This is only a preview of the April 2015 issue of Silicon Chip. You can view 34 of the 96 pages in the full issue, including the advertisments. For full access, purchase the issue for $10.00 or subscribe for access to the latest issues. Items relevant to "Appliance Insulation Tester":
Items relevant to "Build A Low-Frequency Distortion Analyser":
Articles in this series:
Purchase a printed copy of this issue for $10.00. |
Vintage Radio
By Ian Batty
The AWA 897P: Australia’s
first transistor radio
Designated the model 897P, Australia’s first
transistor radio was developed by AWA and
first marketed in November 1957. It uses seven
transistors, is built onto a metal chassis and
uses the same case as its valve predecessor.
W
ITH THE centenary of Amalgamated Wireless Association’s
listing on the Sydney Stock Exchange,
it seems timely to review their first
transistor radio. Formed as a result
of the rivalry between the German
Telefunken and British Marconi companies, AWA has been a pioneering
presence in radio and electronics here
in Australia and around the world.
The first chairman, Hugh Denison, eventually made way for the
better-known Ernest Fisk. Fisk’s later
knighthood and towering presence
84 Silicon Chip
eventually extended to the familiar
“Fisk Radiola” badges on many of
the company’s radios and even to the
naming of the former Imperial Wireless Chain station as “Fiskville”. For a
more complete history on this, see the
Historical Radio Society of Australia’s
Radio Waves, July 2013.
Early transistor radios
Regency (USA) marketed the first
successful all-transistor radio in 1954,
designated the TR-1 (SILICON CHIP,
April 2013). Given the stupendous en-
gineering task (a learning “cliff” rather
than a “curve”), AWA’s offering of the
897P in November 1957 is remarkable.
The initial offering used a mix of “2N”
and “OC” transistor types, the latter
echoing AWA’s early association with
German company Telefunken.
The CSIRO had begun investigating
semiconductors in 1953, initially with
the assistance of Bell Laboratories. Dr
Louis Davies had spent six weeks at
Bell Labs and came back from the US
armed with two essential precursors
to making transistors: the technologies for purifying germanium and for
growing single crystals of germanium.
A subsequent symposium attracted
the attention of industry, rather as the
original Bell Labs seminars had in the
USA. Although all the “big four” Australian companies attended, it was the
work of Ted Watt and Henry Banks that
led to AWA starting local production
in 1958. Watt and Banks had attended
an engineering “apprenticeship” at
RCA and their efforts were pivotal in
AWA’s entry to the local market.
The 897P transistor radio used an
existing valve portable case design
from the model 581PZ. Like many sets
of the era, it also used a pressed and
punched metal chassis and the parts
were all installed by hand. It’s quite
similar the RCA Victor Transistor Seven, even down to the 2-gang volume
control (the RCA set was described in
the October 1956 issue of Radio, TV
and Hobbies).
Two restorations
This article summarises two restorations, as I was very generously loaned a
number of 897 variants by the HRSA’s
Ray Gillett. And while on the subject
of variants, the original 897P used
RCA 2N219 (converter) and 2N218
transistors in the RF/IF section and
Telefunken OC602/604 types in the
audio stages. By contrast, the later
897PX/PY/PZ used all RCA types,
with 2N408 & 2N270 types now in
the audio amplifier. These are all
siliconchip.com.au
Fig.1: AWA 897P is a 7-transistor superhet design with 455kHz IF stages. Transistor VT1 is the converter stage, VT2
& VT3 are IF amplifier stages, VT4 is an audio preamplifier, VT5 an audio driver stage and VT6 & VT7 function as a
push-pull output stage.
“second generation” alloyed-junction
types.
The 897P: first look
As stated, AWA’s 897P uses a press
ed and punched metal chassis just
like the Bush TR82C (SILICON CHIP,
September 2013). It uses seven transistors, five of which are fitted into
chassis-mounted rubber grommets
with their leads then wired to adjacent
solder tags.
By contrast, the two output transistors are held in heatsink clips which
are screw-mounted on the underside of
the chassis. This differs from the later
PX, PY & PZ models which (strangely)
also have their output transistors
mounted in rubber grommets, defeating any possibility of heatsinking.
Unfortunately, the grommet-mounting technique means that the transistor leads are underneath the chassis.
This means that unless the chassis is
removed, the only circuit access, either
for measurement or signal injection, is
at the aerial coil, the volume control
and speaker terminals.
As shown in the photos, the chassis
is mounted in a substantial leather case,
with the wrap-around shell closed off
by stitched-on ends. The front dial
turns easily with a direct drive. It sits
within the front escutcheon which
siliconchip.com.au
also contains the speaker grille. The
volume/power switch is mounted on
the righthand end of the set.
897P circuit details
Fig.1 shows the circuit details of the
AWA 897P. Many of its components
were common to the valve era and
apart from the the transistors and the
low-voltage electrolytic capacitors,
they appear much the same as those
found in portable battery valve sets.
Some models use the classic Philips
tuning gang with rounded edges on its
frame, brass plates and identical aerial
and oscillator sections. The 897P and
897PY models use gangs with 445pF
per section and a 470pF padder, while
the 897PZ and 897PX models use
385pF sections and a 420pF padder.
There are also minor mechanical differences, with the 897PZ and 897PX
models using a different dial scale.
The circuit itself is a fairly conventional 7-transistor superhet design.
The RF signal is picked up by antenna
rod T1 and tuned by C3, one section of
the tuning gang. The other section, C4,
tunes the local oscillator. The tuned RF
signal is then fed to the base of PNP
transistor VT1 via coil T2.
VT1, a 2N219, is the converter stage
(ie, a combined local oscillator and
mixer) and this uses collector-base
feedback (ie, via T3’s tuned primary
and a tapping on T2’s primary) to
maintain oscillation. While this works
reliably, it does increase the amount
of local oscillator (LO) radiation back
out through the antenna rod.
The mixer’s output feeds the primary of the first IF transformer (T3).
This uses a tapped secondary winding
to match into the low base impedance
of the first IF amplifier.
The two following IF transformers
use tapped primaries and secondaries,
with VT2 & VT3 (both 2N218 transistors) functioning as IF amplifiers.
The two IF amplifiers operate similarly to those in most other sets. AGC
action is applied to the first IF amplifier
stage (VT2) alone, reducing base bias
and thus the total collector current
on strong signals. In common with
other designs, reducing the collector
current reduces current gain and thus
the stage gain.
The applied AGC voltage (at VT2’s
base) is quite small, with the base bias
dropping by only about 70mV at full
signal. This is much less than in many
other sets and is due to the voltage
divider connected to VT2’s emitter.
Instead of allowing the emitter voltage to also drop with incoming AGC
(and thus “softening” the response
somewhat like a remote cut-off valve),
April 2015 85
This view shows the general layout of the major parts on the top of the 897P’s
chassis. Transistors VT1-VT5 were mounted by pushing them into rubber
grommets from underneath the chassis.
Unlike VT1-VT5, the two OC604 transistors used in the push-pull output stage
(VT6 & VT7) were attached to the underside of the chassis using metal clamps.
Note the point-to-point wiring technique used to assemble the circuit.
VT2’s emitter voltage is held nearly
constant. This allows the 70mV drop
in base voltage to take VT2 from its
normal forward-bias value of around
130mV down to virtual cut-off at
90mV, much like a sharp cut-off valve
characteristic. So because of the emitter voltage divider, don’t expect to
measure signal strength by the fall in
VT2’s emitter voltage.
The second IF stage operates with
fixed bias (as usual). Note that both IF
amplifiers are neutralised (using C12
& C17) to prevent instability due to
collector-base feedback.
The demodulator uses a conventional diode (MR1) and this feeds
audio to volume control R16. It also
feeds a DC voltage back into the bias
86 Silicon Chip
network for the first IF amplifier, in
common with other designs. Stronger
signals reduce the bias on the first IF,
thus controlling its gain. As with all
AGC systems, the net effect is to keep
the audio signal fairly constant with
varying RF signal strengths.
Audio stages
The 897P uses three audio stages:
preamplifier stage VT4, driver stage
VT5 and a Class-B push-pull output
stage based on VT6 & VT7. In common with the Raytheon T-2500 and
the Bush TR82 radios, the audio section uses transformer coupling. While
this adds complexity and potentially
reduces both high-frequency and lowfrequency audio response, transformer
coupling gives optimal power gain and
thus improves sensitivity.
As an aside, this design choice implies that the set’s RF/IF gain was less
than optimal and that the deficiency
was compensated for in the audio
section.
The biasing in the audio preamplifier and driver stages is similar to that
used in the IF amplifiers and works
identically. However, larger emitter
bypass capacitors are used so that they
are effective at audio frequencies.
The set is unusual in using a 2-gang
volume control and the “original”
897P model is readily identified by
this feature and the use of “OC” series
transistors in the audio section. But
why use a 2-gang volume control? The
articles in the HRSA’s Radio Waves for
July 2013 give two possible reasons:
(1) to prevent overloading and breakthrough at low volume with strong
signals and (2) to reduce the effects of
preamplifier noise.
So which of the two is correct, or
are they both correct? Well, bypassing
the second volume control pot (R22)
still resulted in effective control but
made the set noisy at low volume.
On the other hand, reinstating R22
and bypassing R16 solved the noise
problem but caused serious audio clipping and distortion on strong signals.
So the answer is that both of the possible reasons given for using a 2-gang
volume control are correct.
The output stage uses a conventional transformer-coupled Class-B
push-pull circuit based on VT6 & VT7.
As in the Bush TR82C, feedback is applied from the speaker to each output
transistor’s emitter terminal.
In addition, the output stage uses a
voltage divider to give about 160mV
of base bias to each output transistor.
In common with other Australian
designs, the lower end of this divider
includes a 130Ω NTC thermistor (ie,
its resistance falls as the temperature
increases).
The combined effect of the thermistor and transistor characteristics
ensures a fairly constant collector
current in the output stage, regardless of temperature. This arrangement
minimises crossover distortion and
protects the output transistors from
thermal runaway (due to increasing
current), thereby saving them from
damage due to overheating.
The output stage drives a large 5 x
7-inch oval speaker, which gives good
siliconchip.com.au
This internal view of Telefunken’s
OC604Spez transistor shows the metal
half-cyclinder (at top) that was used
for heatsinking.
efficiency and volume. Power came
from a single Eveready 276P 9V battery and this was capable of powering
the set for some 300-plus hours with
normal use.
The OC604Spez
The most obvious difference between the 897P and its successors
is its use of Telefunken “OC” series
audio transistors. Both the OC602
small-signal and OC604 output types
use glass encapsulation. This glass
encapsulation provides the hermetic
seal that’s vital for germanium devices
but it impedes heat dissipation.
Telefunken’s answer to this was
the OC604Spez(ial) transistor, a glassencapsulated type with internal heatsinking that allows it to deliver up to
500mW from a 6V supply. As shown in
the above photo, the heatsink consists
of a metal half-cylinder that’s attached
to the base slice (the “Germaniumplattchen”).
879PX/PY/PZ differences
By contrast, the PX/PY/PZ models
use RCA-derived 2N408 transistors for
the audio preamp and driver stages
and 2N270 types for the push-pull output stage. Despite this, these variants
perform similarly to the 897P.
Apart from that, they’re recognisable
because they also have their output
transistors mounted in grommets. The
other visible difference is the 897P’s
use of a Philips-style tuning gang with
rounded corners in its frame compared
to the more common square-cornered
types.
Restoring an 879PX
As it came to me, the 897PX set had
a “scratchy” volume control pot and
there was no audio output. A quick
check of the DC voltages revealed
siliconchip.com.au
that all was OK in this department
so the capacitors THEN came under
suspicion. I needed many millivolts
of signal at the volume control to get
even a “squeak” of output and further
checks showed that coupling capacitor
C26 (20µF) was open-circuit, as was its
companion C23.
