Studio 350 Pow

Amplifier Modu
Want an audio power amplifier with some real grunt?
Want an audio power amplifier which is really quiet and
has very low distortion? Here is the one answer for both
desires. The Studio 350 is a rugged power amplifier
module capable of delivering 200 RMS watts into an
8-ohm load and 350 watts RMS into a 4-ohm load, at
very low distortion.
Pt.1: By LEO SIMPSON & PETER SMITH

O
UR first approach on design- voltage amplifier stage. The net result of the output stage (either the NPN or
ing this amp was to decide on is a rugged power amplifier with very the PNP transistors). For the 4Ω load,
the target power output, given low residual noise and distortion. the load line runs up to 17·5A on the
a likely supply rail. We decided to vertical axis, corresponding to the cur-
aim for 200 watts into an 8-ohm load. Load lines and power ratings rent delivered if the active device was
A few back-of-an-envelope calcula- So why did we end up using eight fully turned on (ie, 70V/4W = 17·5A).
tions showed that we would need 200W transistors in order to get just Similarly, for an 8Ω load, the load
supply rails of about ±70V or a total 200W into 8Ω and 350W into 4Ω? line runs up to 8·75A on the vertical
of 140V. It might seem like over-kill but it is axis (ie, 70V/8W = 8·75A). These load
Naturally, we would also want to not. To work out the dissipation in a lines show the instantaneous power
drive 4-ohms loads and with those transistor, you need to draw the load dissipation at any possible signal
same supply rails we would expect lines. These show power dissipation condition (including an output short
to obtain around 350 watts. But how in the active device (in this case, one circuit).
many output transistors and what type half of the output stage, consisting of Also shown on the diagram are two
would be required? As you can see four transistors). The vertical axis is hyperbolas. One represents the maxi-
from the photos and circuit, we have in Amps while the horizontal axis is mum safe power (for one second!)
used eight 250V 200W plastic power Volts. The various load lines for our dissipation of four parallel-connected
transistors: four MJL21193/4 comple- amplifier are shown in Fig.1. MJL21193/94 transistors. Depending
mentary pairs. These are teamed with For a start, we plotted the lines for on the instantaneous voltage across
the high-performance MJL15030/31 8-ohm and 4-ohm resistive loads and the transistors, this can be more than
complementary driver transistors. these are straight lines, showing all 900W for low voltages, reducing to
In addition, we have used some possible conditions. The two resis- 720W at 80V, and ultimately to just
high-voltage low-noise transistors tive lines start at the 70V mark on the 400W at 250V (not shown on the
in the input stage and highly linear horizontal axis, corresponding to the curve). This hyperbola represents
high-voltage video transistors in the supply voltage applied across one half the maximum dissipation the four

10 Everyday Practical Electronics, October 2006

er
le

transistors can withstand because the load lines show

under a non-repetitive one- instantaneous power dissipa-
second pulse, the so-called Safe tion, not average or total power
Operating Area (SOA). dissipation. As long as the load
Since the resistive load lines are well To show this lines are below the SOA curve,
below the one-second SOA hyperbola, effect, we have drawn everything is OK.
you may think that the transistors are 8Ω and 4Ω reactive load All of the foregoing is a shortened
operating far below their maximum lines which represent speakers explanation of the process whereby
ratings and so they would be, if all they with complex impedances of 5·6Ω + we decided to use eight transistors. It
had to drive was resistive loads. Sadly, j5·6Ω and 2·83Ω + j2·83Ω, respectively. shows that eight is a good conservative
loudspeakers are not resistive; they can In the 8Ω case, the 5·6Ω represents the figure whereas six of these transistors
be resistive, inductive or capacitive, voice coil resistance while the j5·6Ω would not be enough.
depending on the signal frequency. Usu- is the coil inductance. The resulting Finally, before we leave the discus-
ally they are inductive which means the curved load lines extend well beyond sion on load lines, we need to mention
load current lags the load voltage. 70V (to almost 110V) and also show short circuit and overload protection.
This has two effects. First, the volt- instantaneous dissipation figures far Apart from fuses, this amplifier circuit
age across the output transistors can go in excess of that for the resistive load has no protection. We could have cho-
much higher than the half-supply value lines. In fact, you can see that in the sen to run with six power transistors
of 70V. Conceivably, it can run to the case of the 4Ω reactive case, there is if we had incorporated “load line”
full supply voltage of 140V (or beyond, far less power margin to spare. protection into the circuit. This uses
if driven into clipping on an inductive We have also drawn the de-rated a pair of transistors to monitor the
load). Second, the instantaneous power power hyperbola (50°C) for four tran- output transistor voltage and current
dissipation across the power transistors sistors on Fig.1 and as you can see, it conditions and then limit the base
can go far in excess of that shown for a touches the 4Ω reactive curve. Does drive signal when the load line is
resistive load line. this mean there is a problem? Well no, exceeded.

