This is only a preview of the July 2010 issue of Silicon Chip. You can view 28 of the 104 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 "An Intelligent 12V Fan Controller":
Items relevant to "PICAXE Project Development System":
Items relevant to "Lab-Standard 16-Bit Digital Potentiometer":
Items relevant to "Dual-Tracking ±19V Power Supply, Pt.2":
Purchase a printed copy of this issue for $10.00. |
Vintage Radio
By RODNEY CHAMPNESS, VK3UG
Reflex receivers – why they
were necessary
Above: the Kriesler 11-41 was a popular
4-valve reflex receiver from the 1950s.
Valves were expensive in the early
days of radio and so designers came up
with clever techniques to minimise the
valve count. One technique was known
as “reflexing” and involved using the
same valve to work as both an RF or IF
amplifier and as an audio amplifier.
C
OMPONENTS such as tuning capacitors, inductors (both fixed and
variable), resistors and fixed capacitors were in common use during the
Spark Era, at the start of last century.
However, valves (when they finally
made an appearance) were initially
extremely expensive and fragile. Suffice to say, they didn’t have a long life.
Initially, obtaining a high vacuum
inside a valve was quite difficult and
most of the early valves were of the
“soft” variety. This meant that they
94 Silicon Chip
had a small amount of gas left inside,
due to manufacturing limitations. As a
result, these valves were rather variable in their performance, even between
supposedly identical types.
Another problem that had to be
overcome was how to maintain a
good vacuum. This could only be
achieved if the glass and metal leads
through the glass envelope had the
same coefficient of expansion. If the
coefficient was different, air would
eventually leak into the valve and it
would become gassy. Occasionally,
even today, a valve with a purple glow
inside it will be seen and this is often
an indication that the glass to metal pin
seal is not perfect and air has leaked
into the valve.
Incandescent light globes were the
first items to have metal pins or wires
protruding through a glass envelope.
However, this created no real problem,
since the vacuum created was satisfactory for their operation and the glassto-metal seals were not as critical. In
some cases, the globe was filled with
an inert gas such as nitrogen to prevent
evaporation of the filament.
One problem with valves was that
the metals used inside them (ie, for
the elements and filaments) had to
be carefully selected, otherwise they
could emit gases when they became
hot. These gases could then “poison”
a valve and adversely affect its performance.
So early attempts at making valves
into viable amplifying devices encountered many difficulties. However, their
potential to revolutionise radio was
obvious and so a great deal of effort
was put into solving these problems.
It is for these and other reasons that
valves were by far the most expensive
and fragile components in early valve
receivers. As a result, the engineers
and experimenters of that era searched
for ways in which each valve could
be made to do more than one job, to
minimise cost.
Reflexing
One of the first to achieve dual
usage of valves was W. H. Priess, a
US Navy worker who patented the
principle of reflexing during WWI.
This technique involves passing a
siliconchip.com.au
signal through the same amplifier
twice, at two significantly different
frequencies – usually once at a radio
or intermediate frequency (ie, RF or IF)
and then at audio frequency (ie, after
the modulated signal from the IF stage
has been detected).
Initially, there was little interest in
reflexing for a couple of reasons. First,
it was no doubt kept a military secret
during the war and so was not widely
known during those years. Second, the
principle, although sound, initially
proved tricky to implement. However,
valves were still very expensive in the
1920s and this led some manufacturers and enthusiastic experimenters to
refine the technique and this eventually yielded good results
In theory, reflexing meant that one
valve did the job of two. However, as
there was always some compromise in
the operating conditions of the valves
for each different frequency, the actual
improvement was always somewhat
less that this.
Reflexing was used for only a relatively short time overseas but in Australia, it was still being used in some
receivers as late as the early 1950s. It
was initially used in high-end receivers in the early 1930s, then in receivers at the bottom end of the market to
reduce valve count (and thus cost).
This is not to say that reflex receivers
performed poorly because they were
aimed at the bottom end of the market. Some sets did leave something to
be desired but others were very good
receivers.
