C Audio xr3801 schematic

XR3801 Circuit Description and Service Notes

Introduction

These notes are intended to assist maintenance and service of the XR3801 amplifier. It is recommended that reference is
made to the relevant schematic diagrams and system diagram whilst reading this document.

The component references of the two amplifier channel electronics are appended “A” and “B” respectively. Shared circuitry
(such as that of the protection system) has no suffix. This document will refer to channel A references only. Operation of
channel B is identical except where explicitly noted. Voltage values in bold mentioned in the text are test voltages which
may be used for diagnostic purposes, although attention is drawn to the surrounding text which explains circuit operation
and may qualify such measurements.

Mechanical Construction

The mechanical structure of the unit is made up of two identical side panels, a rear panel, and a 3mm steel structural front
panel. The rear panel is fixed to the side panels by means of six M5 screws (three each side) which also secure the two
rear rack mounting brackets.

The front panel is fixed to the side panels by means of two side fixing brackets secured to the front by way of the handle
screws and to the sides by three M5 screws each side.

The main circuit board is fixed to the two heatsinks by the power device fixing screws and is spaced from the circuit board
by way of insulating bushes. The circuit board/heatsink assembly is fixed to the rear and side panels by a total of eight M5
screws.

The mains transformer is supported on the internal “transformer tray” which is fixed directly to the structural front panel.
The extruded cosmetic front panel is fixed directly to the structural front panel by two M5 screws.
Top and bottom covers complete the assembly each fixed in place by eight M4 screws.
Access to all the major components may be gained by removal of the top and bottom covers, further disassembly is rarely

required.

Circuit description

Input Stage

The input stage is built around TL071 operational amplifier IC1A, configured as a unity-gain differential amplifier. Its correct
operation is dependent upon both of its input terminals being correctly terminated and, therefore, any gain errors around this
stage may be a result of a fault in the screened cable connection to the input PCB P1030. Preset potentiometer VR2A
adjusts the gain of this stage and thereby provides adjustment to optimise the Common-Mode-Rejection of the input stage.
This preset is factory set for optimum rejection and should not be re-adjusted unless it has been necessary to change any
components around the input stage.

The trimming procedure is as follows :
Inject a common-mode test signal at 1kHz and +4dBu to the channel under test. The common-mode test signal of the Audio

Precision test system is suitable, otherwise connect the signal to both pins 2 & 3 via 51 Ohm resistors. Observe the
amplifier output and adjust VR2A for minimum output.

Power for the input stage is derived from the main +LT supply via 1W resistors R101A and R102A and shunt regulated to a
nominal +18V by D22A and D23A.

The output of the input differential amplifier is fed via 1k “Build-out” resistor R57A and twin screen cable to the front panel
level control and the returns via the “blue” core of the same cable to the main PCB to be fed to the power amplifier stage.

Power Amplifier

The power amplifier consists of a fairly conventional Class A driver stage driving a Class AB MOS-FET output stage with
Class H supply rail modulation. Each stage will be dealt with individually.

Class A Driver

The input signal returned from the level control is fed via DC blocking capacitor C53A and R59A. DC bias current for the
Class A input stage is supplied via R60A, whilst 330pF capacitor C54A prevents any extreme high frequency input signals
from reaching the power amplifier and also provides a low source impedance at high frequencies to ensure frequency
stability.

The first stage of the Class A driver consists of TR52A and TR53A configured as a long tailed pair differential amplifier.
Emitter resistors R62A and R63A de-sensitise the performance of the input stage to parametric variations of the two input
transistors. The quiescent current for the input stage is delivered by current source TR51A. Diodes D11A and D12A
provide a reference voltage of approximately 1.3V which is applied to the base of TR51A. Approximately half of this (0.65V)
will then appear across R61A (330R) which then sets the current sourced from TR51A collector at approximately 2mA. In
the quiescent state half of this current is driven through TR52A and TR53A. Hence the voltage dropped across emitter
resistors R62A and R63A will be approximately equal at 100mV.

