Unusual amp from 1987

Member
Joined 2014
Paid Member
Quick OCR + google translate + quick fix using my rusty Russian language skills not used since primary school.
It's very crude, but it explain certain things, e.g. biasing..

===================================================

SWITCH DISTURBANCES IN AUDIO FREQUENCY POWER AMPLIFIERS

Recent studies have shown that the sound character of transistor
audio power amplifiers a significant influence is exerted by the so-called switching distortion (KI).
They arise in the output stages on bipolar transistors operating in the AB mode with
current cutoff. In this mode, the transistor, as you know, goes from open to closed, and vice versa.
When it is open, at its base, due to the injection of minority carriers from the emitter into it,
it accumulates charge whose value is determined by its diffusion capacity and depends on temperature,
collector current and cutoff frequency of the transistor gain.

In order for the transistor to close, the diffusion capacitance must be discharged through external
ones connected to the base circuit, and this is only possible if the emitter locking time
transition is longer than discharge time.

If it does not exceed it, then conditions are created for the occurrence of uncontrolled base
current and, as a result of this, the appearance at the output of the cascade impulses increasing
the content higher order harmonics in the spectrum amplified signal.

In push-pull cascade during one period such conditions are created twice.
Moreover with increasing frequency, the duty cycle of the pulses decreases,
and therefore their contribution to distortion is increasing.

There are various ways to prevent KI, in particular, the use of a special mode of operation
push-pull terminal cascade of the amplifier with constant through current.
A well-established technical name for such mode does not exist yet.
Some refer it to mode A, others - to a modification of mode B.
The first come from level and nature of distortion, and the latter from energy indicators.


An example of a practical implementation of an amplifier operating in this mode
with direct through current is shown in fig. 1.

It is a simplified version of the amplifier DA-A30 firm Mitshubishi.
With a positive half-wave input signal voltage base of the emitter of the VT8 transistor decreases,
transistors VT6, VT4 open and additional base voltage is supplied to the VT8 base to prevent cutoff
of the emitter current.
With a negative half-wave of the useful signal, the base-emitter voltage decreases
transistor VT7 and through the opening transistors VT5, VT3 additional voltage is already supplied
to his base.

In this way, the cutoff emitter current transistors of the output stage
it is excluded, and the small through current that takes place practically does not impair
the efficiency of the amplifier.

It is only important to ensure a very high degree of symmetry of the transient characteristics of
the individual arms of the amplifier, since its violation leads to nonlinearity of the total
end-to-end characteristic and an increase in the coefficient harmonics.

In order to reduce the KI, special circuits are also introduced to ensure a smooth cutoff of the
emitter current transistors of the output stage.
The discharge time of the diffusion capacitance in this case increases and conditions for occurrence
a short and significant pulse amplitude is not created.
The total harmonic level does not change, but their spectrum shifts into the low frequency region where
they are easy eliminate using OOS.

This mode is widely used in amplifiers, espacially by many Japanese firms.
But this also requires strict complementarity shoulders of the amplifier and the presence of balancing elements.
An example of it circuit implementation can serve as an amplification device described in [1].

Is it possible to reduce the KI more in simple way?
To answer this question. consider what elements of the amplifier (Fig. 2) affect the discharge rate
diffusion capacity and the closing time of the transistors of the output stage.

First of all, it is the resistor R1 [2]. Voltage arising on it at the flow of current I1 is
equal to: I1R1 = (Ug1 + Ug2) - (UbeVT1 + UbeVT2)
(where Ube is the absolute value of the voltage at the emitter junction of the corresponding transistor).
It is the bias voltage for the transistors of the output stage.

Now let’s say that an increasing signal positive (negative) polarity.
Through resistors Rn, and Re1 (Rn and Re2) will flow increasing in time
load current In, and the voltage at the emitter junction of the transistor VT4
(VT3) UbeVT4 = I1*R1 - (UbeVT3 + In*Re1)
[ UbeVT3 = I1*R1 - (UbeVT4 + In*Re2)]
starts to fall and it closes.
Speed of its reduction, is determined by two UbeVT3 and In*Re1
[UbeVT4 and In*Re2], having respectively a logarithmic and linear dependence from load current.
Obviously, a minimum of KI will be observed at Re1 = Re2 = Re = 0
This case corresponds to the minimum KI.