How reflexing works
The mere thought of restoring a
radio receiver with a reflex stage has
sent cold shivers down the backs of
many enthusiasts. As a result, such
sets have either been shunned or used
simply as non-working show items.
I must confess that during my early
days as a serviceman I wasn’t all that
keen on dealing with reflex sets.
There were just so many wires and
components going here, there and
everywhere and a real mix of signal
leads. However, most of the reluctance to service reflex sets (as with
AGC stages) was due to the fact that
servicemen lacked the test equipment
that we have today.
Originally, reflexed stages were included in tuned radio frequency (TRF)
receivers. These sets typically had an
RF stage followed by a detector. The
siliconchip.com.au
V2: 6AD8
6AN7
PLATE
6
IFT1
455kHz
9
1
2
3
IFT2
455kHz
6AN7
SCREEN
7,8
33nF
27k
500pF
10nF
1nF
B
X
15k
250pF
47k
A
X
4.7nF
VOLUME
500k
390Ω
1M
AGC TO
6AN7
+200V
1M
TO GRID
OF 6M5
AUDIO
OUTPUT
150Ω
FROM
SPKR
VOICE
COIL
(NFB)
Fig.1: the IF (intermediate frequency) amplifier, detector and reflexed audio
stage of the Kriesler 11-41 & 11-49 radio receivers. The detected audio
signal at pins 7 & 8 of the 6AD8 are fed back to the grid via the primary
of IFT2, a 47kΩ resistor, the volume control, a 4.7nF capacitor and the
secondary of IFT1.
audio output from the detector was
then fed back to the input of the RF amplifier valve. From there, the resulting
amplified audio signal was typically
applied to an audio output stage.
In a well-designed receiver, this
system could be made to work quite
well. However, most of these early
TRF sets were built breadboard style,
so layout could be (and usually was)
quite critical. Unfortunately, due to
incorrect wiring techniques and the
inevitable stability problems that followed, many people soon came to the
conclusion that reflex circuits were
“cranky” and best left alone.
That wasn’t to be the end of reflex
receivers, however. When superhet
receivers were subsequently developed in the 1930s, the breadboard
style of construction was quickly
abandoned. Instead, the parts were
mounted on a metal chassis, with the
critical components shielded. This
was necessary to ensure stability and
consistent performance and this style
of construction quickly became the
standard technique for Australian
manufacturers.
As a result, some manufacturers
decided to see if more stable highperformance reflex receivers could be
developed using metal chassis and improved shielding. Their efforts proved
successful and an early example is the
Radiolette model 31/32.
There were others, as I quickly
discovered when I looked through
the 1938 edition of the Australian Official Radio Service Manual (AORSM).
There were at least eight mainstream
manufacturers that had at least one reflex model: Aristocrat, Astor, Croyden,
HMV, Hotpoint-Bandmaster, National,
Fisk-Radiola and Westinghouse. In
fact, Fisk-Radiola and its badged stablemate Hotpoint-Bandmaster lead
the way with quite a few models, both
mains and battery operated.
Kriesler 11-41/11-49 receiver
The final volume of the AORSM has
no manufacturers with reflex receivers. In fact, the last reflex receiver
featured in the AORSM is the Kriesler
11-41/11-49 (in Volume 12, 1953), so
Kriesler appears to be the last manufacturer of these sets in Australia.
The Kriesler 11-41 & 11-49 were
4-valve mains-operated mantel receivers, the two models being almost identical. So let’s take a closer look at the
circuit to see how reflexing worked.
Fig.1 shows the IF (intermediate
frequency) amplifier, detector and
reflexed audio circuitry stages of these
models. The front-end uses a 6AN7
converter valve and this feeds a signal
July 2010 95
6G8G
3
RF INPUT FROM
ANTENNA TUNED
100pF
CIRCUIT
speaker’s voice coil to the bottom of
the volume control, to improve audio
quality.