The collector currents of TR52A and TR53A are fed via R67A and R68A to R69A and R70A respectively. Hence, in the
quiescent state, R69A and R70A should each exhibit a voltage drop of 2.7V or so.

Overall voltage feedback of the amplifier is derived through R64A and R66A. R65A and C55A connected in parallel with
R64A provide phase lead compensation to maintain good amplifier frequency stability, and a fault in either of these
components may result in RF signals being present at the output, or in unusually high distortion. C56A connected in series
with R66A gives 100% DC feedback to minimise any DC offset at the output. The resultant feedback signal is applied to the
base of TR53A.

Under normal conditions the signals at the bases of TR52A and TR53A will be identical. However, under fault conditions,
such as a DC offset at the output, the base voltages will become offset also. For example, in the event of a large DC offset
of +50V at the output a positive DC voltage will appear at the feedback point and hence at the base of TR53A. Although this
would, in theory, be the full +50V, owing to C56A being rated at only 25V, the voltage will, in practice, be somewhat lower.
However, the important issue is that the voltage is positive. In the event the voltage is negative this indicates that the
feedback network is faulty (most likely R64A itself).
The voltage at TR53A base being positive whilst the base of TR52A is close to 0V will then reverse bias TR53A base-emitter
hence turning off the transistor. Hence, no voltage should appear across R63A and R70A whilst double the normal voltage
will appear across R62A and R69A (200mV and 5.4V respectively). Should this not be the case, it indicates a fault in the
input stage itself.

The output of the input long-tailed-pair (i.e. the voltages across R69A and R70A) are fed to a second long-tailed-pair TR56A
and TR57A. The bias current for this stage is set by current source TR58A. The base current for TR58A is fed through
R72A. TR59A senses the voltage across the emitter resistor of TR58A R77A and “robs” TR58A of base current to maintain
approximately 650mV across R77A. Hence the collector current of TR58A is set at approximately 4.3mA which is shared
equally between TR56A and TR57A. C58A and C62A provide Miller Feedback around TR56A and TR57A respectively.
These capacitors set the dominant pole of the amplifier frequency response, and are therefore critical for amplifier stability.
It should also be noted that either of these capacitors becoming “leaky” (difficult to measure in circuit) will result in a DC
offset at the output.

The collector of TR57A drives the positive output more-or-less directly (more detail later) whilst the collector of TR56A drives
current mirror TR54A/TR55A via R76A. In the quiescent state R76A will show a voltage drop of around 22V, and the
current mirror emitter resistors R74A R73A and will show equal voltage drops of 320mV. Hence, for the same +50V DC
offset, described earlier, one would expect no voltage drop across any of R76A, R73A or R74A, indicating that the feedback
is attempting to correct the fault. Likewise, for a negative DC offset one would expect these voltages to be twice their usual
value. If this is not the case then the second stage (TR54A-TR59A) is at fault.

The loads for TR57A and TR55A are formed by Bootstrapped current sources TR60A/TR61A and TR63A/TR64A
respectively. Operation of the two current sources is, in principal, identical so the upper current source TR60A/TR61A only
will de described. The load current is sourced from the collector of TR61A, its base being biased through R80A. The
voltage across emitter resistor R79A is sensed by TR60A which then “robs” TR61A of base current to maintain a voltage
drop across R79A of approximately 650mV. This sets the collector current of TR61A at approximately 4.3mA. The
current source is connected to the +HT rail via R78A. The current through R78A is the sum of TR61A collector current
(4.3mA) and the current through R80A (5mA) and, therefore, 43V will be developed across R80A. Capacitors C63A and

C64A Bootstrap the current source end of R78A to the output. Therefore, with signal applied, the voltage at this point will be
approximately 100V DC with the output signal superimposed upon it.

The outputs of the two current sources TR61A and TR63A are fed through D13A and D14A to vbe multiplier circuit TR62A,
which sets the output stage bias. The bias voltage is defined by R81A, R82A and VR1A which is factory preset for the
correct bias setting of 350mV measured between the emitters of TR65A and TR66A.

