................................. <50 (for 8 ohm loads)
Stability ............................ ................. unconditional
all the transistor leads to the board
pattern.
Power up
Before applying power, check all
your work very carefully against the
wiring diagram of Fig.3. This done,
remove the four SA fuses and solder a
CAPACITOR CODES
Value
IEC Code
EIA Code
1µF
105
0.15µF
0.1µF
1u0
220n
150n
100n
820pF
68pF
2.2pF
820p
68p
2p2
0.22µF
224
154
104
821
68
2.2
560Q 5 watt resistor across each of the
on-board fuseholders. These are current limiting resistors which reduce
the likelihood of any damage to the
output transistors in case you have
done something silly like swapped a
TIP142 for TIP147. But of course you
have already checked to see that nothing like that has happened, haven't
you?
Now connect the positive and negative supply leads to one channel of
the amplifier. Set trimpot VRl fully
anticlockwise - this gives the minimum setting for quiescent current
through the output transistors. Set
your multimeter to the 200VDC range
(or no lower than S0VDC if an analog
meter). Do not connect a loudspeaker
or output load at this stage.
Now apply power and measure the
positive and negative supply rails.
RESISTOR COLOUR CODES
D
D
D
D
D
D
D
D
D
D
D
No.
Value
4-Band Code (5%)
5-Band Code (1%)
6
2
4
2
2
2
4
4
10
2
22kQ
15kQ
4.?kQ
4.?kQ
1.8kQ
1kQ
680Q
180Q
100Q
6.8Q
red red orange gold
brown green orange gold
yellow violet red gold
yellow violet red gold
brown grey red gold
brown black red gold
blue grey brown gold
brown grey brown gold
brown black brown gold
blue grey gold gold
red red black red brown
brown green black red brown
yellow viole! black brown brown
yellow violet black brown brown
brown grey black brown brown
brown black black brown brown
blue grey black black brown
brown grey black black brown
brown black black black brown
blue grey black silver brown
FEBRUARY1992
39
The stereo power amplifier module will form part of a complete stereo
amplifier to be described in a future issue of SILICON CHIP. This photo shows
the unit in company with its companion tone control board at bottom left & the
input preamplifier board at right.
temporarily short out the PTC thermistors.
Troubleshooting
They should be within a few volts of
±38.5 volts. Now measure the other
voltages on the circuit. They should
all be within ±10% of the nominal
values, depending also on whether
your 240VAC mains supply is high or
low (it is above 240VAC more often
then not).
The voltage at the output should be
within ±30mV of 0V.
Now switch your multimeter back
to the 200VDC range and connect it
acl'oss one of the 5600 5W resistors.
Adjust trimpot VR1 for a reading of
22.4 volts. This gives a total quiescent
current of 40 milliamps.
After 5 minutes or so, check the
quiescent current again and readjust
VR1 if necessary to get the correct
voltage across the 5600 resistor. (If
you are doing power tests on the amplifier and the heatsink becomes very
hot, you can expect the quiescent current to at least double. When it cools
down though, the quiescent current
should drop back to around 40mA).
Measure the voltage across each
0.470 3W emitter resistor. They
should all be about 9- lOmV, which
means that each Darlington transistor
is getting its rightful share of the
quiescent current.
Now switch off and connect the
positive and negative supply rails to
the other channel. Measure the
voltages as before and adjust VRi in
40
SILICON CHIP
that channel for the correct quiescent
current. If all is well, switch off, remove the 5600 5W resistors and fit
the 5A fuses. The amplifier module is
now ready for work.
Nate that if you intend running continuous power tests on the module,
the PTC thermistor will operate before you can get full power into a 40
load. They will let full power be delivered on music signals but not for
continuous sinewave signals. To do
such full power tests, you will have to
INSULATING
MICA
WASHER
-~~JI
'SCREW
r
mnmfs
--.._ HEATSINK
1
T0220
DEVICE
Fig.4: transistors Q7-Qll are
each isolated from the L-shaped
heatsink using a mica washer &
insulating bush. Smear the
transistor tabs & mica washers
with heatsink compound before
bolting the assemblies together &
use your DMM to check each
transistor as it is mounted to
ensure correct isolation.
What happens if one of the amplifiers is not working? If the other channel is working correctly then you have
an ideal cross-check. Check the
voltages in the good channel and then
in the bad channel and you can usually get a fair idea of what the problem is. It is unlikely that you will get
the same fault in both channels, unless you have made the same assembly mistake in both!
And now we'll give you a few clues
which may help you solve any problems. First of all, let's assume that
most of the amplifier voltages are correct but you have zero quiescent current. Look for a short across VR1 or
Q7. If you have lots of current through
the 5600 resistors and cannot control
it with VR1, look for an open circuit
in the 6800 base resistor to Q7 or a
defect in that transistor.
What if the output of the amplifier
is fully latched up at either +38V or
-38V? The most likely cause of this is
a defect in the first or second differential pair of transistors, or something
silly like the wrong transistor, say a
BF469 where a BF470 should be. Solder bridges between tracks can also
cause this fault.
The above are the more common
problems with build-it-yourself amplifiers. Most times though, you can
expect the modules to work perfectly
at switch on.
SC
|