Everyday Practical Electronics, October 2006 11

 Fig.1: this diagram
shows the resistive
and reactive load
lines for both 4Ω and
8Ω loads. Also shown
are two hyperbolas.
The blue curve shows
the maximum safe
operating area of four
parallel-connected
MJL21193/MJL21194
transistors, while
the red curve shows
the derated power
curve for 50°C case
temperature.

Such circuits can work quite well to If the fuses were increased in rating, The alternative approach is to mount
protect the output stage but in practice the amplifier could ostensibly drive a the output transistors vertically on a
their rapid switching action causes a 2Ω resistive load without damage, so single-sided or fan heatsink, in which
burst of high frequency oscillation to we think the 5A fuses should provide case the PC board could be trimmed
be superimposed on the output signal. adequate short circuit protection. Oh, to 240mm wide by 100mm deep. This
This means that not only do you get but we don’t recommend driving a latter approach takes up less chassis
horrible distortion but the amplitude 2Ω load! space. Both approaches will be de-
of the burst can be enough to overload scribed in the constructional details
and burn out tweeters if the overdrive Amplifier module to be presented next month.
situation persists. Two versions of this amplifier mod-
Therefore, while we regard load ule are possible, both using the same Performance
line protection as important for PA PC board pattern. The one presented As already noted, the Studio 350 de-
amplifiers (which can easily have their here employs a cast aluminium heat- livers up to 200W RMS into an 8-ohm
output leads shorted), it is not desir- sink with an integral shelf which is load and up to 350W into a 4-ohm load.
able for a hifi amplifier. If you do short convenient for mounting the power Music power figures are substantially
the outputs of this amplifier when it transistors. This heatsink is 300mm higher, around 240W into an 8-ohm load
is under full drive, there will be a big wide and the PC board itself is 240 and 480W into a 4-ohm load. These fig-
spark and hopefully the only thing × 136mm so the overall assembly is ures apply only for the suggested power
to be damaged will be the 5A fuses. quite large. supply, which we will come to later.

Fig.2: total harmonic distortion versus power at 1kHz into Fig.3: total harmonic distortion versus power at 1kHz into
an 8-ohm load (10Hz-22kHz measurement bandwidth). a 4-ohm load (10Hz-22kHz measurement bandwidth).

12 Everyday Practical Electronics, October 2006

Fig.4: harmonic distortion versus frequency at 160W into Fig.5: distortion versus frequency at 250W into a 4-ohm
an 8-ohm load (22Hz-80kHz measurement bandwidth). load (22Hz-80kHz measurement bandwidth).

Fig.2 shows the total harmonic are a differential pair using Hitachi resistance of 0·1Ω and a DC output
distortion versus power at 1kHz into 2SA1084 low-noise transistors which offset from the amplifier of just 20mV,
an 8-ohm load while Fig.3 shows have a collector-emitter voltage rating the resulting current through the
distortion versus power at 1kHz into of 90V, necessary because we are using transformer will be 200mA! Not only
a 4-ohm load. As you can see, for 70V rails. Transistor Q1 and diodes will this magnetise the core and de-
an 8-ohm load, distortion is around D1 & D2 make up a constant current grade the transformer’s performance,
0·002% or less up to about 180W, ris- source running at about 1mA to set the it will also result in additional power
ing to around 0·03% or thereabouts current through the differential pair at dissipation of 14W in one half of the
at 200W. At low powers, below 0·5W, 0·5mA each. amplifier’s output stage. This is not
the distortion figure rises but that is Trimpot VR1 in the emitter circuit good! Hence, trimpot VR1 has been
due to residual noise, not distortion. to the differential pair is provided to included.
In reality, at low powers the distortion adjust the offset voltage and thereby Signals from Q2 & Q3 drive another
is well below 0·001%. trim the output DC voltage very close differential pair, Q4 & Q5, which have
Similarly, for a 4-ohm load, distor- to 0V (within a millivolt or so). This a “current mirror” as their collector
tion is around 0·0045% or less for pow- is largely academic if you are driving loads. The current mirror comprises
ers up to around 280W, rising to 0·1% normal 4-ohm or 8-ohm loudspeak- diode D3 and Q6, essentially a varia-
at around 350W. These figures were ers but is particularly desirable if you tion of a constant current load, which
taken with a measurement bandwidth intend driving electrostatic speakers ensures high linearity in Q5. Q4, Q5
of 22Hz to 22kHz. which usually have a high voltage and Q6 are BF469 and BF470 types
Fig.4 shows harmonic distortion step-up transformer with very low which are high-voltage (250V) video
versus frequency at 160W into an primary resistance. transistors, selected for their excellent
8-ohm load while Fig.5 shows distor- The same comment applies if the linearity and wide bandwidth (Ft is
tion versus frequency at 250W into a amplifier is used to drive 100V line 60MHz).
4-ohm load. Both these curves were transformers. Just to explain that, Q7 is a “Vbe-multiplier”, so-called
taken with a measurement bandwidth if you have a transformer primary because it multiplies the voltage
of 22Hz to 80kHz.
All of these distortion curves show Fig.6: this graph
a performance which is outstanding. shows the freq-
This amplifier is also extremely quiet: uency response at
1122dB unweighted (22Hz to 22kHz) 1W into 8Ω. It is
or 1125dB A-weighted. This is far just 1dB down at
quieter than any CD player! 15Hz and 60kHz
Fig.6 shows the frequency response and is virtually
at 1W into 8Ω. It is 1dB down at 15Hz flat between those
and 60kHz. frequencies.