The foregoing is illustrated by oscillograms of harmonic components
useful output signal (the first three harmonics are additionally suppressed for clarity) at K. = 0.47 Ohm
(Fig. 3, c) and К- = 0 (Fig. 3, 6) in the case
supplying a sinusoidal input signal with a frequency of 35 kHz to the amplifier input (Fig. a).

In this way, the KI spectrum was limited to 7, 8 harmonics.

Its main technical characteristics are as follows:
maximum load power output 4 Ohms - 60 W, 8 Ohms - 34 W;
frequency range 10 ... 20 000 Hz;
harmonic factor at output level 0.5 (3)
dB from maximum at frequency
1000 Hz - 0.005 (0.004). 20,000 Hz - 0.025 (0.012)%;
rated input voltage - 0.5
slew rate with off capacitor C2: 15 V / μs;
signal to noise ratio 96 dB;
quiescent current at a temperature of heat sinks of output transistors 15..90°С:
150 ... 200 mA.

Amplifier circuit diagram shown in fig. 4.

Preliminary amplification is provided by the op amp, further should cascade on transistors VT1,
VT2 included according to the OE — OB scheme,
having a high utilization factor of the supply voltage and having a low level of coefficient
harmonics. Its load is the current source on the transistor VT5.

The bias voltage of the output stage transistors is created by the cascade on
transistors VT2, VT4, described in [3], controlled by the difference of potentials
between the bases VT8, VT9.

This technical solution allowed it to make independent of the operating mode of the
VT6, UT7 transistors of the driver stage, which increased thermal stability
amplifier and additionally reduced the amplitude of the pulses of KI more than 1.5 times.

Since the voltage between the bases of transistors VT2, VT4 is less than between the bases
of silicon transistors VT8, VT9, to maintain active mode of operation of the latter,
one of the bias cascade transistors must be germanium.

Temperature stability the quiescent current of the output stage is achieved by placing the VTZ
transistor on heat sink of one of its transistors.
Resistor R23 performs circuit functions discharge diffusion capacity.
Protection mechanism is implemented with diodes VD7, VD8 and transistors VT2, VT5.
The threshold of its operation (6.7 ... 7 A) is determined by the values ​​of the resistors R24, R25.

Capacitors C5, C8 increase amplifier stability, while C6 is aperiodic in nature
attenuation of the transient process that occurs in the amplifier when exposed to a pulse signal.

The amplifier is mounted on a printed circuit -
a board of foil fiberglass 3 mm thick (Fig. 5).

Transistors VT6, VT7 equipped with copper plate heat sinks 0.5 mm thick (Fig. 6).

The area of ​​the cooling surface of the heat sinks output stage transistors 400 cm2.


An unstabilized voltage source is needed to power an amplifier 22..30 V with filter capacitors of at least 10,000 uF.

A. LOMAKIN,
B. Parshin


LITERATURE

1. Mitrofanov Yu. Economic mode And in the amplifier of fraud. - Radio, 1986. No. 5, p. 40 43.

2. Dmitriev N., Feofnlaktov N. Scheme Appliance Power Amplifier Technology. - Radio 1985, No. 5, p 38

3. Akulinichev I. Sound quality at Good. 1979.No 4, p. 26.27
 
Member
Joined 2014
Paid Member
radio_1987_2.png According to the article, only the 'upper', silicon transistor VT3 should be
placed on the heatsink to sense the output device temperature.
Drivers should have their own heatsinks.
OK, now looking for spice model for 'any low-power high frequency germanium transistor'.
Russian transistor GT308B = 2N1854 or 2N2048, there is plenty of them on Ebay...