+50V
100pF
6
4
8
1.75M
7
2
500k
1nF
50pF
AUDIO TO
6V6GT
OUTPUT
70k
200k
+30V
1M
10nF
50nF
10nF
3M
Full AGC
3.3M
5
50nF
+170V
HT
100k
–2 TO –22V
BIAS & VOLUME
CONTROL LINE
Fig.2: the reflexed circuitry in the Astor GR (Football) TRF receiver. In this
circuit, the detected signal on pin 4 of the 6G8G is fed back to the grid via
a 500kΩ resistor, a 10nF capacitor and a 1.75MΩ resistor. A variable bias
(-2V to -22V) line controls the volume (the lower the bias, the greater the
gain).
IF/AF AMPLIFIER
10nF
IF & AUDIO
INPUT
IFT
1nF
TO AUDIO
OUTPUT
VALVE
100k
HT
Fig.3: in this
circuit, the screen
of the IF/AF
amplifier is used
as the plate for the
reflex stage audio
output. This gives
a gain of about 1020, depending on
the valve.
47 µF
centred on 455kHz to an IF amplifier
stage (V2, 6AD8).
In greater detail, the signal from
the converter is fed through a doubletuned first IF transformer (IFT1) to the
grid of a 6AD8 IF amplifier. The signal
is then amplified and applied to another double-tuned IF transformer (IFT2).
The output from this transformer is
then fed to the 6AD8’s detector and
AGC diodes which are tied together
at pins 7 & 8. A 47kΩ resistor, 500kΩ
potentiometer (volume) and 150Ω
resistor form the load for these diodes.
The resulting audio signal is taken
from the wiper of the volume control
and applied via a 4.7nF capacitor to
the top of a 500pF capacitor (which
acts as an RF bypass for the first IF
transformer). IFT1’s secondary winding has virtually no effect on the audio
signal which is now fed directly to the
grid of the IF amplifier.
96 Silicon Chip
As a result, the 6AD8 amplifies the
audio signal along with the IF signal
and the amplified signals appear at the
plate. IFT2’s primary has little effect on
the audio signal which is now developed across a 15kΩ plate load resistor.
The 1nF capacitor at the bottom of
the primary winding acts as an RF
bypass. It’s effective at IF frequencies
but its impedance at the higher audio
frequencies is around 50kΩ. However,
this does shunt the 15kΩ plate load
resistor to some extent, which reduces
the audio performance at higher frequencies.
The amplified audio signal across
the 15kΩ resistor is fed via a 10nF
capacitor and a 47kΩ stopper resistor
(not shown) to the grid of a 6M5 audio
output valve which then drives the
loudspeaker via a transformer. The
390Ω resistor (bottom, right of Fig.1)
provides negative feedback from the
It’s not all that common to see the
full AGC voltage applied to a reflexed
IF/audio stage but Kriesler has done
this here. In this circuit, AGC is taken
from the top of the 47kΩ resistor (at the
top of the volume control) and applied
via two 1MΩ resistors to the grid of the
6AD8 reflex stage. Because its plate
load resistor is only 15kΩ, the 6AD8
does not have high audio gain – only
about 10-12 times.
As a result, the IF amplifier conditions are not far from normal and
the stage is not likely to overload on
strong signals. By contrast, Astor reflex
receivers often used a 70kΩ plate load
resistor in the reflex stage.
Because AGC is applied to the reflex
stage, its gain at both IF and audio
frequencies is reduced with increased
AGC control voltage. If not done correctly, this can mean that the actual audio output from the set can be reduced
with increasing signal (as mentioned
in a previous column on AGC).
Fortunately, Kriesler got it right in
this set. There is no reduction in volume when tuning a strong station, as
compared to that from a weak station.
By the way, it is quite easy to
compare the performance of a reflex
set with a more conventional circuit
without reflexing. In Fig.1, there are
two points marked “A” and “B”. If the
“A” end of the 4.7nF capacitor is lifted
and connected to “B” and the 10nF
capacitor is removed (ie, the wiper
of the volume control now feeds the
audio signal directly to the 6M5 audio
output stage via the 4.7nF capacitor),
then the set reverts to non-reflexed
operation.