Circuit description Reproduced by

The full circuit is shown in Fig.7 arrangement with
and employs 17 transistors and five SILICON CHIP
diodes. magazine 2006.
www.siliconchip.com.au
The input signal is coupled via a
1µF bipolar capacitor and a 2·2kΩ
resistor to the base of Q2. Q2 and Q3

Everyday Practical Electronics, October 2006 13

14 Everyday Practical Electronics, October 2006
Fig.7: the circuit uses eight high-
quality audio output transistors Performance
to give a rugged design with low
distortion. The voltage readings on Output Power . . . . . . . . . . . 200W RMS into 8Ω; 350W RMS into 4Ω
the circuit were taken with no input
signal. Music Power . . . . . . . . . . . 240W into 8Ω; 480W into 4Ω
Frequency Response . . . . . -1dB at 15Hz and 60kHz at 1W (see Fig.6)
between its base/emitter to provide a Input Sensitivity . . . . . . . . . 1·75V for 200W into 8Ω
floating voltage reference to bias the
Harmonic Distortion . . . . . . Typically 0·002% at normal listening levels
output stage and set the quiescent
. . . . . . . . . . . . . . . . . . . . . . (see graphs)
current. Quiescent current is needed
in all class-B amplifiers, to minimise Signal-to-Noise Ratio . . . . . -122dB unweighted (22Hz to 22kHz); -125dB
crossover distortion. In fact, this am- . . . . . . . . . . . . . . . . . . . . . . A-weighted, both with respect to 200W into 8Ω
plifier displays no trace of crossover Damping Factor . . . . . . . . . 75 at 10kHz, with respect to 8Ω
distortion.
Protection . . . . . . . . . . . . . . 5A supply fuses (see text)
We use an MJE340 transistor for Q7
even though a small signal transis- Stability. . . . . . . . . . . . . . . . Unconditional
tor could easily handle the task. The
reason for using a power transistor is
that its package and junction does a Although not shown in the photo- gain is set by the ratio of the 22kΩ
better job of tracking the temperature graphs, one of our prototypes used resistor to the 1kΩ resistor also con-
dependent changes in the junctions non-inductive wirewound emitter re- nected to the base of Q3. This gives a
of the output power transistors and sistors. These have been recommended voltage gain of 23 (+27dB). The 47µF
thereby gives better overall quiescent in some past designs in magazines, in bipolar capacitor in series with the
current control. order to minimise secondary crossover 1kΩ resistor sets the 13dB point of
The driver transistors are the high distortion effects. Our tests showed the frequency response to about 3Hz.
performance MJE15030 and MJE15031 no benefit in this design (probably The other factor in the amplifier’s low
made by On Semiconductor (previous- because of the PC board layout) and frequency response is the 1µF bipolar
ly Motorola). These have a minimum so they are not specified – ordinary input capacitor.
current gain-bandwidth product (Ft) wirewound emitter resistors are OK We have used non-polarised (NP)
of 30MHz. These drive the paralleled in this design. capacitors for the input and feedback
output stage MJL21193/94 transistors Two 1N4936 fast recovery diodes are coupling instead of conventional
which themselves have a typical Ft of reverse-connected across the output electrolytic capacitors because the low
around 6MHz. stage transistors. Normally, these do voltages present in this part of the cir-
Each of the power transistors in the nothing but if the amplifier is driven cuit are insufficient to polarise conven-
output stage has a 5W wirewound into clipping when driving highly tional electrolytics. Incidentally, some
emitter resistor of 0·47Ω. This rela- inductive speakers or transformers, readers may disagree with our choice
tively high value has the disadvantage the diodes safely clamp the resulting of electrolytics in the signal path but
that it causes a slight reduction in back-EMF spikes to the supply rails. the alternative of plastic dielectric ca-
power output but this has been done pacitors is not very attractive; they are
to provide improved current shar- Negative feedback large and expensive and unavailable,
ing between the output transistors Overall negative feedback is applied in the case of 47µF. Nor do we think
– an important factor in a high-power from the output stage via the 22kΩ that electrolytic capacitors, properly
design. resistor to the base of Q3. The voltage used, are the cause of high distortion