150mW; Ucb: 20V; Uce: 12V; Ueb: 3V; Ic: 50mA/ Icmax-150ma/; Tj: 85°C; Ft: 120MHz; Cc: 8PF ;Hfe:50/120

Update:

.model GT308B pnp bf=800 br=.3 eg=0.72 cje=20p cjc=15p tf=1n tr=.4u xcjc=0.4 is=100n ikf=45m vaf=30 mje=0.5 mjc=0.5 rc=20 rb=80 rbm=10 irb=1m ise=25n ne=1.5 isc=.7u nc=1.4 mfg=GERMANIUM_USSR

new sim attached. Biasing with germanium transistor works like a champ.
 

Attachments

  • spice_model_radio1987.zip
    15.9 KB · Views: 137
Last edited:
Here is the Si version "repaired": the bases of Q8 and Q9 are connected almost directly to the OP transistors, but as it is not quite sufficient (with all Si), there is a little help from 4K7 resistors, to slightly shift the bias.
This expedient isn't very clean, but it is sufficient to demonstrate the circuit

attachment.php


Basically, the "bias cell" Q8, 9 monitors the total voltage across the OP bases, and if this voltage falls below 1.2V, it opens, allowing more bias into the OP.

The circuit has the potential to work as a class B switch upstream of the OP (pure but clean class B), as a class AB like here, or as a non-switching controller.

I have developed various amplifiers along the same lines:
This one uses a single transistor with an antisaturation diode to control the conductions of the OP:
Old-fashioned amplifiers with a new twist (1)
It can be made symetrical with a second transistor, but I didn't find any advantage to it.

Here is another variant:
Old-fashioned amplifiers with a new twist (2)
 

Attachments

  • radio_1987_09_v3L.asc
    10.1 KB · Views: 123
  • Minek5.png
    Minek5.png
    64.3 KB · Views: 1,940
Member
Joined 2014
Paid Member
I guess I'm going to build this amp.
I already ordered original GT308 transistors from Ebay; they are plentiful.

Also, this amp sims very well with LatFets - much simpler schema, no biasing needed, excellent results. So I may experiment a little...
Pic attached. Emitter resistors to be skipped.

As for beads - Elvee - one of yours Old-Fashioned amps mentioned above uses them, and you talked about them even before, more recently with Steve in another thread.

As it happens, in several of my amp builds, transistors are connected to the pcb with wires, sometimes short, sometimes longer, so I guess these beads might be helpful.
I was searching on mouser.com for these beads, but there is too many types to make sense from it. Could you help and point me to a specific type/model of the bead suitable for drivers and output transistors?
Is there a cookie cutter formula as for the parameters of these beads?
Or, even better, "one fits all"?
 

Attachments

  • radio1987_latfet.png
    radio1987_latfet.png
    56.8 KB · Views: 1,163
Member
Joined 2014
Paid Member
Also, unrelated question -
This amp from 1987 doesn't have emitter resistors - the 0.22 collector resistor
is there to sense current, for the protection circuit.

I guess it does not help at all with thermal stability, doesn't it?
Or does it have some other function besides 'current sensing'?
 
Or, even better, "one fits all"?
No, not quite: for AF work, something having an Al in the 0.3µH~1.2µH range is generally suitable

I guess it does not help at all with thermal stability, doesn't it?
Or does it have some other function besides 'current sensing'?
No, they are just for overcurrent.
With the transistors of the bias cell directly sensing the Vbe of OP, I think that the thermal stability will be OK for moderate supply voltages

>Here is the Si version "repaired" [...]

Elvee, what is the best spot to place bias pot in your scheme?
Good question: the 470/4K7 resistors are already an inelegant kludge, and throwing another resistor or two into the game is not going to improve on that; I'll need to think about it, to see if something more satisfactory is possible
 
Disabled Account
Joined 2015
A quick view of both options, I think the original one maintains the current of the drivers constant, hence the outputs are biased by base currents making the bias dependent to Hfe thermal variations, no avalanche. The second option is voltage biased, then both BD transistors should be mounted upon the outputs . To my advice, if you want to get a different kind of sound, keep the original.
 
Last edited: