R1=47Kohm R31-32-35-36-39-40=0.22ohm 5W D1-2=15V 1.3W zener
R2-9-27-28=1Kohm R41=10ohm 3W Q1-3-8-9-10-15-18=MPSA06
R3-18=10Kohm R42=10ohm 1W Q2-4-5-6-7-19=MPSA56
R4=18Kohm R43=5.6Kohm Q11-13-14=MPSA93
R5-13=3.9Kohm R44=330Kohm Q12-16-17=2N6515
R6-12=560ohm TR1=22Kohm trimmer Q20-21-22=BD379
R7-8-19=2.7Kohm TR2=2.2Kohm trimmer Q23-24-25=MJ802
R10-20=120ohm C1=10uF 16V Q26-27-28=MJ4502
R11=12Kohm C2=1.5nF 100V MKT Q29-30-31=BD380
R14-21=680ohm 0.5W C3-9-10=100pF ceramic or Mylar L1= see text
R15-22-29-30-33-34-37-38=100ohm C4-5-6-7-8=100uF 25V F1-2=5A fuse fast
R16-17-23-24=220ohm C11-13=220uF 63V .
R25-26=22Kohm C12=220nF 250V MKT .
The simplified circuit [Fig.2] shows that each sub-amplifier consists of two voltage gain stages. The first stage consists of a complementary two�stage common emitter [Q1-7] whose gain is about x2.3. The second stage is a current mirror stage [Q13-14] which drives the voltage across a load resistor tied to 0V. the gain of this stage is about x200. Thus the overall open loop voltage gain is of the order of x460 and so, as the closed loop gain is x26.7, the reduction due to negative feedback is x17.2 or about 24db. The input amplifiers are powered from ± 15V supply rails derived from R14-D1 and R21-D2. The current through the first stage [Q1] is held constant, at about 0.36mA by the floating regulator stage [Q2-3] which provides temperature compensation . The gain of this stage is set by emitter resistor R6 which provides some local negative feedback. The second stage [Q4] is loaded by two series cascode transistor [Q5-6], the first having its base tied to ground and the second having its base tied to the �15V rail. Thus the maximum collector- voltage swing on Q4 is greatly reduced , so reducing the effect of the base-collector capacitance [Miller effect] which would reduce this stages high frequency bandwidth. In summary , the presence of Q5 and Q6 improves the bandwidth and linearity. The load on Q6 is one half [Q12] of the current mirror and can be visualized as a resistor in series with a forward �biased diode. The second half of the current mirror is a common-emitter stage [Q16-17], a simple voltage amplifier except that its collector current equals [or ��mirrors��] the collector current of the other half [Q12].This stage is made up of two transistor in parallel which share the current. This arrangement was found to improve the linearity of the stage. The other sub-amplifier [Q7 to Q14] works in exactly the same way but with opposite polarity. The output stage uses the conventional Darlington emitter follower arrangement, but with three parallel pairs of driver and output transistors. A transistor Q15 is wired across the bases of the pre-driver transistors [Q18-19], providing a bias voltage to set the standing current in the output stage. Q15 is mounted on the heatsink with the aim of keeping this current constant regardless of temperature. The TR2 trimmer is used to set the value of this current.
The output DC offset voltage is set to zero by TR1 trimmer, in the input stage. In theory there should be no DC offset at the output but, because of component tolerances and consequent mismatching , there always is. TR1 is arranged to make the current in the first stage of one ��sub-amplifier�� either higher or lower than in other and so null out any residual offset. A simple low-pass filter is created by an RC networks at the input R2,C2 to reduce the bandwidth of the signal below that of the open loop amplifier and thereby eliminate the generation of any transient intermodulation distortion.
The output stage is quite substantial, using a total of six 250W power transistors. Fairly 'old- fashioned' power transistors have been used [MJ4502/802 family] in preference to some of the higher performance devices now available. They have been chosen because the die used to mount the semiconductor junction is of a large area ; the device is quite rugged and can handle high currents. The short term current capability of the output stage is, in fact, of the order of 90A, somewhat in excess of the current capability of the wiring. The rest of construction is equally massive with a steel chassis supporting six very large heatsinks. However , construction is straightforward provided that the builder has strong arm muscles and circuit alignment simple- there are but two adjustments- quiescent current and DC offset voltage nulling.The coil L1 is wound onto the body of R41. This is not a critical procedure- about 17 to 20 turnoff enameled copper wire should do nicely. Particular care should be taken in mounting the power transistors. Good quality insulating washers and bushes should be used and a generous smearing of thermal paste is essential. These transistors should be bolted to the heatsinks very tightly to ensure good thermal contact at all temperatures. One final point regarding construction. Once the amplifier has been completed and tested, it should be switched on and allowed to reach its normal operating temperature [about 20 minutes]. The amplifier should then be switched off and all the screws tightened up. Differences in thermal coefficients of expansion can result in some of the screws becoming slightly loose, particularly those holding the heatsinks to the top and bottom covers.
My own proposal, for whoever it decides makes the amplifier, is him it manufactures in two separate boxes, one for each channel, [monoblock manufacture ], separating thus and the power supply's. This will facilitate too much the mechanic and electrical manufacture. For all the manufacture, is required certain relative experience, in these sectors. The sound result is sure that it will vindicate, that tries, one and proportional performance amplifiers in the trade, costs exceptionally expensively. ETI 8/81
*
smialytedd