Of course, the audio gain will be
down as there is now only one audio
stage in the receiver rather than two
when it is wired as a reflex set.
As a result, when this is done, the
variation in the audio level is quite
noticeable, particularly if the received
signal is relatively weak. However, if
the signal is strong, there is not a great
deal of difference due to the fact that
the audio output from the reflexed
stage was reduced, due to the AGC
affecting the audio gain.
Astor GR 3-valve TRF receiver
Manufactured from around 1948,
siliconchip.com.au
the Astor GR (Football) is a simple
little TRF receiver (see May 2009 for a
full description). As with the Kriesler
11-41/11-49, it is also a reflex design
but the circuit configuration is simpler.
Fig.2 shows details. In this circuit,
the RF signal from the tuned antenna
circuit is fed to the grid (top cap) of a
6G8G RF stage via a 100pF capacitor.
The amplified RF signal appears on the
plate (pin 3) and is fed to the primary
of the RF tuned coil assembly. The
signal on the tuned secondary is then
applied via a parallel 100pF capacitor and 3.3MΩ resistor to the 6G8G’s
detector diode (pin 4).
From there, the detected signal is fed
via a 500kΩ resistor, a 10nF capacitor
and a 1.75MΩ resistor to the grid of the
6G8G, so that amplification now takes
place at audio frequencies. The 50pF
capacitor at the bottom of the 500kΩ
resistor bypasses any RF signals at this
point, while the series-connected 3MΩ
and 1MΩ resistors go to a bias and volume control line. This line applies a
manually-controlled negative voltage
of between -2V and -22V to the grid
of the valve.
The 6G8G is a variable-mu valve
and the lower the bias the greater the
gain (and thus the greater the volume).
The resulting amplified audio appears at the 6G8G’s plate and is fed
through the primary of the RF coil assembly to a 70kΩ audio load resistor. A
1nF capacitor bypasses any RF signals
that may be present, while the audio
is fed to the grid of the 6V6GT audio
output valve via a 10nF capacitor.
As mentioned, the 70kΩ plate
load resistor used in this set is much
greater than the 15kΩ resistor used in
the Kriesler, although overload does
not appear to be a problem. Both sets
adequately filter any residual RF/
IF signals following the detector, to
prevent them being fed back to the RF
stage. This is vital to ensure stability.
Screen reflexing
Most reflex stages, such as the two
examples given, use plate circuit
reflexing. However, it’s also possible
to use the screen as the plate for the
reflex stage audio output. This usually
involves using the IF amplifier as the
reflex stage.
Fig.3 shows a typical circuit. In this
case, the plate circuit of the reflexed
IF amplifier is the same as for a conventional IF amplifier. The screen,
however, is bypassed at intermediate
siliconchip.com.au
Nicknamed the “Football” because of its cabinet shape, the Astor GR was
a simple 3-valve TRF receiver with a reflexed RF/audio stage (see Fig.2). It
was manufactured from around 1948.
frequencies using a 1nF capacitor,
while the usual bypass capacitor of
around 10nF now couples the audio
to the grid of the audio output valve.
The audio gain using the screen as
the plate will be between about 10 and
20 times, depending on the valve used.
Servicing reflex receivers
Reflex receivers do not usually
present any more servicing problems
than “non-reflexed” sets, nor are they
any more difficult to restore. However,
because they work at both RF and
audio frequencies, it is necessary to
ensure that the component values
around the reflex stage are correct, ie,
all resistors within tolerance, capacitors not leaky and the valves in good
condition.
It’s also important to remember
that relatively few valves are suitable
for use in reflex stages. The 6G8G,
6AR7GT, 6BA6 and 6AD8 are valves
that work well and although substitutes may work, they will usually
not be trouble-free. For example, the
6AD8 in the Kriesler 11-41/11-49 cannot be replaced with a 6N8 although
it appears to have reasonably similar
characteristics, the exception being
the grid cut-off voltage. It could probably be made to work quite well with
a few changes to component values,
however.
