Ecler PAM-1000, PAM-1400, PAM-300, PAM-600 Owners manual

PAM1400/1000/600/300

SERVICE MANUAL

SERVICE MANUAL PAM1400/1000/600/300

INDEX

- BLOCK DIAGRAM

- SCHEMATICS

- COMPONENTS LOCATION SCHEMA

- TESTING AND QUALITY CONTROL

- TECHNICAL CHARACTERISTICS

- WIRING DIAGRAM

- MECHANICAL DIAGRAM

- PACKING DIAGRAM

1

MODULE CIRCUIT 11.0504B OPERATION - DESCRIPTION

The control element is the operational NE5534. This is a very low noise operational, especially designed for
very high quality applications in professional audio equipment, control equipment and telephony channel amplifiers.
The operational is internally compensated for a gain equal to or higher than three. Frequency response can be
optimized with an external compensation capacity, for several applications (unity gain amplifier, capacitive load,
slew-rate, low overshoot, etc...).

Characteristics:

Small-signal bandwidth: 10Mhz
Output drive capability: 600Ω 10V(rms) at Vs="18V
Input noise voltage: 4nV/%Hz
DC voltage gain: 100000
AC voltage gain: 6000 at 10KHz
Power bandwidth: 200KHz
Slew-rate: 13V/µs
Supply voltage range: "3 to "20V

POWER SUPPLY

The BF871 and BF872 transistors are mounted in a common base
configuration, in a current source structure. The current sources have a
double function: polarizing the gate-source links in the MOSFETs to the limit
of the conduction and moving the voltage variations at the operational output
which are refered to ground to voltage variations refered to high voltage
power supply. The polarization point is calculated so the voltage dropout in
Rc (R112+R111) is the limit voltage of conduction of the MOSFETs (.2 to
3V), enough to carry the bias current. If we modulate in AC the base-emitter
voltage, the Ic and VRc will vary proportionally. In our configuration, as the
reference voltage Vref is constant (it is a part of the operational power
supply), we add the operational output voltage to the transistors emitter
through Re (R107-R108).

The Rc value fixes the source output impedance. We do not
recommend to raise it higher than 1KΩ because of frequency response and
slew rate reasons. This voltage circuit's gain is, as usual in a common base
configuration with Rc/Re emitter resistor, 0.45.

2

BIAS CURRENT ADJUST

The bias current adjust is performed through the
variable resistor connected between the emitters of the current
sources R110 (5KΩ). It delivers a supplementary current (it
does not go through the operational) which simultaneously
increases the voltage which falls in the Rc load resistors.

This is the easiest way of acting with just one adjust
over both branches at the same time. In order to adjust the
bias current the adjustable resistor must be varied until a
current of about 80mA circulates through each MOSFET. So,
for instance, for a PAM1400 in which there are six MOSFETs it
will be 80 x 6 = 480mA. The bias current depends on the
MOSFETs temperature and the stabilizing circuit transistors
temperature.

TEMPERATURE STABILIZING CIRCUIT

Temperature affects MOSFETs conduction in two different
ways: first, the conduction threshold voltage has a negative temperature
coefficient; second, the drain-source conduction resistance increases
with temperature. Depending on which of the two things is
predominating the temperature coefficient of the drain can be positive
or negative. In our case, in which the gate-source voltage in the
MOSFETs is very low when they conduct, the temperature coefficient of
drain current -which is positive- is predominating.

To avoid thermal runaway in the polarizing current we must
decrease the gate-source voltage as the MOSFETs get hot.
Temperature stabilization is performed by modifying the reference
voltage of both sources. If the temperature increases the Vref must
decrease so that Ic and VRc decrease and, as a consequence, the
gate-source voltage also decreases.

The circuit used is shown in figure 3. The base-emitter Vbe
temperature/voltage feature is used to obtain the final result we need.
The main idea is adequately choosing R1 and R2 to obtain the right
temperature coefficient.