Diode/Zener clamps D15A-D18A limit the maximum gate to source voltage applied to the output stage thereby setting a
current limit for protection of the output stage. Units fitted with P1042 re-entrant protection daughter boards connect
transistors TR1 and TR2 across D16A and D17A respectively. The turn-on of these two transistors is controlled so as to
reduce the allowed Gate to Source voltage as the Drain to Source voltage increases, thereby providing closer protection of
the output stage.

Emitter followers TR65A and TR66A buffer the Class A driver stage in order to provide more current to drive the output
stage. Although the MOS-FET output stage has very high input resistance, requiring little current, the parasitic capacitances
will impair its performance without the addition drive current available.

Output Stage

The output stage consists of four tiers of seven output devices connected in parallel. TR8A-TR14A form the negative half of
the Class AB output stage, whilst TR15A to TR21A form the positive half. Each device has a gate “stopper” resistor R8A­R21A and a Gate to Drain Miller capacitor C8A-C21A which prevent parasitic oscillation of the output stage.

The supply for the output stage is fed from the +LT rails (approximately 70V) via D6A and D7A for the positive and negative
halves respectively. In addition, the upper tiers of devices TR1A-TR7A and TR22A-TR28A modulate the supply to a voltage
approximately 20V greater than the output voltage when the output approaches or exceeds the 70V LT supply. There are
two important advantages to this system. Firstly, in the “Off” state the output stage is fed from a 70V supply, which reduces
the breakdown voltage requirements for the output stage. Secondly, as the power dissipation in a Class AB output stage is
proportional to the square of the supply voltage, for small (-6dB or lower) output signals, the power dissipation is a quarter of
what it would be with a conventional design.

The upper tiers of the output stage are driven from TR29A-TR32A, TR37A and TR38A for the positive side, and from
TR33A-TR36A, TR39A and TR40A for the negative side. The operation of each half is, in principal, identical so the positive
driver only will be described.

The Class H driver consists of two current sources TR29A/TR30A and TR31A/TR32A. TR29A/TR30A operate from the +HT
supply and are set to source approximately 2mA. TR31A /TR32A operate from the output signal and are set at 4.3mA.
When operating correctly, 650mV will be developed across each of R32A and R33A. Zener diode ZD2A is connected
between the two current sources and will normally show a voltage drop of 20V. Zener diode ZD1A is connected
between the Cathode of ZD2A and the output of the upper tier of output devices. Hence, in the quiescent state, the

Cathode voltage, with respect to 0V, of ZD1A will be approximately +70V, and the Anode to Cathode voltage will be
approximately 650mV. As the two current sources are unbalanced, the lower source will obtain 2mA of its 4.3mA from the

upper current source and the remaining 2.3mA of current through ZD1A, thereby biasing the upper tier of the output stage
into the off state. As the output voltage increases, less voltage will be dropped across TR32A as it maintains its 4.3mA of
collector current. When the output voltage reaches approximately 20V below the +LT rail TR32A will become saturated and
the voltage at the Cathode of ZD2A will begin to increase, biasing the upper tier of output devices into the on state.
Transistors TR37A and TR38A are configured as emitter followers to increase the drive current available to the output stage.
Hence the gate drive applied to the upper tier of devices will follow the output with an additional 20V of DC offset.

Faults in the Output Stage

Output device failure is usually in one of three modes.
A) Device short Drain to Source

B) Device short Gate to Source
C) Device open Drain to Source

Failure mode A) is usually exhibited as a DC offset at the output (or if the device is in the upper tier of devices as the
modulated rail being “stuck” at the full +140V). Such a fault will be revealed as a Drain to Source short circuit across the
offending tier of devices. To identify which device(s) is faulty measure the resistance between Gate and Source of each
device in the faulty tier with the multi-meter set to its 2kOhm range. A faulty device will show as a resistance measurement
of less than 1k1 (usually 0-100 Ohms).