One feature of reflex sets that can
annoy some people is the “minimum
volume effect”. The problem here
is that when the volume control is
turned right down, there is still some
audio output from the loudspeaker.
The Kriesler 11-41/11-49 suffers from
this problem, which is exacerbated by
the 150Ω resistor in series with the
“earthy” end of the volume control.
However, although some people
might think that this is a problem,
most would not even notice. After all,
it’s only rarely (if ever) that the sound
would be turned right down.
By contrast, the Astor GR doesn’t
suffer from this problem, as the bias
can be increased to such a level that
no signal gets through the 6G8G valve.
In fact, this is a very trouble-free little
circuit.
Was it necessary in later sets?
As previously mentioned, reflexing
was used in the early days to keep costs
down. However, as time progressed
and valve prices fell, reflexing was
no longer really necessary. Without
reflexing, circuit layout and design
were not as critical and that suited
many manufacturers whose design
skills were often lacking.
By contrast, in sets with reflex
stages, considerable attention to the
circuit design and layout was necessary if the set was to work well.
Most manufacturers were slow
July 2010 97
high-performance receiver using just
three valves (the rectifier was a single
silicon power diode in a half-wave
circuit).
It’s fair to say that reflex sets were
still being produced long after their
early advantages had been negated by
falling valve costs. Improvements in
other aspects of receiver design also
eventually helped bring about the end
of reflexing.
Certainly, there is no advantage in
modern domestic radios using reflex
circuits. Transistors are cheap and
adding one or two transistors to a
circuit contributes little to the cost of
a receiver.
Summary
The Astor KM (or Astor Mickey) was a 4-valve reflex receiver from the late
1940s. Its reflexed IF/audio stage is similar (but not identical) to that in the
Astor GR and it had the following valve line-up: 6A8G converter, 6B8G IF/
audio reflex stage, 6V6GT audio output stage and 5Y3GT rectifier.
to take advantage of multi-purpose
valves. For example, the 6F7 and its
6P7G octal equivalent weren’t popular,
despite the fact that they contained
two valves in the one envelope which
could be used for a number of different purposes.
One of the first triode-pentode
valves used in receiver audio stages
was the 6AB8, as in the 1953 Tasma
1601. This produced a receiver with
similar performance to the Kriesler
11-41/11-49, despite the fact that the
Tasma 1601 lacked a reflex stage.
Not long afterwards, the ubiquitous
6BM8 came into use and there was no
longer any need for reflexing. The last
Kriesler valve receiver, the 11-99, used
a 6AN7 (converter), 6N8 (IF amplifier, detector and AGC) and a 6GV8
combined triode and pentode output
stage. This was a relatively simple
Reflex circuits filled a niche in the
early days of radio when components,
particularly valves, were quite expensive. Their advantages were lower
costs, lower power consumption (an
important factor in battery and vibrator
sets), less heat in cabinets and smaller
valve inventories for servicemen.
On the other hand, they required
more careful design, were not as easy
to fault-find if test equipment was
scarce and were not as tolerant of
components (including valves) which
drifted out of tolerance. If you want
to learn more about reflex receivers,
take a look at the chapter on reflex
principles in the “Radiotron Designers’ Handbook”.
Finally, reflex receivers are well
worth having in a restorer’s collection.
The Astor GR “Football” is a good example of a receiver that brings quite
SC
high prices.
Photo Gallery: Neumann KM54 Cardioid Condenser Microphone
U
SED FROM THE LATE 1950s
and right through the 1960s, this
microphone included a tiny AC701
valve. Its specifications would easily
match most studio microphones in
use today – 0.6% distortion from 40Hz
to 15,000Hz and 110dB maximum
sound pressure. A matching power
supply provided 4V DC at 100mA for
the valve filament and 120V at 0.5mA
for the plate. The filament supply was
heavily filtered to ensure low noise.
Photograph by Kevin Poulter for the
Historical Radio Society of Australia
(HRSA). Phone (03) 9539 1117. www.
hrsa.net.au
98 Silicon Chip
siliconchip.com.au
|