Replacing both these capacitors immediately brought the set to life. Further checks then revealed that it had
an audio sensitivity of about 200µV for
50mW of output, which is excellent.
Audio response?
What about the audio response? In
a word, it was “rubbish” with a response of only about 190Hz to 1.3kHz.
That just had to be wrong. My chief
suspects were the two top-cut capacitors, ie, C28 (10nF) across the driver
transformer’s primary and C30 (100nF)
across the output transformer’s primary. De-soldering both dramatically
increased the high-frequency response
of the audio stages to 13kHz.
Replacing both capacitors restricted
the response to 2kHz. This was quite
acceptable, especially given the RF/IF
section’s bandwidth of about ±1.7kHz.
As an aside, removing C28 extended
the frequency response to about 6kHz
but worsened the set’s weak-signal
noise figure by over 3dB. So it appears
that the heavy “top-cut” technique was
a quick-and-dirty way to improve the
subjective performance.
One thing of note is that this set has
an unused connection on the ferrite
rod. It turned out to be an “aerial terminal” tap on the tuned winding and
provided a convenient direct signal
injection point for testing.
Now for the 897P
When I obtained the 897P, I found
an envelope containing a paper inside
the case headed “Freddy’s Hire Purchase Contract for AWA Tranny 2412-1957”. This was proof positive that
this set was indeed made in 1957. The
contract, with Industrial Acceptance
Corporation, shows a total amount
payable of £52 13s or about $105.30
before adjusting for inflation.
In today’s money, after inflation
is taken into account, that’s about
$1500 – double the cost of a high-end
smartphone. The purchaser was obligated to pay off the set over a period
of 19 months.
Cosmetically, my 897P came to me
with its leather case in poor condition.
OATLEY ELECTRONICS
APRIL/MAY SPECIALS
4 CHANNEL
UHF REMOTE CONTROL
Rolling Code *Preassembled* Each Channel can be
Momentary/Latching, and be reset by a Switch.
1x RX+1x TX8+
2x Microswitches:
Additional Tx’s:
TX8…. $15Ea
K239P1
$
40
VERY SMALL FM
TRANSMITTER
Complete inc. Mic.
* Only 25mm Wide
* Needs 1XLR44 Batt
(Not Included)
CLEARANCE!
$
5mW/650nM
LASER MODULES
5mW/650nM
*3V operation!
*6mm Diameter
K189
12ea
LASERMODS
PACK OF 3!
$
For Use With Laser Light Shows Only
10
PACK OF 3
MUSIC – AUDIO TRIGGERED
RGB STRIPLIGHT
Audio Triggered
with IR Remote
* Includes DC
connector and a
5M Roll of RGB
Striplight
MUSICRGB:
$
POWER SUPPLY
KIT KC354
15
for the
package!
!!fREE!!*
Mains in; 6-30V DC Out (One R Change)
*Ask for your free power supply when you
purchase either the K239P1 or MusicRGB
20W LED PIR
FLOODLIGHT
20WPIR
$
20
20W 12V
Pure White
Floodlight
with PIR
PO Box 139, ETTALONG BEACH
NSW 2257 PH: (02) 4339 3429
For a firm shipping cost send an email with APR/MAY
as the subject, and include an address/order/tel. no.
Send to: branko<at>oatleyelectronics
April 2015 87
These two photos show sections of the 897PZ chassis. Unlike the 897P, this
model used 2N270 output transistors which were mounted in grommets, just
like VT1-VT5. The other visible difference is the 897PZ’s use of a “non-Philips”
tuning gang with square edges.
The leather was dull, though not badly
scratched, and the stitching on one of
the end cheeks had degraded, allowing
the end to detach.
The case would have been machinestitched during manufacture but I was
able to locate hemp cord at a local craft
shop of the same thickness as the original. After removing the original thread,
I was able to counter-stitch and restore
the case to its original appearance.
Electrically, the set was completely
dead, apart from an audible turn-on/
turn-off “click”. The reason for this
wasn’t hard to determine – audio
transformer T7 had an open-circuit
primary. This was rather odd since it’s
a low-voltage, low-power item.