Failure modes B) and C) will be exhibited as premature clipping on one half-cycle of the output. This fault will not, however,
be shown with the amplifier unloaded. In the event the fault persists with no load connected, the fault is likely to be
elsewhere. Devices suffering from failure mode B) can be easily identified with the simple multi-meter test outlined above.

Failure mode C) is a little more difficult to identify. The simplest method is to connect a “wander” lead to the Source
connection of the offending tier of devices. Touch each gate lead of that tier of devices with the “wander” lead in turn
observing the output waveform. Each time a device is shorted Gate to Source the clipping will become more pronounced. A
faulty device will be revealed by less dramatic additional clipping.

Multimeter Testing of Output Devices

Once a device has been removed from circuit it is comparatively simple to check whether it is operating correctly. The
following routine is for an N-Channel (K1058) device. The routine for testing a P-Channel device is identical with Red and
Black leads swapped.

Set the Multi-Meter to its 200 Ohm range. Connect the Black test lead to the (centre) Source lead of the device under test
and the Red test lead to the (left) Gate lead of the device under test. The meter should show a high (>200 Ohm) resistance
reading. Now move the Red test lead to the (right) Drain lead of the device under test. The meter should show a resistance
reading <2 Ohms. Briefly touch the Black test lead on the (left) Gate lead of the device and then return it to the (centre)
Source lead and check for a high (>200 Ohms) resistance reading. If any of these measurements are not achieved then the
device is faulty.

Replacement of Output Devices

Attention is drawn to C Audio Technical Bulletin “MOS-FETs”. Output devices date coded 4L3 and on exhibit slightly
different self-protecting characteristics from previous production batches. In order to permit use of these devices in this unit
a “re-entrant protection daughter board” has been developed (PCBA042). This PCB is fitted to the underside of the main
circuit board in two positions (one for each channel). Where an output device of date code preceding 4L3 is replaced with a
more recent part, this daughter board must be fitted to the affected channel(s). The fitting procedure is as follows:

1) Remove D16A, D17B, R87A from the circuit board (D16B, D17B, R87B for channel B). Clear the PCB holes of
these components.

2) Fit the daughter board by inserting the leads of the vertically mounted 0 Ohm resistors into the PCB holes vacated
by the above components, and solder in place. Trim the excess lead.

3) Connect the “flying” 0V lead from the daughter board to the main PCB 0v at R94A/B.

4) The unit may now be tested as usual.

Other Causes of apparent Output Stage Faults

Output DC offsets can be caused by faults outside of the output stage itself. In the event that no Drain to Source shorts are
measured at the output stage, it is recommended that the driver stages are checked for correct operation. The fault-finding
routine should start at the feedback point, checking the DC fault is reflected at this point, and then progress through the
driver stages, checking that the relevant current sources are operating correctly, and that the feedback is attempting to
correct the fault.

Premature clipping may also be caused by faults external to the output stage itself. Most often the fault lies within the
current protection circuit D15A-D18A. This can easily be confirmed by removing D15A and D18A from the circuit board and
observing if the fault clears. Note that, counter-intuitively, a fault D15A will show as a problem on the negative half-cycle,
and a faulty D18A will show as a fault on the positive half cycle.

Premature clipping may also be caused by a fault in the Class H driver stage. It is usually difficult to distinguish whether the
fault is caused in the Class AB stage or the Class H stage. It is usually best to confirm that all of the output devices are
operating correctly, and then eliminate the current limit circuit as outlined above. If the problem persists it is likely to be in
the Class H driver. Firstly, correct operation of the current sources should be confirmed by checking for the correct 650mV
across the emitter resistors (R32A, 33A, 34A and 35A). It is usually best to also check all transistors for correct operation by
measuring Base to Emitter and Base to Collector with a multi-meter set to the “diode check” setting. As both ZD1A and
ZD2A can effectively limit the drive to the Class H stage either of these devices being faulty can result in premature clipping.

Output Connections