3

SYMMETRY ADJUST

The threshold voltage varies much, even between
MOSFETs of the same kind. When connecting them in
parallel we must be careful that they all have the same
conduction current if we want equal currents circulating in all
of them. If the conduction voltage of P an N channels
MOSFETs is not the same they will conduct different
currents, even when we apply identical gate-source voltages.
As the bias current of the N MOSFETs must be identical to
the one of the P MOSFETs the feedback will correct the
continuous voltage at the operational output to polarize the
MOSFETs with different voltages until both conduct equal
currents.

If the operational output is not 0 V its capacity to give
voltage and current is not the same in both senses. To avoid
this we must put a symmetry adjust. It is just an adjust which
allows to vary the collector resistance of one of the current
sources (R111).

The symmetry adjust does not correct the
asymmetrical clipping saturation of the power amplifier with
real load. This happens because the conduction resistors
(Ron) of the MOSFETs N and P are not equal. Channel P has
a higher Ron than channel N. This characteristic depends on
the MOSFET's physical construction.

POWER MOSFETs

The MOSFETs used are IRFP9240 (P) and IRFP240 (N). They are assembled in a common source
configuration so they can be completely saturated.

This kind of configuration has two drawbacks compared to a common drain one: less stability (because of the
configuration gain itself) and high output impedance in open loop.

The source resistances (0.22Ω) are needed for the MOSFETs to work in parallel. E.g.: Two MOSFETs excited
by the same Vgs voltage (gate-source voltage) of 5V. If they have different transconductance curves (Id function Vgs)
they will conduct different drain currents; let's say 1A and 3A. The second one will dissipate more power and will get
hotter.

The use of source resistances tends to match the current that each of the MOSFETs connected in parallel is
conducting.

4

This resistance performs a negative feedback on the gate, lowering down the Vgs, relating to the drain current;
like this:

Vgs = Vgg - Id*Rs

The higher the Id, the lower the Vgs voltage. The gate is
protected by a zener, preventing a possible overload during an
unexpected change from overload to real clipping.

Given the high input impedance and the broad frequency response of the MOSFETs there is a high risk of
self-oscillations if all gates are excited connected to the same node. Intercalating serial resistances and ferrite beads at
the gate this possibility is minimized, because the Q of the LC network made by the inductances and gate-source
capacity is reduced.

PROTECTION CIRCUIT

The protection circuit monitors the dissipated power at the MOSFETs stage. It has two basic parts:
MOSFET Id current detection.
MOSFET Vds voltage detection.
The goal is limiting the MOSFET so it works inside an area close to the SOA, as indicated in the figure. We
chose channel N because, due to construction reasons, its SOA is lower.
ZONE A. This zone is for very low loads, around 0Ω. As the load voltage is very low, the voltage held by the
MOSFET will always be high. The protections should be activated with very low current.

Id

Fast protections and some of the slow ones are working in
this zone. The circuit that configures the fast ones is made of: D120,
D121, D123, R174, R175, R176, R177, R178, R179, C127, Q122
and Q123 for the N channel. There is also an equivalent circuit in the
P channel. These start working when there is a sudden current
variation because of a shortcircuit or a transitory. The reaction time

-from the exact moment when these things occur to when the current
stops circulating through the MOSFETs- is about 80µs.
The time constant is given by C127, R174 and R179 and the
load circuit made by the LED diode of the IC104 (opto-coupler).

C B A

Please note that in order for the protection to be activated

Vds

Q122 and Q123 must conduct simultaneously, through which R174
is linked to negative power supply, being C127(1µF) loaded very quickly through this resistance, activating the LED of
the opto-coupler, sending a pulse to the protection circuit, which will open the corresponding channel's relay, being this
way the output from the power amplifier disconnected from the load (0Ω), in this case. Q122, together with the zeners
and the base polarization resistances, configure the voltage detector (this group is in parallel with the Vds voltage of the
N MOSFET).

5

Q123, together with the resistances which make the base divider, configure the current detector (this divider
takes its voltage from one of the source resistances of a N MOSFET, which is proportional to the current circulating
through itself).