Fig.8: the power supply uses a 50V-0-50V transformer

to drive a 35A bridge rectifier and two banks of three
8000µF 75V capacitors to develop supply rails of ±70V.

Everyday Practical Electronics, October 2006 15

 This view shows the fully completed audio amplifier module. The construction details are in next month’s issue.

in audio circuits; there’s no evidence Note that this capacitor can be ce- 6·8µH choke, a 6·8Ω wirewound resis-
of it in the case of this circuit. ramic or polystyrene but must have a tor and a 150nF 250V capacitor. This
The 330pF shunt capacitor and rating of at least 250V. This is because very well-proven filter network was
2·2kΩ resistor in series with the input the signal at this part of the circuit originally developed by Neville Thiele
signal constitute an RC low-pass filter, can be as high as 45V RMS (127V and published in the September 1975
rolling off the high frequencies above peak-to-peak). Other capacitor types issue of the Proceedings of the IREE.
200kHz. The 68pF capacitor between (such as monolithics) are definitely The filter has two benefits: ensuring
Q5’s base and collector rolls off the not recommended. stability of the amplifier with reactive
open loop gain to ensure stability with The output signal to the loudspeaker loads and as an attenuator of RF and
feedback applied. is fed via an RLC filter consisting of a mains-interference signals which are

Scope1: this waveform shows the excellent square wave Scope2: these waveforms show a 150W sinewave at 1kHz
response of the amplifier, taken at 1kHz and 102V p-p into and the resulting total harmonic distortion waveform (ie,
8Ω. This equates to a power output of about 300W RMS. noise and distortion) which is at about 0·0015%.

16 Everyday Practical Electronics, October 2006

Scope3: this is the pulse waveform used to measure music Scope4: the same waveform as in Scope3, but with the
power. Note the excellent stability of the amplifier, partic- scope switched to a faster timebase. In this case, the amp-
ularly the recovery after the pulse. lifier is delivering over 240W RMS into an 8-ohm load.

inevitably picked up by long loud- two banks of three 8000µF 75V capaci- involves laying the copper tracks so
speaker leads. tors to develop ±70V supply rails. The that the magnetic fields produced by
470nF capacitors are used to provide the asymmetric currents in the out-
Power supply high frequency bypassing, while the put stage are cancelled out, as far as
Fig.8 shows the power supply and as 15kΩ 1W resistors are used as “bleeders” possible. These asymmetric currents
you can see, we’ve “gone for broke” on across the electrolytic capacitors. (think of them as half-wave rectified
this one. It’s a vital part of the perform- output signals) and their resultant
ance package and unfortunately, with PC board topology magnetic fields induce unwanted
all those big electrolytic capacitors, is Finally, the PC board has been laid distortion signals into the input stage
likely to be more expensive than the out using distortion-cancelling topol- transistors Q2 and Q3.
amp module itself. The consolation is ogy. It also has “star” earthing whereby This approach is very worthwhile
that the same power supply could be all earth currents come back to a single and constructors will not have to
used for a stereo version with two am- point on the board. This careful sepa- worry about whether the perform-
plifier modules, provided the power ration of output, supply and bypass ance of their module is as good
transformer was uprated. currents avoids any interference with as the prototype featured here.
The 500VA transformer used has two the signal currents and the distortion As long as you follow closely the
50V windings which are connected that this could cause. wiring layout in the construction
together to form a centre tap. This trans- As far as the “distortion cancel- article next month, you can expect the
former drives a 35A bridge rectifier and ling” technique is concerned, this results to be very good indeed. EPE

Get your magazine “instantly” anywhere in the world – buy and

download from the web.
TAKE A LOOK, A FREE ISSUE IS AVAILABLE
A one year subscription (12 issues) costs just $15.99 (US)
www.epemag.com

Everyday Practical Electronics, October 2006 17