Replacing this transformer with
a similar inter-stage transformer restored the set to life.
How good is it?
As noted above, the frequency response is around 190-1900Hz from
the volume control to the speaker and
around 210-1600Hz from the antenna
to the speaker. Removing that pesky
10nF top-cut capacitor (C28) from
VT5’s collector extended the response
to around 2.6kHz, and the difference
in quality was quite noticeable.
The audio performance was otherwise quite good: at 10mW output and
400 Hz, the total harmonic distortion
was just 2%, while at 50mW, the distortion was about 3.3%. This rises to
around 8% as the set just begins clipping at 220mW output and is 12% for
88 Silicon Chip
250mW. Removing the output stage’s
feedback gave a worst-case figure of
8% distortion at just 50mW output.
The Pye Jetliner transistor radio
described in the September 2014 issue
has a diode-biased output stage and
was able to maintain low distortion
down to 50% battery voltage, with
little evidence of crossover distortion.
By contrast, the 897 is unable to cope
nearly as well with falling battery
voltage because it uses voltage-divider
biasing. As a result, it gives audible
crossover distortion when the supply is down to 5V. At this voltage, it
clipped at an output of just 50mW.
The 897’s selectivity is ±13kHz at
60dB down, reflecting the presence
of three double-tuned IF transformers.
Although the 897, like the TR82C, uses
four audio stages, the 897’s design fails
to exceed the TR82C’s performance,
achieving 250µV/metre at 600 kHz and
150µV/m at 1400kHz with the volume
control adjusted for a signal-to-noise
ratio (S/N) of 20dB.
At full gain, the model 897 achieved
125µV/m at 1400kHz with an S/N of
17dB (note: all inputs are for 50mW
output and 30% modulated at 400Hz).
Given that Bush’s TR82C is a later design using alloy-diffused AF116/117
transistors in the RF/IF section, the 897
(which uses alloy-junction transistors)
performs quite well, especially as it
was AWA’s first transistor radio.
The AGC control is average and a
26dB signal increase from 60µV to
1200µV at the aerial terminal gives a
6dB increase in the audio output. The
only reservation is that this set went
into violent oscillation with a radiated
signal strength much above 30mV/m.
Graham Moore (Radio Waves, July
2013) notes the “sharp cut-off” characteristics of a transistor AGC circuit
and the 897, with its emitter tied to
about 0.47V by voltage divider R8 &
R9, certainly exhibits this.
By contrast, most sets use a single
emitter resistor to ground, allowing
something closer to a medium/remote
cut-off. With an IF stage gain of about
30dB, applying AGC only to the first IF
can’t reduce the stage gain by any more
than 30dB before the transistor is left
with virtually zero collector current.
Basically, in order to achieve greater
AGC range, the converter must also be
controlled, either by using an auxiliary
diode circuit as in the Pye Jetliner or
by applying AGC to a mixer that’s fed
by a separate local oscillator.
The 4-valve predecessor
When I dusted off a somewhat
sorry-looking 4-valve AWA 581PZ
and applied power, I was rewarded
with absolute silence. I’ll leave its
restoration details for another article.
However, when I did eventually get it
working, the 581PZ (which looks just
like the 897P) had a sensitivity of about
360µV/m at 600kHz and 250µV/m at
1400kHz.
So the 897 appears to have roughly
double the sensitivity. In practical
terms though, the two sets would
have almost identical performance
except on the lab bench. I expect that
the 581PZ’s audio performance (output power, distortion and frequency
response) will be similar to that of its
all-transistor 897 successor.
Acknowledgments
Many thanks to Ray Gillett of the
HRSA for his very generous loan of a
half-dozen variants of the 897.
Further reading
(1) For a more complete history of
AWA, see Radio Waves, July 2013.
(2) For more detail on the 897P, see
the July 2013 Radio Waves articles by
Graham Moore and Ian Malcolm.
(3) For more detail on early Australian transistor manufacture, see Mark
P. D. Burgess’ article – go to https://
sites.google.com/site/transistorhistory/
and navigate to Australian semiconSC
ductor manufacturers.
siliconchip.com.au
|