The threshold separating zone A from zone B is determinated by the D125 zener. When this zener stops
working and there is no current circulating through it because the Vds voltage is lower (let's remember this circuit is
also in parallel with this voltage) or, what is the same, the load voltage grows because it is not 0O anymore and has a
given value, like 0.5Ω to 1Ω, and the help given by D126 stops so more current will be needed for the shot. We have
climbed the first stair of the stairway of the SOA graphic.
When the zeners D124 and D118 stop working because the load voltage goes on growing (values higher than
1Ω) or -what is the same- the Vds decreases, the Q125 transistor does not receive current anymore in its base and so
it is shorted, allowing Q124 to enter conduction. This way R172 stays in parallel with the base-emitter of Q121, making
up a voltage divider with R173. This divider will climb another stair of the stairway and enter the ZONE C.

The link between the module's protection circuit and the relays' control circuit is made through IC103 and
IC104 which are, as mentioned earlier, opto-couplers, just to insulate the existing high voltages at the power amplifying
module, "90V in the case of the PAM1400, and the power supply voltage of the existing logic circuits in the relays'
control card.
Once the pulse generated by the protections is detected, the control circuitry resident in the protection card,
appart from opening the corresponding relay, returns the signal A.O. SUPPLY CONTROL to the module,
which cuts by means of Q119, Q120 and IC102 the operational's power supply.
This is the way to insure a fast and safe cut of the Id current in the MOSFETs (around 80µs time), because
they stop receiving their respective reference voltages and, consequently, their Vgs polarization voltages so they are
cut. The circuit is shown in figure 9 and its operation is very simple.
When the A.O. SUPPLY CONTROL (+10V) signal appears, the Q119 transistor starts conducting,
shortcircuiting to ground the positive power supply of the operational. On the other hand, the signal is also applied to
the IC102's LED (opto TIL112 (4N35)), which puts its internal transistor and Q120 into conduction, connecting the
negative power supply of the operational to ground.

6

7

8

ZOBEL NETWORK

This circuit tries to get a constant load impedance for the power module, in spite of the amplifier's load and
frequency, to avoid phase shifting of the feedback signal.

The values have been experimentally calculated through a study with square signal by trying to minimize the
power amplifier's ringing with very capacitive loads (2,2µF//4Ω).

The Zobel Network eliminates possible oscillations of the MOSFETs between 5MHz and 10MHz, too. This is
why it must be physically placed at the module's output, avoiding long wiring. Great care must be taken for the signal
not to be too shifted at the output, because the feedback could turn negative.

FEEDBACK

The whole amplifier is compensated with just one capacity, which places the amplifier's general pole at:

1

Fp = -------------- = 140KHz

2*π*Rf*Cf

Rf = R106 Cf = C109-C110

9

PROTECTION CIRCUIT 11.0411 OPERATION - DESCRIPTION

The circuit is configured by:

- A POWER SUPPLY.

- A THERMAL PROBE DC AMPLIFIER.

- A TEMPERATURE DETECTOR.

- A DC OUT DETECTOR PER CHANNEL.

- A CLIP CIRCUIT PER CHANNEL.

- A RESET (TURN OFF/TURN ON) CIRCUIT.

- A BINARY COUNTER PER CHANNEL.

- TWO MONOSTABLE CIRCUITS PER CHANNEL.

The circuit power supply is performed through various sources: +V, module's power supply. This voltage feeds
the relays circuit, manual reset circuit and part of the clip circuit. Alternate voltage from a transformer's
secondary (manual reset circuit).
There is also a stabilized 10V power supply which feeds the card's circuitry, made of IC301 (7805) plus the
zener D302 (Z4.7) 4.7+5 . 10V. We will also need a regulated power supply to get 14Vmax at 0.7A, which can be
obtained with IC302 (7805) plus an auxiliary circuitry that will be analysed below.
The cooling fan speed is automatically regulated in relation to the power module's temperature, which is read
by a thermal probe (LM35D), jointly linked to the heat sink.
This high sensitivity thermal probe gives variations of 10mV for every EC. This voltage is picked up and
amplified by the IC305 (LM358). Of course, there is a probe for each L and R heatsink. The output of both amplifiers is
linked through two diodes D304 and D305, making an O gate, whose cathodes go to the regulator, applying the DC of
any of them to the regulator. This provides a variable voltage at its output which oscillates from a minimum of
approximately 7V for a temperature of 20EC (cold heatsink) to a maximum of 14V for temperatures of 76EC or higher.
The gain of the amplifiers has been calculated for this temperatures window. The maximum voltage allowed by
the heatsink in order to work properly is 14V. This maximum is given by the zener D305 (Z9.1/1); as the regulator is a
7805 the voltage will be -as maximum- 9.1+5 = 14.1V. When the zener is not working (not enough voltage) the voltage
on the fan will be the output amplifiers', less 0.6V (diodes fall), plus the 5V of the IC302.

10

11

TEMPERATURE DETECTOR

This circuit is calculated to operate over the output relay opening it if any of both modules' temperature
excedes 90EC, approximately. It is made with a comparator per channel (L-R), resident in the same IC306. Both share
a reference voltage provided by D306 (TL431A), which gives excellent stability at that voltage "1%. These comparators
reveive, like the DC amplifiers, the signal from their probes, comparing them with the Vref. Once this voltage is
surpassed by any of both probes, the output of the corresponding comparator is balanced to the power supply (+10V),
acting through D307, R318, D308 and R319 over the respective bases of transistors Q301 y Q307, which makes the
corresponding relay open. This output is also connected to the THERMAL LEDs, which light up as the relays are open.

Note that each time the relay is open through any of the variables which act upon it the PROTECT LED must
light up. The circuit acting over this LED is made of R327, R328, R329, R4, R5 and Q303. When Q302 stops
conducting (open relay), Q303 receives its base current through R327, R326, R6 and the relay's coil, putting this
transistor into saturation. This way the LED is linked to the power supply (+V) by means of the group of resistances
R328, R329, R4 and R5.

12

DC OUT CIRCUIT

The circuit shown in the figure corresponds to the DC OUT of channel L. The goal of this circuit is protecting
the loudspeakers when, because of a module fail, there is some DC appearing at the output. The voltages indicated in
the figure correspond to rest state and they are given by the dividers made of R320-R322 and R332-R323.
The resistances R323-R322 are linked by their extreme to the leg 7(Q) of the monostable IC310 (4538), which
has +10V at rest state. On the other hand R320-R321 are linked by their extreme to the L output, which, in these
conditions, has 0V respect to ground. If we apply Ohm's Law to these dividers we will obtain the above mentioned
voltages.

Let's remember briefly the function of a NOR gate like the HEF4001B.

A B C

0 0 1

0 1 0

1 0 0

1 1 0

Let's suppose there is a continuous voltage appearing at the module output, because of any malfunction.
This makes the voltage dividers lose balance, no matter if the above mentioned voltage is positive or negative,
the gate goes to 0V, the base Q302 loses the current stream and, as a consequence, the relay K301 opens. The aim
of the zeners D309 and D310 is protecting the gates, avoiding the voltage in them to be higher than 8.2V when the
voltage is positive and lower than -0.6V when it is negative; as you can see, the zener plays the role of a diode.

13

CLIP CIRCUIT

The other half of IC307(4001) is used in the clip circuit. Given that we have two gates more and we just need
one for our purposes we will connect them in parallel for a higher output current and a more effective LED lighting up.

The clip threshold or point where we want the LED to light up is determined by the zener D313. In our case it is
between 0.5 and 1dB or, what is the same, when the output signal level over the load reaches a value close to that of
the power supply (+V), exactly Vout = V - 5.6, moment in which Q304 loses the base-emitter voltage stopping
conduction; this makes the zener D312 voltage disappear (0V) and the output from IC307 go to "1" logic (+10V),
making the LED light up.

14

GENERAL RESET CIRCUIT (TURN OFF/TURN ON)

TURN OFF RESET. This circuit starts working when the AC current from the transformer secondary
disappears or, what is the same, when we turn the power amplifier off by pushing the power off switch, actually
disconnecting it from mains.
Circuit operation: The AC signal present at the anode D321 is rectified by this, attenuated and filtered by R13,
R348, R347 and C322, apllying it to the base of Q306, which is conducting into saturation and, as a consequence,
Q305 is cut. When this signal disappears Q306 is cut and then Q305 has its base feeded through R345, R346 and
R14 from the +V power supply, which has begun to lose voltage -because we have just cut the mains- but, because of
the high capacity value of the filter condensers, there is enough time to saturate Q305, which puts the resistances R15
and R344 (50Ω) in parallel with the power supply (+10V) of the logic circuitry, completely discharging the capacities of
the circuit, leaving it ready for a new reset pulse -the connection one-, what warranties the new turn-on, even with very
short time intervals (.1s) between turn-off and connection pulses.

CONNECTION RESET

This is made of C315, R336 and D314. It is the classical reset circuit, used in lots of applications.
In the exact connection moment the condenser C315 is not charged, with a high amount of current circulating
through it, or a high voltage in R336. This current decreases as the condenser is charging until it disappears. At the
same time, the voltage -in the extremes- of the resistance goes from maximal, in the beginning, to 0V. This way we get
a pulse whose duration depends on the time constant RC. The aim of the diode D314 is a fast discharging of C315
during disconnection.

15

BINARY COUNTER HEF4520

This is a 4-bit double binary counter. Configured in a way in which when there is the binary code equivalent to
decimal number 5 at its output -so this is 1 0 0- it is blocked in this position, until it receives a new MR reset pulse. The
blocking action is performed by the NAND gate between legs Q2 and CP1. At this state Q2 becomes "H" one logical,
the NAND changes its state putting the leg CP1 to "L" zero logical and -as you can see in the table of functions- the
mode can not change in this conditions.

CPO CP1 MR MODE

8

L

9

X

8

H

X X H Q0 to Q3=low

The general turn-on reset initializes the counter. Every time it receives a pulse from the module opto-couplers
because of a protections shot it is counted. If during an interval of approximately 5 minutes it does not receive any
other pulse, the counter will go back to the original zero state, because it receives a new MR reset pulse from the
monostable IC311, whose time constant is approximately 5 minutes (R342,C319). This monostable begins counting
from the very first pulse received by the counter, because both are linked to the PROTECT SIGNAL from the module
and, consequently, activated at the same time.
If during this time interval (about 5 minutes) a minimum of 5 successive pulses are received, these will make
the counter block at that position. This translates into a logical "1" at the Q2 leg of the counter, a "0" at the NAND
(IC308) output; this zero makes a "1" at the output of the next NAND, giving a result of "0" at the collector of Q301, so
Q302 is not conducting and the relay K301 remains open. It will stay this way until the reset from the monostable
happens or there is a manual mains disconnection by pushing the power off button.
The reset circuit associated with the monostable is made of C320, D320, R339 and D318 (above we have
always been refering to channel L). By means of diodes D317 and D318 we build an "O" gate, with which we apply any
of the above mentioned reset pulses to the counter.

H L counter advance

9

X L no change

8

L L no change

9

L counter advance

L no change

L no change

16

17

STANDBY MONOSTABLE

The only thing left is the function of the monostable made of IC310 (4538).
Like the counter and the monostable IC311 (4538), this circuit is connected to the PROTECT SIGNAL, too. Its
output is "1" in rest state and becomes "0" during an approximate time of 1.3 seconds, which is given by the constant
RC of the circuit R341 C316.

This leads to two situations: First, putting a "0" in one of the legs of the NAND (IC308) generates the
immediate opening the relay, as we have seen before. On the other hand the voltage divider of the DC OUT circuit is
put off balance. The monostable time is calculated to be long enough to unload the capacities of C312 and C313. This
way we get a DC OUT circuit initialization as we had done a manual reset (disconnection from mains), causing the
tipical turn-on STANDBY time for each of the disconnections of the relays because of the protections shooting. Let's
take into account that the system is locked after the fifth disconnection.